Teradata Corporation et al v. SAP SE et al

Filing 603

ORDER RE 467 , 472 MOTIONS FOR SUMMARY JUDGMENT AND 466 , 470 , 474 , 476 , 480 , 495 MOTIONS TO EXCLUDE EXPERT TESTIMONY (redacted of confidential information) by Judge William H. Orrick. (jmd, COURT STAFF) (Filed on 11/8/2021)

1

UNITED STATES DISTRICT COURT

2

NORTHERN DISTRICT OF CALIFORNIA

3
4

TERADATA CORPORATION, et al.,
Plaintiffs,

5
6
7
8

v.
SAP SE, et al.,
Defendants.

Case No. 18-cv-03670-WHO
REDACTED – ORDER RE MOTIONS
FOR SUMMARY JUDGMENT AND
MOTIONS TO EXCLUDE EXPERT
TESTIMONY
Re: Dkt. Nos. 466, 467, 470, 472, 474, 476,
480, 495

9
10

Defendants SAP SE, SAP America, Inc., and SAP Labs, LLC (collectively “SAP”) move

United States District Court
Northern District of California

11

for summary judgment on plaintiffs Teradata Corporation, Teradata US, Inc., and Teradata

12

Operations, Inc.’s (collectively “Teradata”) technical trade secret claims, business trade secret

13

claims, and tying claim. Teradata also moves for summary judgment and argues that

14

counterclaim-plaintiff SAP SE’s U.S. Patent No. 8,214,321 (“’321 Patent”) is invalid under 35

15

U.S.C. § 101. It argues that SAP is not entitled to damages for the alleged infringement of U.S.

16

Patent Nos. 7,617,179 (“’179 Patent”), and 9,626,421 (“’421 Patent”) before May 19, 2019.

17

Teradata also moves to exclude portions of four of SAP’s expert’s opinions: Tim Kraska, Stephen

18

Horn, Gregory Leonard and Ouri Wolfson, and Sharad Mehrotra. SAP moves to exclude portions

19

of one of Teradata’s expert’s opinions, John Asker.

20

For the reasons explained below, SAP’s motion for summary judgment on Teradata’s

21

technical trade secret claims and tying claim is GRANTED. Its motion related to Teradata’s

22

business trade secret claims under the DTSA is DENIED as moot. Its motion to exclude portions

23

of Asker’s report is GRANTED in part and DENIED in part. Teradata’s motion for summary

24

judgment on the invalidity of the ’321 Patent is GRANTED. Its motion for partial summary

25

judgment against an award of damages for infringement of the ’179 and ’421 Patents before May

26

21, 2019, is also GRANTED. Its motion to exclude portions of Kraska’s expert report is DENIED

27

as moot. Its motion to exclude portions of Horn’s report is GRANTED in part and DENIED in

28

part. Its motion to exclude portions of the Leonard and Wolfson reports is DENIED in part as

1

moot and DENIED in part on the merits. Its motion to exclude portions of Mehrotra’s report is

2

DENIED.1
BACKGROUND

3
4

I.

FACTUAL HISTORY
Teradata conducts research, development, engineering, and other technical operations

5
6

related to its Enterprise Data Analytics and Warehousing (“EDAW” or “EDW”) products. See

7

Dkt. No. 67 (“SAC”) ¶ 4. Teradata’s flagship product is the Teradata Database, a relational

8

database management system designed for EDW. SAC ¶ 16. Teradata was the first commercial

9

EDW vendor to utilize massively parallel processing (“MPP”) through Teradata Database to
execute high volumes of analytical queries on massive amounts of data for EDAW customers.

11

United States District Court
Northern District of California

10

Dkt. No. 528-9 (“Walter Decl.”) ¶ 3.

12

SAP is best known for Enterprise Resource Planning (“ERP”) software, historically

13

designed to run on transactional databases such as those by Oracle, IBM, and Microsoft. Dkt. No.

14

462-5 (“Anicich Decl.”) ¶ 39. SAP’s ERP applications do not, and have never, run on top of

15

Teradata’s analytical database. Dkt. No. 467-5 (“Mehrotra Decl.”) ¶ 127.
ERP Applications allow companies to manage data required to conduct their day-to-day

16
17

operations across numerous aspects of the business enterprise and are typically designed around a

18

relational transactional database that can ensure users have access to a uniform and current set of

19

data. Id.; Dkt. No. 452-11 (“Stiroh Rep.”) ¶ 10; Anicich Decl. ¶¶ 24–28. Transactional databases

20

are also known as “online transactional processing” (“OLTP”) databases and are typically

21

“row-based,” which is advantageous for processing transactions, such as pay roll data, and running

22
23
24
25
26
27
28

1

The parties have also filed 21 motions to seal. Dkt. Nos. 462, 465, 468, 471, 473, 475, 479, 506,
515, 518, 522, 524, 528, 531, 536, 550, 551, 554, 560, 568, 599. I will issue a separate order
addressing these motions. Suffice it to say, the parties have sought to seal a great deal of
information that does not meet the compelling interest standard that applies to dispositive motions.
See Ctr. for Auto Safety v. Chrysler Grp., LLC, 809 F.3d 1092, 1096–99 (9th Cir. 2016), cert.
denied sub nom. FCA U.S. LLC v. Ctr. for Auto Safety, 137 S. Ct. 38 (2016). While I will address
all of the sealing requests in a separate order, what is not sealed in this Order does not meet the
applicable standard. Further, with respect to those portions of this Order that are sealed, the
parties should not assume that I have concluded that they have provided a sufficient basis to seal
the information. These redactions are preliminary and should not be taken as an indication about the
merits of sealing.
2

1
2

a large number of simple transactions concurrently. Mehrotra Decl. ¶ 53.
In contrast, analytical applications are designed to run on a second type of database, known

3

as an analytics or “OLAP” database. Dkt. No. 463-15 (“Sell Depo.”) at 18–19. These databases

4

typically store data in columns to optimize the running of a small number of queries with a large

5

number of complex records. Mehrotra Decl. ¶ 60; Dkt. No. 562-6 (“Kraska Decl.”) ¶ 22. There

6

are three different types of analytical databases: (1) data marts; (2) enterprise data warehouses

7

(“EDWs”); and (3) data lakes. Sell Depo. at 14. EDWs are large-structured analytics databases

8

that draw data from different sources, e.g., transactional databases, across an enterprise. Id. at 13.

9

In 2009, SAP and Teradata entered into a partnership referred to as the “Bridge Project” to
combine SAP's ERP Applications and SAP BW tool interface with Teradata’s MPP architecture

11

United States District Court
Northern District of California

10

that it uses in Teradata Database for EDW. Kraska Decl. ¶ 161. During the Bridge Project,

12

Teradata provided SAP with access to its confidential information. SAC ¶ 35. The parties

13

executed two agreements to formalize the Bridge Project, the Software Development Cooperation

14

Agreement (“SDCA”) and the Technology Partner Agreement (“TPA”). SAC ¶ 32. These

15

agreements restricted disclosures of each parties’ confidential information. Id. The parties also

16

entered into a mutual non-disclosure agreement (“Mutual NDA”) in December 2008 and June

17

2009 (“NDAs”). Id.

18

Through the Bridge Project, SAP and Teradata jointly developed “Teradata Foundation”

19

which enabled SAP's ERP applications to use Teradata for the transactional database and

20

data-analytics for EDW activities. Dkt. No. 528-5 (“Graas Decl.”) ¶¶ 6–9. While the Bridge

21

Project was underway, SAP was developing another EDW product called SAP HANA (“HANA”).

22

Dkt. No. 530-39 (“Primsch Depo.”) at 362. By June 2011, HANA was commercially available.

23

After nearly three years in the Bridge Project, and two months after HANA was made available,

24

SAP unilaterally terminated the joint venture and stopped supporting, selling, and marketing

25

Teradata Foundation. Dkt. No. 529-25 at 068.

26

In February 2015, SAP launched its latest version of ERP Application, SAP S/4HANA and

27

combined its ERP Application and EDAW products into a single sales offering. S/4HANA is

28

integrated to operate on top of SAP’s HANA database, a translytical database with both
3

1

transactional and analytical functionalities. Anicich Decl. ¶ 40. Customers can purchase HANA

2

either with a full-use license, with no restrictions on how the data within HANA can be used, or a

3

lower-cost limited-use “runtime” license, with database use limited to supporting S/4HANA. Dkt.

4

No. 467-53 (“Zenus Depo.”) at 105–115. In other words, if customers want to export their own

5

data from HANA for use with third-party products, they must pay an additional license fee, i.e., an

6

exit fee. Dkt. No. 532-41 at 583.

7

II.

PROCEDURAL HISTORY
On June 19, 2018, Teradata filed a complaint against SAP alleging, among other things,

9

misappropriation of its trade secrets and violation of antitrust laws.2 Dkt. No. 1. On December

10

12, 2018, I granted in part and dismissed in part SAP’s motion to dismiss. Dkt. No. 65 (“MTD

11

United States District Court
Northern District of California

8

Order”). On December 21, 2018, Teradata filed a second amended complaint alleging that SAP

12

disingenuously entered a joint venture with it to steal its trade secrets and develop a competing

13

product, HANA, misappropriating trade secrets, and violating antitrust laws in the process. See

14

Dkt. No. 67 (“SAC”). The following claims remain at issue: whether SAP misappropriated

15

Teradata’s trade secrets related to the Batched Merge method and whether SAP unlawfully tied its

16

ERP applications to its HANA product.3 SAP answered on January 11, 2019 and filed

17

counterclaims related to five of its patents on May 29, 2019. See Dkt. Nos. 72, 106. SAP’s

18

remaining patent infringement counterclaims concern the following patents: the ’421 Patent, the

19

’321 Patent, and the ’179 Patent. On June 12, 2020, I held a claims construction hearing, and

20

issued an order on July 15, 2020. Dkt. No. 279 (“Claim Construction Order”). On August 25,

21

2021, Teradata and SAP filed all of the motions at issue.
LEGAL STANDARD

22
23

I.

SUMMARY JUDGMENT
A party is entitled to summary judgment where it “shows that there is no genuine dispute

24
25
26
27
28

2

This case is related to Teradata v. SAP, No. 20-CV-06127-WHO.

3

The parties dispute whether Teradata asserted a new tying theory during the summary judgment
briefing, i.e., that SAP unlawfully tied S/4HANA to HANA’s analytical capabilities through
licensing. See infra Part I.B.3.
4

1

as to any material fact and [it] is entitled to judgment as a matter of law.” FED. R. CIV. P. 56(a). A

2

dispute is genuine if it could reasonably be resolved in favor of the nonmoving party. Anderson v.

3

Liberty Lobby, Inc., 477 U.S. 242, 248 (1986). A fact is material where it could affect the

4

outcome of the case. Id.
The moving party has the initial burden of informing the court of the basis for its motion

6

and identifying those portions of the record that demonstrate the absence of a genuine dispute of

7

material fact. See Celotex Corp. v. Catrett, 477 U.S. 317, 323–24 (1986). Once the movant has

8

made this showing, the burden shifts to the nonmoving party to identify specific evidence showing

9

that a material factual issue remains for trial. Id. The nonmoving party may not rest on mere

10

allegations or denials from its pleadings but must “cit[e] to particular parts of materials in the

11

United States District Court
Northern District of California

5

record” demonstrating the presence of a material factual dispute. FED. R. CIV. P. 56(c)(1)(A); see

12

also Liberty Lobby, 477 U.S. at 248. The nonmoving party need not show that the issue will be

13

conclusively resolved in its favor. Id. at 248–49. All that is required is the identification of

14

sufficient evidence to create a genuine dispute of material fact, thereby “requir[ing] a jury or judge

15

to resolve the parties' differing versions of the truth at trial.” Id. (internal quotation marks

16

omitted). If the nonmoving party cannot produce such evidence, the movant “is entitled

17

to . . . judgment as a matter of law because the nonmoving party has failed to make a sufficient

18

showing on an essential element of her case.” Celotex, 477 U.S. at 323.
On summary judgment, the court draws all reasonable factual inferences in favor of the

19
20

nonmoving party. Liberty Lobby, 477 U.S. at 255. “Credibility determinations, the weighing of

21

the evidence, and the drawing of legitimate inferences from the facts are jury functions, not those

22

of a judge.” Id. However, conclusory and speculative testimony does not raise a genuine factual

23

dispute and is insufficient to defeat summary judgment. See Thornhill Publ'g Co., Inc. v. GTE

24

Corp., 594 F.2d 730, 738–39 (9th Cir. 1979).

25

II.

FEDERAL RULES

26

Federal Rule of Evidence 702 allows a qualified expert to provide an opinion where:

27

(a) the expert's scientific, technical, or other specialized knowledge
will help the trier of fact to understand the evidence or to determine a
fact in issue;

28

5

1
2
3

(b) the testimony is based on sufficient facts or data;
(c) the testimony is the product of reliable principles and methods;
and
(d) the expert has reliably applied the principles and methods to the
facts of the case.
FED. R. EVID. 702.

4
5
6

Expert testimony is admissible under Rule 702 “if it is both relevant and reliable.” See
Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 589 (1993). “[R]elevance means that the
evidence will assist the trier of fact to understand or determine a fact in issue.” Cooper v. Brown,

7
510 F.3d 870, 942 (9th Cir. 2007). Under the reliability requirement, expert testimony must
8
9
10

United States District Court
Northern District of California

11
12
13

“relate to scientific, technical, or other specialized knowledge, which does not include
unsubstantiated speculation and subjective beliefs.” Id. To ensure reliability, the court must
“assess the [expert's] reasoning or methodology, using as appropriate such criteria as testability,
publication in peer reviewed literature, and general acceptance.” Primiano v. Cook, 598 F.3d 558,
565 (9th Cir. 2010). These factors are “helpful, not definitive,” and a court has discretion to
decide how to test reliability “based on the particular circumstances of the particular case.” Id.

14
(internal quotation marks and footnotes omitted).
15

The inquiry into the admissibility of expert testimony is “a flexible one” where “[s]haky

16
but admissible evidence is to be attacked by cross examination, contrary evidence, and attention to
17

the burden of proof, not exclusion.” Id. at 564. “When the methodology is sound, and the

18
evidence relied upon sufficiently related to the case at hand, disputes about the degree of relevance
19
or accuracy (above this minimum threshold) may go to the testimony's weight, but not its
20

admissibility.” i4i Ltd. P'ship v. Microsoft Corp., 598 F.3d 831, 852 (Fed. Cir. 2010). The burden

21
is on the proponent of the expert testimony to show, by a preponderance of the evidence, that the
22
admissibility requirements are satisfied. FED. R. EVID. 702 advisory committee notes.
23
24
25
26

“Trial courts must exercise reasonable discretion in evaluating and in determining how to
evaluate the relevance and reliability of expert opinion testimony.” United States v.
Sandoval-Mendoza, 472 F.3d 645, 655 (9th Cir. 2006). A district court serves as “a gatekeeper,
not a factfinder.” Id. at 654.

27
28
6

DISCUSSION

1
2

I.

SAP’S MOTION FOR SUMMARY JUDGMENT

3

SAP moves for summary judgment on Teradata’s technical trade secret claims, business

4

trade secret claims, attempted monopolization claim, and tying claim. Dkt. No. 467 (“SMSJ”).

5

During the briefing, however, Teradata stipulated to the dismissal of its attempted monopolization

6

claim. Dkt. No. 545. It also voluntarily dropped its federal Defend Trade Secrets Act (“DTSA”)

7

claim regarding its business trade secrets. Dkt. No. 542 (“Opp. SMSJ”) at 25. It continues to

8

assert claims related to trade secrets 54–56 under the California Uniform Trade Secrets Act

9

(“CUTSA”). Id. SAP’s motion for summary judgment related to Teradata’s attempted
monopolization claim and business trade secret claims under the DTSA is therefore DENIED as

11
United States District Court
Northern District of California

10

moot. Teradata opposes SAP’s motion related to its technical trade secret claims and its tying

12

claim.

13

A.

14

SAP moves for summary judgment on Teradata’s technical trade secret claims. SMSJ at

Trade Secret Claim

15

2–12. Over the course of this case, Teradata’s trade secret allegations have significantly narrowed

16

and now focus on only one category of technical trade secrets: the Batched Merge method

17

(“Asserted Trade Secret”),

18

Dkt. No. 528-9 (“Walter

19

Decl.”) ¶¶ 3–12. SAP asserts that Teradata does not have standing to sue, it failed to mark the

20

Asserted Trade Secret as confidential, as required by its agreements with Teradata, and that SAP is

21

contractually authorized to use the product that incorporates Teradata’s Asserted Trade Secret.

22

SMSJ at 2–12.

23
24

1.

Standing

SAP asserts that Teradata lacks standing to assert its technical trade secret claims because

25

Teradata assigned the claims to Marlin Equity, a third-party firm. SMSJ at 10. On April 22, 2016,

26

Teradata sold to Marlin a portion of its assets related to its “Marketing Execution and Marketing

27

Operations” business, or Teradata’s Marketing Applications Business (“TMA Business”) as

28

defined in the Asset Purchase Agreement (“APA”). Dkt. No. 543-3 (“APA”) at 15. In the APA,
7

1

Teradata assigned to Marlin contracts that were material to the TMA Business or “Material

2

Contracts” as defined in the APA. Id. at 42. Teradata expressly identified the SDCA as a

3

“Material Contract.” Dkt. No. 464-26 at (“APA Schedule”) at 163. In addition, Teradata’s

4

general counsel wrote to SAP that Teradata was assigning away “all of Teradata’s rights, title and

5

interests in, to, and under the” SDCA and that it “should deal solely” with Marlin. Dkt. No. 464-1

6

(“Lanier Decl., Ex. 58”). Teradata acknowledges that it provided the Asserted Trade Secret to

7

SAP “[d]uring the Bridge Project, subject to the terms of the parties’ agreements” such as the

8

SDCA. SAC ¶ 34. SAP asserts that because the SDCA is material to the TMA Business under the

9

APA, Teradata’s alleged trade secrets claims are also material to the TMA Business and therefore

10

United States District Court
Northern District of California

11

that Teradata assigned all such claims to Marlin. SMSJ at 11.
Teradata contends that no assignment occurred because (1) any assignment required SAP’s

12

consent and SAP never consented and (2) the listing of the SDCA in the APA was a mistake and

13

not part of the assignment. Opp. SMSJ at 11. Section 14.9 of the SDCA states that “Neither party

14

may assign this Agreement . . . except with the express written consent of the other Party.” SDCA

15

¶ 14.9. Similarly, Section 5.5(b) of the APA states, “Notwithstanding anything in this Agreement

16

to the contrary, this Agreement shall not constitute an assignment, sale, transfer, conveyance etc.,

17

with respect to any Transferred Asset, or any right thereunder if an assignment, sale,

18

transfer . . . without the Third-Party Consent of, or other action by, any third party, would

19

constitute a breach or other contravention of the terms of such Transferred Asset.” APA § 5.5(b);

20

Dkt. No. 596 (“Hearing Tr.”) at 6.

21

It is undisputed that SAP did not consent to the assignment of the SDCA under the APA.

22

Dkt. No. 532-6 (“Weber Depo.”) at 81–83. But SAP asserts that its consent is irrelevant to the

23

issue of assignment under the SDCA’s choice-of law, New York law. Reply SMSJ at 2–3. New

24

York courts have “consistently held that assignments made in contravention of a prohibition

25

clause [e.g., a contractual provision prohibiting assignments without the written consent of a

26

party] in a contract are void if the contract contains clear, definite, and appropriate language

27

declaring the invalidity of such assignments.” Sullivan v. Int'l Fid. Ins. Co., 465 N.Y.S.2d 235,

28

237 (1983) (collecting cases); see also Purchase Partners, LLC v. Carver Fed. Sav. Bank, 914 F.
8

1

Supp. 2d 480, 505 (S.D.N.Y. 2012), on reconsideration in part, 2013 WL 1499417 (S.D.N.Y.

2

Apr. 10, 2013) (holding that a transfer was valid because even though the agreement prohibited

3

transfers or assignment without the written consent of the other party, it did not state that “any

4

such transfer or assignment would be invalid or void.”). The SDCA does not contain clear,

5

definite, and appropriate language declaring the invalidity of an assignment made without SAP’s

6

consent. The assignment of the SDCA to Marlin is therefore valid.
Even if SAP had consented and its claims were subject to the assignment, Teradata

7
8

contends that summary judgment should be denied because whether the SDCA was listed in the

9

APA by mistake is a disputed factual issue and that the APA should be reformed to rectify the
mistake. Opp. SMSJ at 11 n.15. It argues that because the SDCA is not related to the TMA

11

United States District Court
Northern District of California

10

Business, its inclusion in the APA was a mistake.
The APA is governed by Delaware law. APA § 11.3(a). “Claims for contract reformation

12
13

require proof by clear and convincing evidence.” Parke Bancorp Inc. v. 659 Chestnut LLC, 217

14

A.3d 701, 710 (Del. 2019). In cases of unilateral mistake, reformation is permissible only when

15

“the other party knew of the mistake but remained silent.” Scion Breckenridge Managing

16

Member, LLC v. ASB Allegiance Real Est. Fund, 68 A.3d 665, 680 (Del. 2013). In cases of

17

mutual mistake, reformation is permissible only if the parties “came to a specific prior

18

understanding … that differed materially from the written agreement.” Parke Bancorp, 217 A.3d

19

at 710.

20

Teradata does not show by clear and convincing evidence that Marlin knew of Teradata’s

21

mistake and remained silent or that Marlin and Teradata expressed an intent to agree to terms that

22

differed from the terms in the APA. Instead, it contends that Steven Weber, its Global Head of

23

Deal Management, testified that the SDCA or the Teradata Database “does not have anything to

24

do with the TMA products” and that the letter from Teradata’s general counsel was sent in error.

25

Opp. SMSJ at 11. But such self-serving testimony is not evidence that there was either a unilateral

26

or mutual mistake necessary for contract reformation. Reply SMSJ at 3. The APA is

27

unambiguous and so “the writing itself is the sole source for gaining an understanding of intent.”

28

Penton Bus. Media Holdings, LLC v. Informa PLC, 252 A.3d 445, 461 (Del. Ch. 2018). Because
9

1

Teradata fails to identify sufficient evidence to create a genuine dispute of material fact, whether

2

the SDCA was listed in the APA by mistake cannot preclude summary judgment.

3

In contrast, Teradata contends that neither the APA nor the SDCA assigns Teradata’s trade

4

secrets to Marlin and “without a clear assignment of the underlying intellectual property,” it

5

cannot have assigned its trade secret claims to Marlin. Id.at 12. SAP asserts that the question

6

about whether the APA assigned the Asserted Trade Secret itself is irrelevant; the relevant

7

question is whether Teradata assigned the right to sue for misappropriation of the Asserted Trade

8

Secret. Reply SMSJ at 3–4. It argues that Teradata did because it assigned to Marlin “[a]ll

9

Intellectual Property that is Related to the TMA Business,” e.g., “trade secrets and confidential
proprietary business information” and Teradata’s claim is based on trade secrets purportedly

11

United States District Court
Northern District of California

10

provided subject to the SDCA’s terms, which is “primarily related” to the TMA business. APA §

12

2.1(e); APA at 8. It also assigned to Marlin “[a]ll rights to Any Actions of any nature available to

13

or being pursued by any member of [Teradata] to the extent related to the TMA Business” and

14

“[a]ll goodwill and the going concern value of the TMA Business or the Marks included in the

15

Transferred IP, and the right to sue for and recover for damages and profits for past and future

16

infringements and misappropriations by any third party of any part of any of the Transferred IP

17

owned by any member of [Teradata].” Id. §§ 2.1(h), (k). Teradata responds that its trade secret

18

claims are not primarily related to the SDCA because its claims are for misappropriation and not

19

breach of contract. Hearing Tr. at 7–8. It also contends that no actual intellectual property was

20

ever conveyed. Id.

21
22
23
24

At the very least, there is a genuine dispute of whether Teradata’s trade secret claims fall
outside the scope of the assignment. Teradata has standing to pursue its trade secret claims.
2.

Marking the Asserted Trade Secret Communications as Confidential

SAP asserts that Teradata’s Batched Merge method trade secret claims also fail, however,

25

because Teradata never marked as confidential any of the communications that purportedly

26

disclosed the trade secrets as required by the NDAs. SMSJ at 13. The NDAs governed the

27

sharing of confidential information during the Bridge Project and stated that “Confidential

28

Information” shall mean the following:
10

1
2
3
4
5
6
7
8

“[A]ll information which Disclosing Party protects against
unrestricted disclosure to others, furnished by the Disclosing Party . .
. to the Receiving Party . . . in writing or in other tangible form and
clearly identified as confidential or proprietary at the time of
disclosure marked with an appropriate legend indicating that the
information is deemed confidential or proprietary by the Disclosing
Party . . . Where the Confidential Information has not been reduced
to written or other tangible form at the time of disclosure, and such
disclosure is made orally or visually, the Disclosing Party agrees to
identify it as confidential or proprietary at the time of disclosure and
to summarize the Confidential Information in writing and deliver such
summary within thirty (30) calendar days of such oral or visual
disclosure provided . . . .”
Dkt. No. 463-27 (“Mutual NDA 1”) ¶ 2; Dkt. No. 463-28 (“Mutual NDA 2”) ¶ 2.
There are two documents at issue, different versions of the same document created by John

9

Graas, a Teradata employee, that were marked confidential (“Marked Document”). See Dkt. Nos.

10
464-3 (version 1), 463-24 (version 6). Teradata contends that the first version of the Marked
11

United States District Court
Northern District of California

Document from July 2008 identifies the Batched Merge method and explained how the method
12
13
14
15
16
17
18

could resolve SAP’s problems. Opp. SMSJ at 14. But its witness, Graas, concedes that the first
version of the Marked Document “does not contain the details of the overall batched merge
method that was conveyed – conveyed to SAP” and only “listed the batched merge method . . . as
a reference.” Dkt. No. 462-17 (“Graas Depo.”) at 106, 109. He explained that “the entire
explanation of the batched merge method . . . would not have been in writing” and that he would
“have conveyed it verbally within the meeting explaining to [SAP] what I meant with the [batched
merge method] and how it worked.” Id. at 84, 105.

19
The sixth version from August 2010, however, explained the manner in which the Batched
20
Merge method was implemented to address deficiencies that prevented SAP from processing large
21
batches of data. Opp. SMSJ at 13; Dkt. No. 463-24 § 4.2. SAP does not dispute that the sixth
22
version contains the details of the Batched Merge method but argues that there is no evidence that
23
24
25
26

Graas ever sent the sixth version to SAP. Graas Depo. at 268 (“Q: Do you have any records, any
evidence at all, of version 6 being shared with SAP in any way? A: I don’t recall.”). It relies on
Prostar Wireless Grp., LLC v. Domino’s Pizza, Inc., 360 F. Supp. 3d 994 (N.D. Cal. 2018), to
argue that because “there was no evidence that the defendant had access to the trade secrets” the

27
sixth version of the Marked Document cannot be the basis for a trade secret claim. Reply SMSJ at
28
11

1

6 (citing Prostar Wireless, 360 F. Supp. at 1002). Teradata objects and contends that Prostar

2

Wireless is distinguishable because SAP does not dispute that it received the five prior versions of

3

the document. Opp. SMSJ at 13 n.17. But Section 4.2 of the sixth version, which conveys how

4

the Batched Merge Method was implemented, is not in any other version. Reply SMSJ at 6 n.3;

5

see Dkt. Nos. 530-19 at 224–25, 464-3, 529-27, 529-28, 529-29. The first version of the Marked

6

Document therefore fails to put SAP on notice about the allegedly confidential Batched Merge

7

Method; there is no evidence that SAP received the sixth version.

8

The question then is whether Teradata’s trade secret claims fail because of its failure to

9

mark as confidential the communications that allegedly conveyed the Batched Merge Method.
Teradata contends that its claims do not fail because, as SAP admits, Graas testified that he

11

United States District Court
Northern District of California

10

conveyed the Batched Merge Method orally to SAP employees in relation to the first version of

12

the Marked Document. Opp. SMSJ at 15–17. It argues that the MNDA does not require marking

13

subsequent oral discussion of confidential information already marked as confidential and even if

14

it did, the parties waived this requirement through their conduct. Id. at 17. SAP responds that its

15

argument is not that the NDAs require marking subsequent oral discussions of confidential

16

information already marked as confidential but that Graas never disclosed the claimed trade secret

17

in a writing marked as confidential in the first place. Reply SMSJ at 6.

18

It also asserts that the parties have not waived the marking requirement through their

19

conduct. Id. at 7. The NDAs contain no-waiver provisions and therefore “Teradata must prove

20

that SAP intentionally relinquished the marking provision and the no-waiver provision itself.” Id.;

21

Mutual NDA 1 ¶ 15; Mutual NDA 2 ¶ 15. Under New York law, waiver “‘should not be lightly

22

presumed’ and must be based on ‘a clear manifestation of intent’ to relinquish a contractual

23

protection.” Kassab v. Kasab, 195 A.D.3d 832, 838 (N.Y. App. Div. 2021).

24

There is no clear manifestation of intent to relinquish either provision here. Teradata

25

contends that there is waiver based on the conduct of its and SAP’s employees “that shows that the

26

parties intended that subsequent oral discussions of information already identified as confidential

27

would be treated confidentially.” Opp. SMSJ at 17–18. In support, it emphasizes testimony from

28

various SAP employees, including SAP’s CTO, where they state that they were required to protect
12

1

Teradata’s confidential information. See, e.g., Dkt. No. 532-1 (“Sikka Depo.”) at 58–59 (“Q: So

2

if it turned out that somebody on your team took Teradata confidential information and used it in

3

the development of NewDB, would that . . . be consistent with your understanding of the

4

confidentiality obligations between SAP and Teradata? [Objection to form] A: No, it would not

5

be consistent with it.”); Dkt. No. 530-25 (“Holetke Depo.”) at 120 (“Q: So you cannot imagine

6

sharing Teradata’s confidential information with other groups at SAP? [Objection to form] A:

7

Yes.”); Dkt. No. 530-39 (“Primsch Depo.”) at 354–55 (“Q: Employees working on the Bridge

8

Project would not have shared [internal] information outside of SAP or Teradata? A: To -- to the

9

extent, yes.”). It also emphasizes an email from an SAP employee telling colleagues not to share
“internal information of TD [Teradata]” with IBM. Dkt. No. 529-30 at 437. These are not

11

United States District Court
Northern District of California

10

evidence of a clear manifestation of SAP’s intent to relinquish either provision in the Mutual

12

NDAs. As SAP points out, it is not apparent from the testimonies what information was

13

“confidential” under the Mutual NDAs, only that information reduced to writing and marked

14

confidential is confidential. Reply SMSJ at 7–9.

15

SAP asserts that Teradata’s trade secret claims therefore necessarily fail because Teradata

16

failed to comply with its contractual obligation to designate information as confidential when it

17

disclosed the alleged Batched Merge Method trade secret to SAP. It requests that I reconsider my

18

analysis of PQ Labs, Inc. v. Yang Qi, No. 12–CV–0450–CW, 2014 WL 334453, at *4 (N.D. Cal.

19

Jan. 29, 2014) in the prior MTD Order, where I rejected SAP’s arguments that failure to satisfy the

20

contractual marking requirement requires the dismissal of Teradata’s claims because I concluded

21

that there may be other ways for Teradata to have disclosed its trade secrets to SAP. MTD Order

22

at 8–9. For example, the Mutual NDAs “were only two of four contracts involved in the Bridge

23

Project to ensure that Teradata’s proprietary information would not be misappropriated or reverse

24

engineered.” Id. at 9. But with the record on summary judgment, it is clear that PQ Labs case is

25

distinguishable from this case.

26

In PQ Labs, the court held that the marking requirement was irrelevant because PQ Labs

27

had “presented evidence that it used other means to notify its employees and agents that its

28

technological and customer information was confidential.” PQ Labs, 2014 WL 334453, at *4.
13

1

But unlike this case, there was no contractual marking requirement in PQ Labs; instead, the

2

marking requirement derived from non-precedential Tenth Circuit case law. Id.

3

In contrast, a case like Convolve Inc. v. Compaq Computer Corp., 527 F. App’x 910 (Fed.
Cir. 2013) is more analogous. In Convolve, there was a contractual marking requirement “to

5

confirm in writing, within twenty (20) days of the disclosure, that the information was

6

confidential.” Id. at 923. The Federal Circuit affirmed the district court’s decision that there was

7

no misappropriation of trade secrets because the appellant had failed to protect the confidentiality

8

of its information. Id. at 921–22. It also held that there was no waiver or modification of the

9

marking requirement because “the testimony of a single Seagate employee that he believed that all

10

disclosures were confidential . . . is not indicative of the mutual intent of both parties.” Id. at 924.

11

United States District Court
Northern District of California

4

Likewise, Teradata’s technical trade secret claims fail because it failed to protect the

12

confidentiality of its information. Even if PQ Labs was analogous to this case, there is no

13

evidence that Teradata notified SAP of the confidentiality of the Batched Merge method through

14

other means. Reply SMSJ at 8–9.

15
16

3.

Contractual Right to Use the Asserted Trade Secret in Any SAP
Product

Even if Teradata had sufficiently protected the confidentiality of the Batched Merge
17
18
19
20

method, SAP asserts that it is contractually authorized to use any “Confidential Information”
under the NDAs in any product. SMSJ at 16. Teradata has sued SAP for “using the proprietary
information conveyed by John Graas pertaining to Teradata’s batched merge method.” Dkt. No.
464-14 (“Lanier Decl., Ex. 72”) at 8. The Batched Merge method “is alleged to have been

21
incorporated into the Bridge Project software (the MaxDB Bridge, also called the Teradata
22
23

Foundation)—and then allegedly into the interface between SAP applications and HANA (“Native
FAE”),” also known as the conceptual design. Id. at 17.

24
According to SAP, both the software and its conceptual design are SAP property. Id.
25
26
27

Section 10 of the SDCA outlines the “Proprietary Rights of the Parties.” SDCA § 10. It states
that SAP owns all rights to “the Conceptual Design [and] the SAP Interface in the form originally
supplied by SAP as well as any modified versions,” and the “software code that is necessary to

28
14

adapt its software to” Teradata’s database, including SAP’s Interface. Id. § 10.1. The

2

“Conceptual Design” is defined as “the description of the functional specifications of the SAP

3

Interface or any other architecture, guideline or specification developed by or with SAP

4

concerning or related to the integration of the [Teradata Database] with the [SAP BW product].”

5

Id. § 1.2. “SAP Interface” is defined as “an application interface developed by or with SAP that

6

resides on or in the SAP Software and which, when activated will give access to the Partner’s

7

Solution [i.e., Teradata’s database].” Id. § 1.11. And section 10.3 states that “any and all

8

Intellectual Property Rights to or arising out of any Newly Developed Materials shall belong to

9

SAP” and “Newly Developed Materials” is defined as “any software . . . developed by SAP and/or

10

[Teradata] in connection with or as a result of a party’s interaction with the other party within the

11

United States District Court
Northern District of California

1

context of this Agreement.” Id. §§ 1.8, 10.3.

12

Teradata responds that SAP’s arguments depend on the Batched Merge method falling

13

under certain SDCA provisions, but resolving which provision applies depends on disputed factual

14

issues. Opp. SMSJ at 21. It argues that the Batched Merge trade secrets are not part of SAP’s

15

Interface or Conceptual Design but rather Partner Materials under the SDCA, and therefore when

16

SAP ended the Bridge Project, its license to use the Batched Merge method terminated. Id. The

17

SDCA defines “Partner Materials” as “any and all Intellectual Property Rights in any programs,

18

tools, systems, data or materials utilized or made available by Partner [Teradata] in the course of

19

the performance under this Agreement,” which “shall remain vested exclusively in [Teradata]” but

20

“[s]ubject to any rights expressly granted to SAP hereunder.” SDCA § 10.2. Section 9.2 of the

21

SDCA limits use of Teradata’s Partner Materials to five specific purposes, none of which include

22

the development of SAP’s HANA product; therefore, SAP was only allowed to use these materials

23

“during the Term” of the SDCA. Opp. SMSJ at 21–22. Section 9.2 provides SAP a limited

24

license to “the Partner Solution, related Documentation, and any other programs, tools, or other

25

materials provided by Partner to SAP under a Project Plan.” SDCA § 9.2.

26

SAP points out that Teradata does not and cannot explain how Graas’s suggestions are any

27

of the above. Reply SMSJ at 11. Graas’s suggestions are not the Partner Solution, which is

28

defined as only the Teradata Database itself, not the Database, not the documentation related to the
15

1

database such as manuals, not a program and not a tool. See SDCA § 1.9. As SAP asserts,

2

Graas’s suggestions do not fall under section 9.2, the purpose of which is to prevent SAP from

3

using Teradata’s Database itself. Id. Instead, section 9.4 encompasses Graas’s input, as explained

4

below.

5

Teradata argues that the Batched Merge method was neither developed in connection with

6

or as a result of the parties’ interactions within the context of the SDCA, nor developed by or with

7

SAP as the Conceptual Design. Opp. SMSJ at 22; see SDCA §10.03. Accordingly, Section 10.3

8

(“Newly Developed Materials”) could not apply because those intellectual property rights existed

9

prior to the Bridge Project. But whether Teradata owned the Batched Merge method and
incorporated it into its own software before the Bridge Project is irrelevant. SAP does not argue

11

United States District Court
Northern District of California

10

that it owns the Batched Merge method but rather that “it owns the new software that includes the

12

optimizations based on SAP’s interactions with Graas.” Id. at 10.

13

Section 10.2, which provides an exception to the rights expressly granted to SAP under the

14

SDCA, does not change the fact that Section 10.1 expressly licenses to SAP the right to use

15

Graas’s input—i.e., his conversation with SAP employees—in any product. SMSJ at 18. Both the

16

SDCA and the Mutual NDAs permit SAP to use any Teradata feedback or input regarding SAP’s

17

products, even if such information was marked confidential. Id. (citing SDCA §§ 9.4, 12; Mutual

18

NDA 1 § 7, Mutual NDA 2 § 7).

19
20
21
22
23
24
25
26

For example, Section 9.4 of the SDCA provides,
“Partner [Teradata] grants to SAP a worldwide, nonexclusive,
royalty-free fully paid up, perpetual and irrevocable license to use,
reproduce, display, distribute, create derivative works, or sublicense
any Input submitted by Partner [Teradata] to SAP with resect to any
deliverables or other items that SAP provides or shall provide to the
Partner . . . To the extent that any such Input is incorporated into an
SAP product, any inherent disclosure of Confidential and/or trade
secret Information of Partner through the exercise of the license grants
set forth in this Section 9.4 shall not constitute a breach of this
Agreement including, but not limited to, any agreement between the
Parties with respect to such Confidential or trade secret information
referenced herein.”
SDCA § 9.4. The SDCA states that “Input” means “suggestions, comments, and feedback

27
(whether in oral or written form), including any included ideas and know-how, voluntarily
28
16

1

provided by one Party to the other Party with respect to the work performed under this

2

Agreement.” Id. § 1.6. Similarly section 7 of the Mutual NDAs state,

3
4
5
6
7
8
9

“During the course of this Agreement, Company [Teradata] may
provide or SAP may solicit Company’s input regarding SAP’s
Software, products, services, business or technology plans, including,
without limitation, comments or suggestions regarding the possible
creation, modification, correction, improvement or enhancement of
SAP Software, products and/or services . . . (collectively, ‘Company
Feedback’) . . . In order for SAP to utilize such Company Feedback
Company grants to SAP a non-exclusive, perpetual, irrevocable,
worldwide, royalty-free license . . . SAP shall be entitled to use
Company Feedback for any purpose without restriction or
remuneration of any kind with respect to Company.”
Mutual NDA 1 § 7, Mutual NDA 2 § 7.
According to SAP, Graas’s suggestions to SAP engineers about how to approach a

11

United States District Court
Northern District of California

10

command/query coming from SAP applications to work more efficiently with the Teradata

12

database qualifies as “Input” under the SDCA and “Company Feedback” under the Mutual NDAs.

13

SMSJ at 18–19. Teradata does not dispute that Graas’s disclosures fall within “input,” but it

14

argues that the Batched Merge method was not mere “Input.” Opp. SMSJ at 23; Reply SMSJ

15

at 10. It argues that the trade secret is a proprietary method developed over many years, is

16

something that could not be fixed through a mere “thought” or “offhand comment,” and took SAP

17

more than a year to understand that it was necessary and months more to implement it. Opp.

18

SMSJ at 23. As SAP points out, however, the license is not limited to thoughts or offhand

19

comments but rather distinguishes Teradata software, which SAP could only use for the purposes

20

of the Bridge Project, and changes made to SAP software, which could be used in any SAP

21

product under section 9.3 of the SDCA. Reply SMSJS at 10–11. As a result, SAP has the right

22

under the agreements to use the alleged Batched Merge method in its products outside of the

23

Bridge Project. SMSJ at 19.

24

Finally, the parties dispute whether SAP’s interpretation of the SDCA contradicts the

25

implied covenant of good faith and dealing. “In every contract there is an implied

26

covenant that neither party shall do anything which will have the effect of destroying or injuring

27

the right of the other party to receive the fruits of the contract, which means that in every

28

contract.” Kirke La Shelle Co. v. Paul Armstrong Co., 188 N.E. 163, 164 (1933). But it cannot be
17

1

used “to add contract terms that contradict the unambiguous provisions of the written contracts.”

2

Atlas Equity, Inc. v. Chase Bank USA, N.A., 403 F. App’x 190, 192 (9th Cir. 2010). Teradata

3

contends that the SDCA’s purpose was to develop a joint solution that connects its hardware and

4

software with SAP Business solutions and jointly promote the solution. SDCA, Preamble.

5

According to Teradata, “SAP’s interpretation of SDCA’s license provisions—that it could use

6

what it took from Teradata to develop and sell a competing product simultaneously—would

7

destroy these benefits.” Opp. SMSJ at 24. SAP responds that Teradata’s argument contradicts its

8

intent when it entered into the SDCA. Reply SMSJ at 12. Teradata knew that SAP was not

9

working exclusively with Teradata on updates to its MaxDB database and knew that under the
SDCA, all developments of SAP products would be owned by SAP even if made by Teradata. See

11

United States District Court
Northern District of California

10

SDCA § 2.3 (“This Agreement is not exclusive. SAP or Partner may enter into similar agreements

12

with other partners.”); SDCA § 10.1.

13

Accordingly, there is no genuine dispute of fact that Teradata not only failed to protect the

14

confidentiality of its alleged trade secrets but also that SAP has a contractual right to use the

15

alleged Batched Merge method in its own product. SAP’s motion for summary judgment on

16

Teradata’s technical trade secret claims related to the Batched Merge method are GRANTED.

17
18

4.

Related Motions to Exclude Expert Testimony

Two of Teradata’s motions to exclude expert testimony relate to its trade secret claims.

19

The first is Teradata’s motion to exclude ten paragraphs in the report of SAP’s computer science

20

expert Tim Kraska. Dkt. Nos. 466, 495 at 1. Teradata asserts that Kraska improperly offers

21

opinions regarding his interpretation of the SDCA and the Mutual NDAs. Id. Because none of

22

Teradata’s technical trade secret claims survive, I DENY Teradata’s motion to exclude portions of

23

Kraska’s testimony as moot.

24

The second motion relates to Teradata’s business trade secrets and seeks to exclude certain

25

paragraphs in the report of SAP’s data management expert Stephen Horn. Dkt. No. 474 (“Horn

26

Mot.”) at 1. Horn is SAP’s rebuttal expert to Teradata’s damages expert, Paul Meyer.

27

Specifically, Teradata moves to exclude Horn’s opinions on whether the allegedly stolen Teradata

28

confidential information includes trade secrets, whether use of the confidential information
18

1

contributed to any sales of SAP HANA, whether Teradata took reasonable measures to protect its

2

confidential information, and what examples are of “reasonable” measures taken by data

3

management companies to protect confidential information. Id. (citing Dkt. No. 473-4 (“Horn

4

Reb. Rep.”) ¶¶ 19, 21, 51-57, 63-84, 97-105).

5

First, Teradata asserts that Horn relies in part on documents that he and SAP refuse to

6

produce, which Horn claims show that Teradata’s trade secrets were publicly available. Horn

7

Mot. at 4; see Dkt. No. 473-5 (“Horn Tr.”) at 74 (“Q: Did you do any investigation to see if that

8

information was publicly available in 2011? A: Yes. Actually I was able to use some of my own

9

folders of information . . . Q: But your materials in your folders are not cited in this report; right?
A: Correct, because I wanted to keep them confidential.”). Teradata seeks to exclude Horn’s

11

United States District Court
Northern District of California

10

opinions based on these documents in paragraphs 63–84. SAP responds that Horn bases his

12

opinion on materials he referenced in his report as well as publicly available documents such as

13

articles, websites, industry reports, laws, and statutes. Dkt. No. 523 (“Horn Opp.”) at 2, 9.

14

Further, SAP points out that under the parties’ stipulation (Dkt. No. 235 ¶ 3), the parties are only

15

required to produce materials underlying the expert report rather than all materials an expert ever

16

considered and therefore Horn is not required to produce the documents at issue. Id. at 10. It

17

contends that Horn does not and will not offer any opinion based on documents that are not

18

available to Teradata. Id. at 2. With this understanding, Teradata’s motion to strike paragraphs

19

63–84 because they allegedly include Horn’s opinions based on unproduced documents is

20

DENIED. Teradata may question Horn about this issue during cross examination as it goes to the

21

weight of his testimony, but it is not a basis for excluding the testimony.

22

Second, Teradata argues that I should exclude Horn’s opinions that present a legal

23

conclusion based on a fundamental misunderstanding of the law. Horn Mot. at 1. Specifically,

24

Teradata asserts that Horn should not be allowed to testify to what is or is not a trade secret

25

because his understanding of the law is incorrect and to allow his testimony would mislead the

26

jury and confuse the issues at trial. Id.

27
28

Under the CUTSA, “[c]ombinations of public information from a variety of different
sources when combined in a novel way can be a trade secret.” O2 Micro Int’l Ltd. v. Monolithic
19

1

Power Sys., Inc., 420 F. Supp. 2d 1070, 1089–90 (N.D. Cal. 2006), aff’d, 221 Fed. Appx. 996

2

(Fed. Cir. 2007). When asked whether “information that’s collected through public sources or is

3

otherwise public, when collected together, can still be [a] trade secret,” Horn testified that this

4

standard was “totally incorrect.” Dkt. No. 473-5 (“Horn. Depo.”) at 24. As a result, Teradata

5

argues that Horn should be precluded from testifying about what types of information would be

6

considered trade secrets and that the allegedly stolen confidential Teradata information cannot be a

7

trade secret because he purportedly found snippets of information from those documents in

8

various public or customer-facing documents. Horn Mot. at 8.

9

SAP responds that Horn does not provide any legal conclusions in his report and that he is
allowed to challenge the factual issue of whether or not Teradata’s alleged trade secrets could be

11

United States District Court
Northern District of California

10

ascertained by others outside Teradata. Horn Opp. at 2. But as Teradata points out, Horn does

12

offer legal opinions that Teradata’s information are not trade secrets or proprietary to Teradata.

13

See Horn Reb. Rep. ¶¶ 73–74, 77, 82. Horn can address the factual issue of whether Teradata’s

14

purported trade secret information was ascertainable to others outside of Teradata, but he cannot

15

testify that Teradata allegedly stolen confidential information are not trade secrets. Teradata’s

16

motion to exclude Horn’s legal conclusions is GRANTED.

17

Third, Teradata asserts that Horn’s opinion that its confidential information did not lead to

18

sales of SAP HANA is insufficiently supported because he admits that he failed to conduct any

19

investigation of the sales. Horn Mot. at 1. SAP contends that Horn is not required to replicate

20

Meyer’s investigation and that Horn properly relied on SAP’s damages expert Leonard, who

21

examined each of the six sales for purposes of his damages analysis. Horn Opp. at 2, 16. Teradata

22

responds that Horn does not simply rely on Leonard’s opinions but endorses them by opining that

23

they “are consistent with the commonly prevailing principles in the industry, and with [his]

24

experience and expertise.” Horn Reb. Rep. ¶¶ 101–02. It asserts that in the cases on which SAP

25

relies, the “courts have been careful to either require independent investigation or to strictly limit

26

their testimony to critiquing methodology or assumptions of an opposing expert.” Horn Opp. at 7

27

(citing TCL Comm’cns. Tech. Holdings Ltd. v. Telefonaktenbologet LM Ericsson, 2016 WL

28

7042085, at *5 (C.D. Cal. Aug. 17, 016) (holding “it is proper for [rebuttal] experts to utilize their
20

1

own independent analyses and methodologies to” rebut expert opinions); Cmty. Ass’n for

2

Restoration of the Env’t, Inc. v. Cow Palace, LLC, 80 F. Supp. 3d 1180, 1215 (E.D. Wash. 2015)

3

(“recogniz[ing] the limited bases for [rebuttal expert’s] rebuttal opinions” given the lack of

4

independent investigation)). Teradata’s motion to exclude paragraphs 21, 99–105 of Horn’s report

5

is DENIED because Horn properly relies on Leonard’s analysis, but Horn may not otherwise

6

endorse or offer any affirmative opinions about Leonard’s analysis.

7

Finally, Teradata argues that Horn is not qualified to opine on industry standards regarding
the protection of confidential information. Horn Mot. at 1. According to Teradata, Horn has no

9

experience drafting or developing protocols for the protection of confidential information, and

10

therefore cannot base his opinions on what reasonable measures are taken by data management

11
United States District Court
Northern District of California

8

companies. Id. at 11. SAP responds that Horn has decades of experience implementing, applying,

12

and working with confidentiality policies of data management companies and, based on that

13

experience, has conducted a more than sufficient review to rebut Meyer’s assumptions regarding

14

the alleged confidentiality of Teradata’s purported business trade secrets. Horn Opp. at 17–18.

15

Even if Horn did not have the experience, lack of particularized expertise goes to weight rather

16

than admissibility. Teradata’s motion to exclude paragraphs 51–57 and 97 is DENIED.

17

B.

18

SAP moves for summary judgment on Teradata’s tying claim. Before I address SAP’s

Tying Claim

19

motion, I will address two motions to exclude expert testimony and Teradata’s objections to

20

SAP’s reply evidence.

21
22

1.

Motion to Exclude Asker Testimony

SAP moves to strike the opinions of Teradata’s liability and damages expert, Dr. Asker.

23

Dkt. No. 470 (“Asker Mot.”) at 1. Asker opines that the relevant product market for the tying

24

market is “core ERP products for large enterprises.” Dkt. No. 468-20 (“Asker Rep.”) ¶ 46.

25

Market participants include SAP and Oracle, with Workday and Microsoft appearing as leaders of

26

a fringe of participants. Id. He defines the tied market as “EDW products with OLAP capabilities

27

for large enterprises” with market participants such as

28

Id. ¶ 78. He opines that SAP has economically significant market power in the tying market, that
21

1

SAP has caused harm to competition in the tied market, and that there are no procompetitive

2

benefits of the alleged tie. Id. ¶¶ 12, 105, 171. Finally, he asserts that Teradata has lost significant

3

profits and will experience significant future losses due to the alleged tying arrangement. Id.

4

¶¶ 181, 192.

5
6
7

a.

Tying Product Market

SAP argues that Asker’s methodology for defining the tying product market is unreliable

8

because instead of showing cross-elasticity, Asker’s primary methodology is “to interpret ordinary

9

course documents produced in the case” and “buttress this qualitative approach with a quantitative
‘aggregate diversion analysis’” of the Customer Relationship Management (“CRM”) data from

11

United States District Court
Northern District of California

10

SAP and Oracle. Asker Mot. at 5–6; see Asker Rep. ¶¶ 63, 64, 70.

12

In a tying arrangement the seller conditions one product, the tying product, on the buyer's

13

purchase of another product, the tied product, to extend its market power in a distinct product

14

market. See Cascade Health Sols. v. PeaceHealth, 515 F.3d 883, 912 (9th Cir. 2008). A tying

15

arrangement is “forbidden on the theory that, if the seller has market power over the tying product,

16

the seller can leverage this market power through tying arrangements to exclude other sellers of

17

the tied product.” Id.

18
19

i.

Cross-Elasticity of Demand

First, SAP asserts that Asker’s methodology is flawed because he failed to calculate the

20

cross elasticities for demand among various ERP products. Asker Mot. at 5. As the Supreme

21

Court has instructed, “The outer boundaries of a product market are determined by the reasonable

22

interchangeability of use or the cross-elasticity of demand between the product itself and

23

substitutes for it.” Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962). The Ninth Circuit

24

has recognized that “[t]he principle most fundamental to product market definition is

25

‘cross-elasticity of demand’ for certain products or services. Commodities which are ‘reasonably

26

interchangeable’ for the same or similar uses normally should be included in the same product

27

market for antitrust purposes.” Kaplan v. Burroughs Corp., 611 F.2d 286, 291–92 (9th Cir. 1979).

28

Cross-elasticity of demand occurs where “an increase in the price of one product leads to an
22

1

increase in demand for another”; in that circumstance, “both products should be included in the

2

relevant product market.” Olin Corp. v. F.T.C., 986 F.2d 1295, 1298 (9th Cir. 1993). As I have

3

previously acknowledged, “[n]umerous cases have recognized the importance of cross-elasticity to

4

determining what products should be included in or excluded from the relevant antitrust market.”

5

United Food & Com. Workers Loc. 1776 & Participating Emps. Health & Welfare Fund v.

6

Teikoku Pharma USA, 296 F. Supp. 3d 1142, 1167 (N.D. Cal. 2017) (collecting cases).
Teradata contends that Asker was not required to measure cross-elasticity of demand,

7
8

especially where, as here, it was not possible to calculate cross-elasticities. Dkt. No. 537 (“Asker

9

Opp.”) at 7. It points to three district court cases where the court relied on an expert’s
methodology that did not use cross-elasticities and instead used “practical indicia” as outlined by

11

United States District Court
Northern District of California

10

the Supreme Court in Brown Shoe to determine the boundaries of a product market. See Epic

12

Games, Inc. v. Apple Inc., No. 20-CV-05640-YGR, 2021 WL 4128925, at *85 (N.D. Cal. Sept. 10,

13

2021); In re Live Concert Antitrust Litig., 863 F. Supp. 2d 966, 984–86 (C.D. Cal. 2012); Nobody

14

in Particular Presents, Inc. v. Clear Channel Commc'ns, Inc., 311 F. Supp. 2d 1048, 1082 (D.

15

Colo. 2004). Teradata’s reliance on these three cases, however, is misplaced.
In Epic Games, the court focused on practical indicia and not cross-elasticities when

16
17

determining the submarket. Epic Games, 2021 WL 4128925, at *85. This is proper under Brown

18

Shoe and Ninth Circuit precedent. “In limited settings . . . the relevant product market may be

19

narrowed beyond the boundaries of physical interchangeability and cross-price elasticity to

20

account for identifiable submarkets or product clusters.” Thurman Indus., Inc. v. Pay 'N Pak

21

Stores, Inc., 875 F.2d 1369, 1374 (9th Cir. 1989). “The boundaries of such a submarket may be

22

determined by examining such practical indicia as industry or public recognition of the submarket

23

as a separate economic entity, the product's peculiar characteristics and uses, unique production

24

facilities, distinct customers, distinct prices, sensitivity to price changes, and specialized vendors.”

25

Brown Shoe, 370 U.S. at 325. Here, Asker is not defining a submarket but the tying product

26

market. He also does not address the practical indicia under Brown Shoe.4

27
4

28

During the hearing, Teradata’s counsel asserted for the first time that Asker’s methodology was
23

1

In In re Live Concert, the court held that “while calculating the cross-elasticity of demand

2

(and supply) is the preferred methodology, it is not an absolute requirement” and found that “it is

3

usually necessary to consider other factors that can serve as useful surrogates for cross-elasticity

4

data” because “it is ordinarily quite difficult to measure cross-elasticities of supply and demand

5

accurately.” In re Live Concert, 863 F. Supp. at 984. Likewise, in Clear Channel, the court found

6

“that a plaintiff may, through sufficient evidence of other indicia of market definition, define a

7

relevant market without economic study of cross-elasticity of demand, especially when economic

8

analysis of cross-elasticity of demand is infeasible based on pricing data.” Clear Channel, 311 F.

9

Supp. at 1082. Both courts then evaluated the sufficiency of the expert’s methodology that was
based on the Brown Shoe practical indicia factors. See, e.g., Clear Channel, 311 F. Supp. at 1083

11

United States District Court
Northern District of California

10

(finding that the expert’s methodology is sufficient). The court in In re Live Concert recognized

12

that the Ninth Circuit “has never expressly held that . . . a plaintiff’s expert economist[] can define

13

the relevant product market exclusively by reference to these ‘practical indicia.’” In re Live

14

Concert, 863 F. Supp. at 986, 985. But for the purposes of the motion, it assumed that an expert

15

economist could and found that the expert’s purported market definition was “neither sufficiently

16

reliable nor sufficiently helpful to the trier of fact to warrant admission under Rule 702” because

17

the expert’s analysis (1) fails to comport with his “chosen methodology (i.e., the “SSNIP”

18

methodology); (2) is effectively predicated on the analysis of a single Brown Shoe factor; and (3)

19

fails to consider the cross-elasticity of supply.” Id. at 994.

20

Teradata contends that it was “not possible to calculate cross-elasticities” here because

21

third parties such as Oracle, IBM, and Microsoft were not “ordered to produce the type of granular

22

data required to calculate cross-elasticities of demand.” Asker Opp. at 7. It also argues that SAP’s

23
24
25
26
27
28

proper because market definition can be determined based on practical indicia. Hearing Tr. at 31.
Its counsel expressly identified one factor, industry or public recognition of the market, as a
separate economic entity. Id. But the case on which Teradata relies holds that, “[t]he existence of
three or four of these indicia has been held ‘sufficient to delineate a submarket,’” not one. In re
Live Concert Antitrust Litig., 863 F. Supp. at 989. As explained in the subsequent sections, there
is no evidence of three or four of these practical indicia. See Brown Shoe, 370 U.S. at 325
(practical indicia are “industry or public recognition of the submarket as a separate economic
entity, the product’s peculiar characteristics and uses, unique production facilities, distinct
customers, distinct prices, sensitivity to price changes, and specialized vendors.”).
24

1

expert, Dr. Stiroh, admits that such data is unavailable. Id. at 8; see Dkt. No. 541-33 ¶ 62

2

(“Econometric methods include the estimation of the cross-price elasticity of demand. However, I

3

have not seen data in this case that can be used to reliably estimate actual lost sales and diversion

4

ratios in response to price changes of different ERP products.”). SAP responds that Teradata did

5

not request any data from third parties that would have permitted analysis of cross-elasticity of

6

demand. Dkt. No. 555 (“Asker Reply”) at 2.

7

Regardless, Teradata contends that Asker does, in fact, analyze cross-elasticity of demand.

8

Hearing Tr. at 29. According to Teradata, although Asker does not have an econometric

9

estimation of cross-elasticity, he looks at cross-elasticity, i.e., substitutability, from a quantitative

10

United States District Court
Northern District of California

11
12

and qualitative standpoint. Id.
ii.

Qualitative Analysis

SAP asserts that Asker’s qualitative approach is unreliable because it is based on “his own

13

subjective interpretation of ordinary course documents” and his inconsistent use of evidence.

14

Asker Mot. at 5, 7. In particular, it criticizes Asker’s definition of “core ERP” and “large

15

enterprises” (“LEs”) in his tying product market definition of “core ERP products for large

16

enterprises,” composed of SAP and Oracle. Asker Opp. at 9.

17

For “core ERP,” SAP asserts that Asker changes its definition to fit his needs. Id. at 7. For

18

example, Asker defines “core ERP” as products that “are identified with reference to the finance

19

modules of ERP software.” Asker Rep. ¶ 9. According to Exhibit 2 in Asker’s initial report,

20

however, only 30% of “core EP” is finance and the other 70% is human resources, procurement,

21

R&D sales, supply chain, and travel. See id. at 33, Ex. 2. But Asker mainly focuses on finance

22

when defining “core ERP.” He explains that “[w]hile there are various core ERP definitions, a

23

consistent feature of core ERP is that it includes finance.” Id. ¶ 48. And none of SAP’s three

24

economic experts dispute Asker’s relevant market definition. See, e.g., Dkt. No. 532-3 (“Stiroh

25

Depo.”) at 108–09 (“Q: And you don’t dispute in your report that Dr. Asker’s opinion that the

26

relevant product market for S/4HANA is limited to core ERP products; is that right? A: I don’t

27

take that on. . . . The opinions that I have in my report are not dependent on a specific definition of

28

what is included or excluded in core ERP.”).
25

1

Moreover, Asker’s focus on finance in his definition of “core ERP” is supported by SAP’s

2

own witness testimony. SAP’s Vice President for competitive market insights for business

3

applications and industries testified that the solutions included within “core ERP” are “general

4

ledger and some of the other financial – you know, financial close, that type of activity you start to

5

– I’d also say that master data governance types of products are – may be considered part of digital

6

core.” Dkt. No. 530-16 (“Dover Depo.”) at 30. Likewise, SAP’s Senior Vice President of

7

S/4HANA testified that when SAP decided to build S/4HANA it “obviously started in the finance

8

area because that’s the center of every ERP system.” Dkt. No. 530-22 (“Grigoliet Depo.”) at 24.

9

Teradata’s industry expert, Paul Pinto also opined that “large enterprises build their systems
around their financial ERP, which is why it is often referred to as ‘core ERP.’” Dkt. No. 528-8

11
United States District Court
Northern District of California

10

(“Pinto Decl.”) ¶ 31.

12

SAP also objects to Asker’s use of applications such as “treasury management” “when it is

13

expedient to do so” because it is outside his definition of core finance. Asker Mot. at 7 (citing

14

Asker Rep. ¶ 122 (mentioning that SAP

15

). The column titled “core ERP” in Exhibit

16

2 in Asker’s report does not mention “treasury management.” Asker Rep. at 33, Ex. 2. Teradata’s

17

own expert also testified that “treasury management” is not part of “core ERP.” See Dkt. No. 555-

18

3 (“Pinto Depo.”) at 83 (“Q: What about cash and treasury management, would you consider that

19

part of core ERP? A: I would not.”). That said, column 1 in Exhibit 2 in Asker’s report, titled

20

“Digital Core” includes “Treasury Management” under “Core Finance.” Asker Rep. at 33, Ex. 2.

21

Asker also testified that he considered treasury management part of “core ERP.” Dkt. No. 536-6

22

(“Asker Depo.”) at 20–22 (“[T]he left-hand side column it – this is labeled . . . ‘Digital Core.’ It

23

says ‘Core Finance plus Enterprise Risk & Compliance, Treasury Management, Real Estate

24

Management, Indirect Tax.’ I interpret that as incorporating the articulation of “Core Finance”

25

that’s located in the middle column.”). In addition, SAP’s own documents include “treasury

26

management” in its definition of “core ERP.” Dkt. No. 536-12 at 69–70. As a result, contrary to

27

SAP’s assertion, Asker’s definition of “core ERP” is proper.

28

Next, SAP argues that Asker’s definition of “large enterprises” is problematic. Asker Mot.
26

1

at 8. Asker defines “large enterprises” as “companies with over 1,000 or 1,500 employees and

2

over 125 users of the ERP product”; his own sources, such as SAP’s internal documents, show

3

that there is no commonly accepted categorization of SAP’s customers. See, e.g., Asker Rep. ¶ 50

4

n.110–11 (SAP internal presentation defining large enterprise as “Revenue: €250 [million] + Size

5

250 employees”); id. ¶ 50 n.111 (SAP presentation defining large enterprises as companies with

6

“over 1000, 5000, or 10K”); id. ¶ 38 n.84 (large enterprises: over 500 employees, $1 billion in

7

annual revenue, and an ERP user count of over 250); id. ¶ 20 n.24 (large enterprise: companies

8

with over $1 billion in revenues in North America and over $250 million or $500 million in Latin

9

America).

10

Further, according to SAP, Asker testifies that there are approximately 100,000

United States District Court
Northern District of California

11

companies in his proposed relevant market but he does not sufficiently explain why he then

12

focuses only on documents discussing the largest 500 or 2,000 companies in the world, e.g.,

13

companies in Forbes Global 2000, Global Fortune 500, DAX stock index, and MDAX index.

14

Asker Mot. at 8; see Dkt. No. 512-2 (“Asker Depo.”) at 32 (“Q: And global . . . is fewer than

15

100,000 companies would qualify as large enterprises? A: It may be a little more than that . . . Q:

16

So best estimate is, give or take, somewhere around a hundred thousand? A: [A]s I sit here today,

17

that would be my sense, but I want to be very clear that it may be a fair bit less, it may be

18

somewhat more.”). Teradata responds that SAP mischaracterizes Asker’s testimony and that he

19

repeatedly testified that SAP does not count its own customers. Dkt. No. 536-6 (“Asker Depo.”)

20

at 30 (“I note that even in their own documents, SAP doesn’t count customers; but, rather, they

21

talk about total market opportunities). It also highlights SAP’s own documents to its investors,

22

which show that SAP relies on the Forbes Global 2000 index and the DAX stock index to assess

23

and report its market position. See Dkt. No. 537-8 at 7 (“S/4 is further gaining market share, and

24

we see positive software license growth and high double-digit cloud revenue growth. 80% of the

25

DAX companies and 65% of the Forbes’ Global 2,000 companies already rely on SAP

26

S/4HANA.”).

27
28

SAP also relies on United States v. Oracle Corp., 331 F. Supp. 2d 1098 (N.D. Cal. 2004)
to argue that a product market limited to “large” ERP customers is improper. Asker Reply at 3. In
27

1

Oracle, the court evaluated evidence after a two-week trial and rejected the plaintiffs’ product

2

market that only included products sold by Oracle, PeopleSoft, and SAP, and did not include mid-

3

market products. Oracle, 331 F. Supp. at 1158. The court rejected the proposed product market

4

in part because there was “no ‘quantitative metric’ that could be used to determine the distinction

5

between a high function product and a mid-market product.” Id. For example, it found that

6

Microsoft would be a viable substitute after examining its entry into the high function product

7

market. Id. at 1160. Today, however, Microsoft
Dkt. No. 543-44 at 7. Teradata asserts that the case is

8

therefore distinguishable because the ERP market has changed since Oracle, e.g., mid-market

10

competitors that the Oracle court relied on have now been acquired by larger ERP vendors or

11
United States District Court
Northern District of California

9

disappeared from the market. Asker Opp. at 11 (citing Oracle, 331 F. Supp. at 1159–61).

12

SAP contends that the case is persuasive for rejecting a proposed product market where, as

13

here, “there is no clear line separating those companies or the products they buy from others.”

14

Asker Reply at 3. Despite Asker’s admission that there is no common definition of “large

15

enterprises,” even among SAP’s own internal documents, he concludes, without further

16

explanation, that “‘large enterprises’ are generally companies with over 1,000 or 1,500 employees

17

and over 125 users of the ERP product.” Asker Rep. ¶ 50. He bases his conclusion on two SAP

18

documents that show that it markets different ERP products based on customer size, namely

19

S/4HANA to large enterprises that have over 1,000 employees. See Dkt. No. 537-6 at 572. But he

20

ignores the other SAP documents that indicate otherwise. As a result, Asker’s limitation of the

21

product market to “large enterprise” customers “stands on infirm ground” because Teradata

22

“makes no other effort to reconcile Dr. Asker’s distinct separate market with the broad continuum

23

of customers and varied and flexible approach to customer size taken by the industry.” Asker

24

Reply at 3.

25
26

iii.

Quantitative Analysis

More importantly, Asker’s quantitative analysis, which he uses to corroborate his

27

qualitative analysis, is flawed because contrary to his claims, Asker does not apply a “hypothetical

28

monopolist” test (“HMT”) as contemplated in the Department of Justice and the Federal Trade
28

1

Commission’s (“FTC”) Horizontal Merger Guidelines (the “Guidelines”).5 Asker Mot. at 9. This

2

test asks whether a hypothetical monopolist over a group of products could profitably impose a

3

small but significant and non-transitory increase in price (“SSNIP”) of 5%; if a significant number

4

of customers respond to a SSNIP by purchasing substitute products, then the SSNIP would not be

5

profitable and the market definition must be expanded to include those substitute products. See

6

Saint Alphonsus Med. Ctr.-Nampa Inc. v. St. Luke’s Health Sys., Ltd., 778 F.3d 775, 784 (9th Cir.

7

2015).

8

Asker states that he conducts a quantitative hypothetical monopolist test using aggregate

9

diversion (“ADR”) analysis of “Customer Relationship Management” (“CRM”) data from SAP
and Oracle, based on the number of times competitors are mentioned in sales representatives’ sales

11

United States District Court
Northern District of California

10

report.6 Asker Rep. ¶¶ 63, 64, 71. He opines that “CRM databases can be informative for market

12

definition to the extent that they provide some information on how frequently a business

13

encounters various potential competitors.” Asker Rep. ¶ 65. But he admits that CRM data “may

14

not always be a reliable indicator of the actual competitor faced by a company because the data is

15

often incomplete or the salesperson may have only a limited view into competition.” Id. He

16

concedes that, “CRM databases may also lack detail that allow precise evaluations of specific

17

markets” and that “this is the case in this matter, where every CRM data set [he has] examined has

18

limitations.” Id. As a result, he explains that he views the CRM data “ as merely providing a way

19

to corroborate the patterns that are present in the deposition testimony and documentary evidence

20
21
22
23
24
25
26
27
28

In its opposition to SAP’s motion for summary judgment, Teradata asserts that SAP’s
argument—“that a product market must include economic substitutes, i.e., products that would see
increased demand in response to a price increase in another product”—contravenes established
economic principles. Dkt. No. 542 at 30. It argues that the Guidelines “make clear that even if a
significant number of customers (even two thirds) would switch to other suppliers’ products in
response to a price increase, that does not require their inclusion in a properly defined relevant
market.” Id. (citing Dkt. No. 543-64 (“Guidelines”) § 4.1). But the Guidelines do not say this.
Instead, Section 4.1 of the Guidelines make clear that although a product market need not include
every competitor, it must “contain enough substitute products” to satisfy the SSNIP test.
Guidelines § 4.1; see Dkt. No. 552 at 20.
5

“Aggregate diversion analysis finds the threshold where a hypothetical monopolist imposing a
[SSNIP] would lose enough sales (‘actual loss’) compared the “critical loss” such that the SSNIP
would be unprofitable for the hypothetical monopolist. When the estimated actual loss due to a
SSNIP is smaller than the critical loss, the candidate market is considered a relevant antitrust
market.” Asker Rep. ¶ 71.
29
6

1

on the record” and expresses caution about using the data to form conclusions. Id. That said, he

2

concludes that the results of his analysis is “consistent with the deposition testimony and

3

documentary record that is my primary foundation for concluding that SAP and Oracle are each

4

other’s primary competitors for core ERP opportunities for large enterprises. Asker Rep. ¶ 71.

5

SAP’s expert, Stiroh, asserts that the ADR analysis is flawed because “[s]uch an analysis

6

requires data and inputs that can be used to reliably estimate actual lost sales and diversion ratios

7

in response to price changes of different ERP products” but such data was not available in this

8

case. Dkt. No. 554-9 (“Stiroh Rep.”) ¶ 58. Moreover, “[t]he CRM data that Asker uses to

9

calibrate his ADR model do not show actual diversion from one company to another, do not
reflect changes in purchasing patterns in response to price changes, and do not account for the

11

United States District Court
Northern District of California

10

competitive effects of emerging competitors and technologies or potential changes to SAP’s

12

expected competitive significance over the decade.” Id. ¶ 59.

13

Teradata contends that ADR analysis is an accepted methodology and that disputes

14

concerning an expert’s decision about what data to use in their analysis “bear on the weight, not

15

the admissibility, of expert testimony.” In re Qualcomm Antitrust Litig., 328 F.R.D. 280, 305

16

(N.D. Cal. 2018) (collecting cases). Although courts often conclude that “‘experts’ decisions

17

about what data to use’ in their analysis bear on the weight, not the admissibility, of expert

18

testimony,” id., ADR analysis has rarely been accepted by courts.

19

Teradata only cites to two district court cases that allowed an expert to use this

20

methodology to determine a product market. Asker’s ADR analysis, however, is distinguishable

21

because the experts in those cases relied on data sets that measured a customer’s response to

22

changes in price, e.g., actual win/loss data or bidding data, when using ADR analysis. In Federal

23

Trade Commission v. Sysco Corp., 113 F. Supp. 3d 1 (D.D.C. 2015), the FTC moved to enjoin a

24

potential merger between two food distribution companies. Sysco, 113 F. Supp. at 15. FTC’s

25

expert had “calculated the actual aggregate diversion based on three different data sets” and “built

26

a database for each company that tracked, for each bidding opportunity, the incumbent distributor,

27

the winning distributor, and the competing bidders.” Id. at 35. Like SAP in this case, the

28

defendants had objected to the expert’s methodology in part because the data on which he relied
30

1

did not describe whether the two companies “lost a customer for a price-based reason or some

2

reason having nothing to do with price.” Id. at 36. The court expressed its hesitancy to rely on the

3

expert’s findings but concluded that “when evaluated against the record as a whole” the expert’s

4

“conclusions are more consistent with the business realities of the food distribution market than”

5

the defendants’ expert. Id. at 37.

6

Similarly, in Federal Trade Commission v. Wilh. Wilhelmsen Holding ASA, 341 F. Supp.

7

3d 27 (D.D.C. 2018), the FTC moved to block a potential merger between two large providers of

8

marine water treatment chemicals. Wilh. Wilhelmsen, 341 F. Supp. at 39. FTC’s expert “used

9

three kinds of data—revenue information provided by marine suppliers, [] salesforce data, and [the
providers’] win-loss data.” Id. at 57. The court accepted the FTC expert’s market definition in

11

United States District Court
Northern District of California

10

part because the defendants’ expert did not contest that the FTC’s expert’s methodology was

12

flawed, did not present any alternative calculations or HMT results, and “the gap between critical

13

loss and aggregate diversion in every trial was so large as to ensure the stability of the HMT's

14

qualitative result against any but the gravest of statistical errors.” Id.
Teradata asserts that like both cases, Asker’s methodology “confirmed the market realities

15
16

evident in the record” and his findings ensured “the stability of the HMT’s qualitative result

17

against any but the gravest of statistical errors.” Asker Opp. at 12 (citing Dkt. No. 468-21 (“Asker

18

Reb. Rep.”) ¶ 89) (opining that “the CRM data would have to overstate aggregate diversion by a

19

factor of 2.5 to 3.2 for the conclusions for the aggregate diversion ratio analysis to change.”). But

20

Asker’s methodology is less reliable than those of the FTC experts because unlike the FTC

21

experts, Asker did not build a database of the type prices or rely on the price ultimately paid by the

22

customer. See Asker Depo. at 66–67 (“My recollection is that the final pricing is not available in

23

Oracle’s CRM data, and my recollection is that it’s not available in SAP’s CRM data.); id. at 67

24

(“Q: And in instances in which a competitor is listed, neither SAP’s nor Oracle’s CRM data

25

indicate the pricing offered by competitors; right? A: As I sit here today, that’s my recollection of

26

those data sets.”).7 His evaluation of CRM data did not and cannot consider pricing because the

27
7

28

In fact, as opposed to the “hundreds of thousands” of entries in the CRM data used by the expert
31

1

CRM data does not measure customer responses to changes in price. Id. at 68–69. As a result,

2

Asker’s ADR analysis of SAP’s CRM data cannot measure the most fundamental principle in

3

defining a market: cross-elasticity of demand.8

4

Teradata also contends that Asker conducts a robust quantitative analysis of SAP’s pricing

5

data to analyze price discrimination, which corroborates his conclusion that large enterprises form

6

a separate market. Asker Opp. at 10. As the Guidelines state, “[t]he possibility of price

7

discrimination influences market definition [], the measurement of market shares [], and the

8

evaluation of competitive effects.” Guidelines at 6. Teradata argues that even SAP’s expert

9

admits that there is price discrimination between large and small enterprises. Id. For pricing
discrimination to exist there must be (1) differential pricing; and (2) limited arbitrage. Guidelines

11
United States District Court
Northern District of California

10

at 6. According to Teradata, Stiroh’s pricing analyses
, and she does not dispute

12
13

that there is limited arbitrage. Asker Rep. ¶ 79; Dkt. No. 532-3 (“Stiroh Depo.”) at 109. But SAP

14

points out that its differential pricing is unrelated to customer size. Asker Reply at 4. “Per-unit

15

and per-user pricing confirm that SAP charges equivalent prices for large, mid-sized, and small

16

companies.”9 Id. (citing Stiroh Rep. ¶¶ 48–57). Further, Teradata’s argument that Asker does not

17

need to demonstrate that SAP currently charges higher prices of large customers and only needs to

18
19
20
21
22
23
24
25
26
27
28

in Sysco, Asker relied on fewer than 7,700 entries in SAP’s CRM data because almost 85% of the
data lacked any competitor information. Asker Rep. at 46, Exhibit 14; see Sysco, 113 F. Supp.
at 35.
Teradata contends that SAP’s expert, Murphy confirmed the proprietary of Asker’s
methodology. Asker Opp. at 3 n.9; Hearing Tr. at 29. But Asker relied on CRM data, which
addresses the “relative frequency with which those firms compete for Core ERP sales
opportunities,” whereas Murphy confirmed the methodology of looking at “win/loss” data to
indicate substitutability. Asker Opp. at 3; see Dkt. No. 530-36 (“Murphy Depo.”) at 44-45. SAP
also points out that Murphy does not use the CRM data to define any antitrust markets. Hearing
Tr. at 56.
8

SAP points out that in his initial report, Asker also analyzes “per-unit prices” for large
customers, based on the “‘size’ of installation (number of users),” and concludes that they vary.
See Asker Rep. ¶ 76; id., Ex. 7. Stiroh responded that the per user prices paid by small and
mid-sized customers also vary, but tend to be higher than those paid by large customers, and
therefore there is no evidence of price discrimination against large customers. Stiroh Rep. ¶¶ 51–
57. In his reply report, Asker criticizes Dr. Stiroh for using “per-user” prices and claims that the
appropriate measure is “total spend.” Asker Reb. Rep. ¶ 77. According to SAP, “neither Dr.
Asker nor Teradata explain this flip-flop.” Asker Reply at 5.
32
9

1

demonstrate that future price discrimination is “feasible” and “reasonably likely” undermines its

2

argument. Asker Reply at 4. If SAP can charge higher prices to larger customers in the future

3

because SAP negotiated different prices with customers in the past, it could also charge higher

4

prices to small customers. Id. But this does not make them antitrust markets; in the absence of

5

evidence of actual current price discrimination against large customers there is no basis to assume

6

that future price discrimination is feasible or likely. Stiroh Rep. ¶ 57.

7

Asker’s methodology in defining the tying market is unreliable. Contrary to Teradata’s

8

assertion, he does not measure the cross-elasticity of demand or the substitutability of products

9

based on reliable quantitative and qualitative analyses. Because his methodology for defining the
relevant tying market is unreliable, his conclusions that SAP has market power in his proposed

11
United States District Court
Northern District of California

10

market should also be excluded.

12

b.

Tied Product Market

13

Asker’s proposed tied market is “EDW products with OLAP capabilities for large

14

enterprises.” Asker Rep. ¶ 10. For the same reasons as above, SAP objects to this definition; his

15

qualitative analysis fails to consider the appropriate universe of documents and his quantitative

16

analysis is not a result of any reliable methodology. Asker Mot. at 14.

17

First, SAP asserts that Asker fails to consider all of the relevant documents when

18

determining the tied market. Id. For example, Asker excludes an EDW vendor Snowflake from

19

the market because he found it did not compete for EDW use cases for large enterprises. Asker

20

Rep. ¶ 91. But Teradata’s documents show that Snowflake was one of Teradata’s primary

21

competitors, if not the largest competitor, in 2019. See Dkt. No. 468-22 at 8-9
; Dkt. No. 468-23 at 4, 5–7

22
23

. But to claim that Asker ignored evidence regarding Snowflake is incorrect. Teradata

24

responds that this one document is contrary to the testimony of SAP’s competitive intelligence

25

team and SAP documents which characterize Snowflake as a
Opp. Asker at 17 (citing Dkt. No. 536-10 at 548; Dkt. No. 530-46 at

26
27

147–48
. Asker also analyzed the CRM data

28
33

1

himself and recognized that Snowflake was not a significant competitor. Asker Depo. at 97–100.

2

SAP also asserts that Asker’s conclusion that SAP’s HANA is in his tied market is

3

inconsistent with Teradata’s own admissions. Id. at 15. Its Senior Vice President of Global

4

Marketing, Chris Twogood, testified that Teradata does not compete frequently against HANA for

5

sales of EDW because HANA was not “designed to be an enterprise data warehouse,” and

6

Teradata does not consider SAP to be a primary competitor in the EDW space. Dkt. No. 468-17

7

(“Twogood Depo.”) at 20–22. Teradata points out that Twogood clarified, however, that once

8

SAP tied HANA to S/4HANA, SAP was able to “leverage[] all their ERP customers to grow

9

market share.” Dkt. No. 543-38 (“Twogood Depo.”) at 312, 315–16. But Twogood’s testimony
describes HANA being used as a transactional database under SAP ERP applications and

11

United States District Court
Northern District of California

10

S/4HANA, not as an EDW. Twogood Depo. at 314–15 (“[T]hey weren’t successful with HANA

12

only or HANA alone in the marketplace. So they bundled it in with their ERP solution and to

13

really ride a leverage for (verbatim) install base and force people to the HANA platform.”).

14

Teradata also emphasizes that SAP omits the testimony of Teradata witnesses and ordinary

15

course documents identifying SAP as a key EDW competitor. See, e.g., Dkt. No. 530-9 (“Boerger

16

Depo.”) at 303 (“IBM, Oracle, and SAP HANA compete for large enterprise data warehousing

17

types of customers”); Dkt. No. 543-26 (“Lea Depo.”) at 59 (“Q: Who are the primary competitors

18

to Teradata Vantage, based on your experience today, with large enough customers looking for an

19

EDW solution? A: It is more our traditional vendors, Oracle, IBM with Netezza and with Db2,

20

and HANA”); Dkt. No. 543-37 (“Susag Depo.”) at 20 (“Q: Who do you consider to be Teradata’s

21

main competitors in the enterprise data warehouse space? A: IBM, Oracle, SAP, Microsoft at the

22

lower end of the enterprise data warehouse space.”). As a result, Asker’s conclusions are not

23

inconsistent with Teradata’s own admissions.10

24
25
26
27
28

Further, Teradata asserts that SAP’s contention that HANA does not compete with Teradata
contradicts its prior statements. Hearing Tr. at 38–39; see, e.g., Dkt. No. 543-54 at 367 (a
technical textbook explaining to customers how to use HANA as an EDW). SAP responds that
the relevant question is not whether SAP tried to sell HANA as an EDW but whether customers
purchase and use HANA for this purpose. It asserts that Asker’s “false assumption that, because
SAP tried to market SAP as an EDW, customers necessarily use it as an EDW, runs throughout
Asker’s EDW-related opinions and renders them unreliable.” Reply at 8.
34
10

1
2

Asker Mot. at 17.
Teradata responds that Asker is not required to quantify damages for every participant in

3

the relevant market in order to opine that there are anticompetitive effects due to the tie. Asker

4

Opp. at 21–22. It points to SAP’s own economist, Murphy, who admits that database vendors like

5

Oracle and IBM are losing sales for database products that include OLAP/EDW capabilities as a

6

result of SAP’s tie and that SAP’s licensing restrictions have an anticompetitive effect, as they

7

reduce customers’ demand for using Teradata. Dkt. No. 530-36 (“Murphy Depo.”) at 145–46;

8

Dkt. No. 541-31 (“Murphy Rep.”) ¶ 224. But Murphy’s statement concerned transactional

9

databases, not products that include OLAP/EDW capabilities. Asker Reply at 11.

10

Teradata asserts that “Tying arrangements are forbidden on the theory that, if the seller has

United States District Court
Northern District of California

11

market power over the tying product, the seller can leverage this market power through tying

12

arrangements to exclude other sellers of the tied product.” Cascade Health Sols. v. PeaceHealth,

13

515 F.3d 883, 912 (9th Cir. 2008). It argues that “the injury caused by an unlawful tying

14

arrangement is ‘whether a total amount of business, substantial enough in terms of dollar-volume

15

so as not to be merely de minimis, is foreclosed to competitors by the tie.’” Datagate, Inc. v.

16

Hewlett-Packard Co., 60 F.3d 1421, 1425 (9th Cir. 1995) (internal citations omitted). For

17

example, Asker relies on SAP revenue data
as evidence of a

18
19

distortion due to a tie and “not simply the result of competition on the merits. Asker Rep. ¶¶

20

145,147. Asker reviewed evidence that also showed that customers are not allowed “to use

21

S/4HANA and a third-party EDW without also purchasing HANA. In particular, the customer

22

must still purchase the ‘full use’ HANA license in order for it to use a competing third-party

23

EDW.” Id. ¶ 158. He opined: “If a condition of purchasing a product is the simultaneous

24

purchase of a product of a competitor, economic reasoning indicates that the product’s competitive

25

position is weakened.” Id. Teradata therefore argues that Asker’s opinions about the alleged harm

26

to competition are proper.

27

There needs to be a showing of “substantial” harm; de minimis harm is not enough under

28
36

1

rule of reason analysis.12 See Qualcomm, 969 F.3d at 991 (Under § 1, “the plaintiff has the initial

2

burden to prove that the challenged restraint has a substantial anticompetitive effect that harms

3

consumers in the relevant market.”). Asker failed to show this; he presented no evidence of harm.

4

In addition, SAP asserts that Asker relies on a series of unwarranted assumptions,

5

specifically that because HANA has OLAP capabilities, it is necessarily always sold as an EDW.

6

Asker Reb. Rep. ¶ 127. When HANA is “sold together with S/4HANA, [it] is almost always

7

bundled with S/4HANA under a runtime license,” which precludes use of HANA as an EDW.

8

SMSJ at 28. With a runtime license, HANA can be used only to support the SAP application

9

running on top of it; in other words HANA is the transactional database that supports the
application, S/4HANA. Stiroh Decl. ¶ 176. It cannot be an EDW, as defined by Teradata,

11

United States District Court
Northern District of California

10

because it does not bring data from multiple sources across an enterprise and then use

12

sophisticated analytics tools to conduct analysis of that combined data. See SAC ¶ 16.

13

Teradata does not dispute that approximately 88% of SAP’s customers have purchased

14

HANA with a runtime license. SMSJ at 29. And it does not present any evidence that a single

15

customer has taken S/4HANA together with HANA pursuant to a full use license and used that

16

HANA installation as an EDW. Id. Teradata does not provide any instance where a customer

17

who used Teradata Database replaced it with HANA for the same purpose. Because these

18

undisputed facts render Asker’s opinion unreasonable and because Teradata’s opposition is based

19

on an incorrect legal standard, SAP’s motion to exclude portions of Asker’s opinions related to

20

alleged harm to competition is GRANTED.
SAP also objects to Asker’s opinions that HANA’s adoption is not being driven by any

21
22

procompetitive benefits of the alleged tie. Asker Mot. at 4. According to SAP, Asker lacks the

23

expertise necessary to evaluate evidence of the design benefits of S/4HANA and admits that he

24

did not understand much of the relevant evidence. Id.; see Asker Rep. ¶ 171 (“I do not have the

25

expertise to evaluate whether there is a technical benefit from combining S/4HANA with

26

HANA.”). Teradata responds that Asker is not opining that there are no technical benefits for the

27
28

12

See infra Part I.B.4.a for discussion on the application of either the rule of reason or per se
analysis to this case.
37

1

integration of S/4HANA and HANA but rather that the “documentary and deposition evidence

2

indicates that there is no technical reason for the tie and that the decision to tie was made by

3

SAP’s board of directors on business grounds.” Asker Opp. at 22–23. SAP replies that this

4

distinction is nonsensical because if S/4HANA is designed to work with HANA such that

5

S/4HANA “is wholly incompatible with other transactional databases” as Teradata alleges, then

6

this is the technical reason why customers must license S/4HANA and HANA together. Asker

7

Reply at 12.
Teradata asserts that no SAP witness or expert has provided a technical justification—or

8
9

any justification—for tying S/3HANA to HANA’s analytical capabilities.13 Asker is qualified to
assess the economic realities of SAP’s business decisions to tie S/4HANA to the OLAP

11

United States District Court
Northern District of California

10

capabilities of HANA. Asker Depo. at 226. His testimony related to the alleged lack of

12

procompetitive benefits should not be excluded.14

13

2.

Motion to Exclude Mehrotra Testimony

14

Teradata’s moved to exclude three out of four of SAP’s expert, Dr. Sharad Mehrotra’s

15

opinions in sections VI and VII of his report, which rebut the opinions of Teradata’s technical

16

expert Hosagrahar Jagadish: (1) that SAP could achieve significant benefits by designing

17

S/4HANA for its HANA database product; (2) that SAP could not have achieved these same or

18

similar benefits by designing S/4HANA to run on other databases; and (3) that “porting”

19

S/4HANA to third-party databases would be challenging. Dkt. No. 472 (“Mehrotra Mot.”) at 1. It

20

contends that Mehrotra lacks the necessary factual foundation for these opinions and that he did

21

not follow any reliable methodology in reaching them. Id. It argues that Mehrotra never reviewed

22

any of the source code for S/4HANA or HANA, has never used or examined the products, has

23

never used or examined the third-party database products he compares to HANA, disregarded

24

testimony from SAP executives, and relies on cherry-picked documents for sweeping conclusions.

25
26
27
28

The reason for this lack of evidence, however, is Teradata’s allegation of a new tying theory
during the summary judgment briefing. See infra Part I.B.3.
13

Because Teradata’s tying claim fails, see infra Part I.B.4, I will not address SAP’s motion to
exclude Asker’s analysis of lost profits due to the alleged tying arrangement.
38
14

1

Id. at 2. SAP responds that Mehrotra’s methodology is reliable, and that the rest of Teradata’s

2

arguments go to the weight and not the admissibility of evidence. Dkt. No. 533 (“Mehrotra

3

Opp.”) at 1.

4

First, Teradata argues that Mehrotra’s opinion in section VI of his report—that SAP was

5

able to realize multiple technical and practical benefits by designing S/4HANA to work closely

6

with HANA—should be excluded because it is unsupported and unreliable. Mehrotra Mot. at 2.

7

Mehrotra admits that he has never used or even examined the S/4HANA or HANA software, any

8

SAP ERP applications, or source code. Dkt. No. 483-4 (“Mehrotra Depo.”) at 42–44 (“Q: Have

9

you ever used S/4 Hana? A: Personally, no . . . Q: Have you ever used any part of SAP’s Business
Suite? A: No, I have not. Q: Have you ever used SAP’s HANA database? A: I have personally

11

United States District Court
Northern District of California

10

not used SAP HANA database. It’s not an open source database. So it’s not free.”). When asked

12

whether he had ever used any SAP ERP software, he responded, “No. I am an academic. We

13

normally do not deal with the operational aspect of the problem, so where companies sort of run

14

these things.” Id. at 43–44. Teradata contends that SAP “is silent on the issue” and “cites no case

15

where an expert was allowed to opine on the design, capabilities, performance, and compatibility

16

of products without ever having even looked at them.” Dkt. No. 551-4 (“Mehrotra Reply”) at 1.

17

SAP does not respond directly to the argument that Mehrotra did not use any of the SAP

18

products. It does assert, however, that Mehrotra reviewed the architecture and design of

19

S/4HANA, relying on a series of 28 architectural guidelines of all of the versions starting with the

20

first in May 2014 through March 2019. Mehrotra Opp. at 18. It also asserts that it was unfeasible

21

and not useful for Mehrotra to review all 300 million lines of source code. Id. at 10. Teradata

22

responds that the argument that Mehrotra “cannot look at everything does not mean it is proper to

23

look at little to nothing” and at the very least, he “should have identified some representative

24

queries in S/4HANA that would require porting to third-party database.” Mehrotra Reply at 1.

25

Mehrotra did not have to review the source code because he reviewed the architecture and

26

design of S/4HANA instead. Mehrotra Depo. at 147. For example, he stated that he is “intimately

27

aware” of the “architectural aspects of things, but [] not [as] aware of the exact software

28

implementation.” Id. But he testified that source code is simply “one aspect of the system
39

analysis” and that he understood “the system and its properties” by the architectural diagrams. Id.

2

SAP contends that reviewing S/4HANA and HANA at a design and architectural level is “a

3

common and accepted method of software analysis,” as evidenced by the academic and expert

4

works that Mehrotra cites in Appendix B of his report, which “rely on exactly this architectural

5

level of analysis.” Mehrotra Opp. at 11 (citing Dkt. No. 483-3 (“Mehrotra Rep.”), Appendix B).

6

Teradata responds that SAP does not point to any specific methodologies that are supposedly

7

found in any of these works. Mehrotra Reply at 2. It asserts that “[t]he reality is that the cited

8

works provide only general software background, not any methodology for the sort of software

9

analysis required in this case.” Id. SAP also, however, contends that Mehrotra’s reliance on the

10

architecture and design of S/4HANA is proper as evidenced by Teradata’s expert Jagadish also

11

United States District Court
Northern District of California

1

relying on architecture-level analyses. Mehrotra Opp. at 11 (citing Dkt. No. 531-21 (“Jagadish

12

Rep.”) ns. 290–92, 306, 319–33, 336, 341–45, 357–60, 365–70). Teradata does not respond to

13

this argument.15
The following cases provide a helpful analysis of whether Mehrotra’s approach is proper.

14
15

Teradata relies on a Seventh Circuit case in support of its argument that Mehrotra’s approach is

16

flawed, but the case is distinguishable. In Autotech Tech. Ltd. P’ship v. Automationdirect.com,

17

471 F.3d 745 (7th Cir. 2006), an expert testified “[b]ased on his 26 years of experience in software

18

development, review of the EZTouch software, and review of advertisements about C–More . . .

19

that the features of C–More could not be developed independently of EZTouch” but he had “never

20

conducted tests on the product.” Autotech, 471 F.3d at 749. The Seventh Circuit affirmed the

21

district court’s decision that this methodology was unreliable because “computer experts must do

22

more than read advertisements.” Id. The court held that “[t]o qualify as an expert on software, an

23

expert should, at a minimum, examine the product and software upon which the expert bases his

24

opinion.” Id. In this case, while Mehrotra did not use the product or examine the source code, he

25
26
27
28

15

Teradata does assert that unlike Mehrotra, Jagadish examined the source code and software.
Mehrotra Mot. at 10. SAP responds that nowhere in the sections of Jagadish’s report, to which
Mehrotra responds, does Jagadish refer to S/4HANA source code. Mehrotra Opp. at 13–14.
Instead, Jagadish discusses the source code only in relation to trade secrets. Compare Jagadish
Rep. § X.C ¶¶ 247–75 (discussing trade secrets) with Jagadish Rep. § X.E ¶¶ 336–77 (discussing
antitrust opinions).
40

1

reviewed the architecture and design of S/4HANA over the course of five years.
Iconics, Inc. v. Massaro, 266 F. Supp. 3d 461 (D. Mass. 2017) is more factually analogous

2
3

to the case here. In Iconics, the court declined to exclude the expert’s testimony regarding the

4

“core architecture” of the software products, which was based on “three architectural diagrams,”

5

even though the expert failed to inspect the software code or review technical documents. Iconics,

6

266 F. Supp. at 470. The court held that because the expert “illuminates aspects of the core

7

architecture trade secret” any challenge went to the credibility of the testimony and not

8

admissibility. Id. Teradata contends that Iconics is distinguishable because there the expert relied

9

on the same architecture documents that the plaintiff cited to whereas here Mehrotra relies on
“hand-selected” documents by SAP’s counsel and not Teradata’s materials. Mehrotra Mot. at 4–5.

11

United States District Court
Northern District of California

10

But in Iconics the product at issue belonged to the plaintiff whereas here the S/4HANA product

12

belongs to SAP. It is unclear why it is improper for Mehrotra to rely on technical documents from

13

SAP itself when reviewing its product. Mehrotra Opp. at 18. As the Iconics court held,

14

“[r]egardless of the benefits of any alternative approaches,” for example those found in Jagadish’s

15

report, Mehrotra’s opinion is sufficiently reliable. Iconics, 266 F. Supp. at 470. “Any questions

16

on the comparative weight or credibility of these two analyses are questions for a jury to resolve.”

17

Id.

18

As for Teradata’s other objections—“that Mehrotra could not identify with sufficient

19

specificity the academic literature he relied on, did not cite to the particular documents that

20

Teradata thinks he should have, and did not interview the individuals that Teradata thinks he

21

should have”—all go to the weight of his testimony are not grounds for excluding his opinions.

22

Mehrotra Opp. at 16; In re Korean Ramen Antitrust Litig., 281 F. Supp. 3d 892, 931 (N.D. Cal.

23

2017) (holding that an expert’s failure to “address (or review) deposition testimony where

24

defendants' employees testified to matters that purportedly undermine some of his opinions or

25

assumptions does not make his testimony excludable. Those are grounds for cross-examination.”).

26

For example, Teradata asserts that Mehrotra’s opinions are flawed in part because he did not

27

interview the SAP employees that Rudolf Hois spoke with in preparation for his 30(b)(6)

28

deposition. Mehrotra Mot. at 4. SAP responds that Mehrotra did not have to speak with the
41

1

individuals that Hois spoke with because Hois’s experience is in the area of ERP applications

2

whereas Mehrotra’s experience is in databases. Mehrotra Opp. at 23. SAP also points out that

3

Mehrotra had the deposition transcripts of Hois and other SAP employees related to the interface

4

of S/4HANA with HANA. Id. (citing Mehrotra Rep., Appendix B). Its argument is well-taken:

5

Teradata’s motion to exclude Section VI of Mehrotra’s expert report is DENIED.

6

Finally, Teradata moves to exclude section VII of Mehrotra’s report which opines that
SAP could not have realized the same benefits by designing S/4 for multiple databases and that

8

porting S/4HANA to another database would be challenging and unpredictable because it is

9

unsupported and unreliable. Mehrotra Mot. at 8. Teradata asserts that in support of his opinions,

10

Mehrotra could and should have reviewed some of the analytical queries in S/4HANA in order to

11
United States District Court
Northern District of California

7

provide at least one specific example of a query that purportedly requires the use of HANA, and

12

not another database. Mehrotra Reply at 5–6.

13
14

SAP contends that it is unclear how Mehrotra was supposed to do this or what purpose it
would serve. Mehrotra Opp. at 14. Mehrotra and Jagadish do not dispute that
Id. at 7;

15
16

Mehrotra Reply at 5. But

17
18
19
20
Mehrotra Opp. at 14 (citing Dkt. No. 531-8 (“Hois Depo.”) at

21
22

14–16, 70–74.

23
24

Mehrotra Depo at 131–32. SAP points out that review of this code is unnecessary

25

because both Jagadish and Mehrotra agree that the key issue is not how much code must be ported

26

to another database but how difficult it would be. Mehrotra Opp. at 16 (citing Mehrotra Depo. at

27

186–88; Dkt. No. 531-14 (“Jagadish Depo.”) at 227).

28

To determine how difficult porting would be, Mehrotra relies on “SAP’s past experience
42

1

parroted from SAP documents selected by SAP’s counsel.” Mehrotra Reply at 12. It also points

2

out that Mehrotra failed to reconcile this allegation with results of

3

Jagadish’s opening report. Id. (citing Jagadish Rep. ¶¶ 342, 372). But again, Mehrotra’s failure to

4

consider contrary evidence goes to weight and not admissibility. Teradata’s motion to exclude

5

portions of Mehrotra’s expert report is therefore DENIED.

6

3.

as noted in

Teradata’s Objections to SAP’s Reply Evidence

7

The final preliminary matter I must address is Teradata’s objections to SAP’s reply

8

evidence, namely its declaration of Rudolph Hois, Dkt. No. 552-1 (“Hois Declaration”). Dkt. No.

9

568-4 at 1. Teradata asserts that I should strike the declaration because it is impermissible and
highly prejudicial. Id. The Hois Declaration concerns a key issue underlying Teradata’s theory—

11
United States District Court
Northern District of California

10

that SAP’s requirement that S/4HANA customers license HANA’s analytical capabilities violates

12

federal antitrust law. Id. Teradata argues that even if it were true that SAP first heard of this

13

theory from Asker’s reply report, as it claims, SAP should have submitted the Hois declaration

14

with its motion, a full month after Asker’s reply report. Id. Instead, it asserts that SAP improperly

15

waited until its reply brief to submit the evidence. Id.

16

SAP responds that the Hois Declaration was necessary because it was not aware of

17

Teradata’s new theory until its opposition to SAP’s summary judgment motion. Dkt. No. 585 at 1.

18

There, Teradata abandoned the tying theory pleaded in its complaint and asserted a new one, after

19

the close of fact discovery. Dkt. No. 585 at 1. In the SAC, Teradata alleged that SAP tied

20

S/4HANA to HANA by making it “wholly incompatible with other transactional databases,”

21

forcing customers that purchase S/4HANA to also adopt HANA. SAC ¶¶ 89, 132. But in SAP’s

22

summary judgment motion, SAP showed how it and other leading vendors achieved

23

procompetitive benefits by integrating their ERP applications with their databases. Dkt. No. 585

24

at 1. Then, in its opposition, Teradata argued that the tie was different; it was between S/4HANA

25

and HANA’s analytical capabilities, which offer EDW functions. Dkt. No. 542 at 26. As a result,

26

the mechanism at issue is no longer a technological incompatibility but licensing terms that SAP

27

allegedly forces upon its customers. Id. at 31. Teradata asserts that SAP has failed to show

28

procompetitive justifications for the licensing practices that tie S/4HANA to HANA’s analytical
44

1
2

capacities. Id. at 27.
In support of its argument that Teradata changed its tying theory, SAP points to Asker’s

3

opening expert report that expressly and repeatedly defined the “tied product” to mean HANA, not

4

its analytical capabilities. Asker Rep. ¶¶ 5–6, 35. But in his reply report, Asker suggests that SAP

5

should have to justify the tie of S/4HANA to the EDW capabilities of its HANA database. Asker

6

Reb. Rep. ¶ 4. SAP contends that “this shift did not put SAP on notice that Teradata had changed

7

its legal theory regarding the alleged tie” and therefore it did not file the Hois Declaration with its

8

summary judgment motion. Dkt. No. 585 at 2.

9
10

United States District Court
Northern District of California

11
12
13
14
15
16

Teradata maintains that its theory has not changed. It emphasizes paragraph 95 of the
SAC, which states:
“SAP’s Top-Tier ERP Applications customers were free to choose
how to manage their data needs, those locked-in customers will now
be forced to adopt HANA.
Given the costs of licensing,
implementing, and maintaining EDAW products, the vast majority of
large-scale customers will have no choice but to abandon their prior
EDAW providers because they cannot support dual EDAW providers.
Thus, because HANA purports to offer some or all of the functionality
offered by Teradata, SAP is effectively coercing its customers into
leaving Teradata and adopting the full stack of SAP products
(including HANA).”
Dkt. No. 599 (quoting SAC ¶ 95) at 1. Teradata explains that throughout its complaint, the

17
18
19
20
21

“functionality offered by Teradata” that HANA purports to offer to replace Teradata’s products is
HANA’s analytical (or EDAW) functionality. Id.; see, e.g., SAC ¶ 45 (HANA purports to provide
“EDAW functionality that SAP claims can enable enterprise analytics similar to those offered by
Teradata” and “[t]hus, with HANA . . . SAP now positions itself as a direct competitor in the
EDAW market”). It contends that it has never alleged that HANA’s transactional functionality

22
competes with EDAW products. Dkt. No. 467 at 4.
23

Teradata’s arguments do not address SAP’s point—that Teradata initially challenged the

24
technological integration of the ERP application and HANA, not the licensing practice. Teradata
25
26
27

asserts that its allegation that SAP’s “sales practice” is “directly contrary to the practices of other
ERP applications” is regarding its licensing. But in actuality, the alleged “sales practice” in the
SAC does not refer to licensing but a design change, i.e., “tying upgrades of customers’ ERP

28
45

1

Applications to customers’ adoption of HANA (while ending support for older versions of ERP

2

Applications).” SAC ¶ 58. Furthermore, references to “licensing” in the SAC concern the exit

3

fee, not the licensing of HANA’s analytical capabilities. See, e.g., SAC ¶ 90 (“SAP’s licensing

4

agreements further restrict the ability of customers to read and copy S/4HANA ERP data to any

5

other database); id. ¶ 151 (“This rate will only rise more rapidly as more customers upgrade to

6

S/4HANA and are foreclosed from either licensing alternative EDAW products or accessing their

7

SAP ERP data for use with Teradata’s EDAW products.”).

8

Teradata also claims that Hois Declaration contradicts his deposition as a corporate witness

9

and should be struck as undisclosed expert testimony. Dkt. No. 568-4 at 3–5. It asserts that in his
declaration, Hois explains HANA’s capabilities as unique and opines about the comparisons

11

United States District Court
Northern District of California

10

between HANA and other databases. Hois Decl. ¶¶ 4–6. But during his deposition he repeatedly

12

claimed that he lacked the requisite knowledge or expertise to compare the databases and deferred

13

to other experts. Dkt. No. 568-6 (“Hois Depo.”) at 64; Dkt. No. 568-8 (“Hois Depo.”) at 12, 14,

14

18. In the his declaration, however, Hois is not comparing databases; instead, he explains a

15

feature of Oracle databases in a manner that is consistent with his deposition testimony. Compare

16

Hois Decl. ¶ 6 with Hois Depo. at 34–35. Moreover, Hois’s high-level opinions are based on his

17

personal knowledge and therefore are proper. Hois Decl. ¶¶ 2–3.

18

SAP contends that “Teradata cannot oppose summary judgment on the basis of an unpled,

19

and prejudicially-late change in theory.” Dkt. No. 552 at 13 (citing Navajo Nation v. U.S. Forest

20

Serv., 535 F.3d 1058, 1080 (9th Cir. 2008) (“where . . . the complaint does not include the

21

necessary factual allegations . . . raising such a claim in a summary judgment motion is

22

insufficient to present the claim to the district court”)). I agree. The Hois Declaration is proper,

23

even though it is new evidence, as a “reasonable response to the opposition.” Hodges v. Hertz

24

Corp., 351 F. Supp. 3d 1227, 1249 (N.D. Cal. 2018).

25
26

4.

Motion for Summary Judgment

I will now turn to SAP’s motion for summary judgment on Teradata’s tying claim. To

27

state a sufficient tying claim under Section 1 of the Sherman Act, Teradata must prove: (1) a

28

contract, combination or conspiracy among two or more persons or distinct business entities; (2)
46

1

by which the persons or entities intended to harm or restrain trade or commerce []; (3) which

2

actually injures competition.” Kendall v. Visa U.S.A., Inc., 518 F.3d 1042, 1047 (9th Cir. 2008).

3

a.

Per Se or Rule of Reason Analysis

4

The first dispute between the parties is whether the per se rule or rule of reason test applies

5

in this case. To determine whether a practice unreasonably restrains trade, courts sometimes apply

6

a “rule of reason” analysis. Bhan v. NME Hosps., Inc., 929 F.2d 1404, 1410 (9th Cir. 1991).

7

Under the rule of reason test, courts “analyze the degree of harm to competition along with any

8

justifications or pro-competitive effects to determine whether the practice is unreasonable on

9

balance. The focus is on the actual effects that the challenged restraint has had on competition in a
relevant market.” Id. “Some practices, however, are so likely to interfere with competition that

11

United States District Court
Northern District of California

10

they violate the Sherman Act per se. In these cases, [courts] do not require evidence of any actual

12

effects on competition because [they] consider the potential for harm to be so clear and so great.”

13

Id. Under the per se test, Teradata must prove: (1) that the defendant tied together the sale of two

14

distinct products or services; (2) that the defendant possesses enough economic power in the tying

15

product market to coerce its customers into purchasing the tied product; and (3) that the tying

16

arrangement affects a “not insubstantial volume of commerce” in the tied product market.

17

Cascade Health, 515 F.3d at 913.

18

“Restraints that are not unreasonable per se are judged under the ‘rule of reason.’” Fed.

19

Trade Comm'n v. Qualcomm Inc., 969 F.3d 974, 989 (9th Cir. 2020). “[N]ovel business

20

practices—especially in technology markets—should not be conclusively presumed to be

21

unreasonable and therefore illegal without elaborate inquiry as to the precise harm they have

22

caused or the business excuse for their use.” Id. at 990–91 (internal quotation marks omitted).

23

“Because innovation involves new products and business practices, courts[’] and economists’

24

initial understanding of these practices will skew initial likelihoods that innovation is

25

anticompetitive and the proper subject of antitrust scrutiny.” Id. at 991. In this case, the rule of

26

reason applies because this is not a case that “has so little redeeming virtue, and that there would

27

be so very little loss to society from its ban, that an inquiry into its costs in the individual case [can

28

be] considered [] unnecessary.” United States v. Microsoft Corp., 253 F.3d 34, 94 (D.C. Cir.
47

1
2

2001) (internal quotation marks and citations omitted).
Contrary to Teradata’s argument, there are procompetitive justifications from SAP’s

3

design of S/4HANA to run on HANA rather than on multiple databases. Reply SMSJ at 14. For

4

example, under Teradata’s original theory, SAP’s expert, Mehrotra, explains how SAP achieved

5

efficiency gains such as improved performance and functionality with S/4HANA by designing it

6

for only HANA. Mehrotra Rep. ¶¶ 138–98. Under its new theory, Teradata implies that there is

7

no reason SAP could not separately license HANA’s analytical capabilities, SMSJ Opp. at 35, but

8

SAP explains that unlike Oracle and Microsoft, SAP lacks the ability to license analytical and

9

transactional functionalities separately because they operate on the same set of data and are
intertwined. Hois Decl. ¶¶ 5–6. It asserts that SAP “achieved procompetitive benefits by

11

United States District Court
Northern District of California

10

designing S/4HANA to run on all of HANA, including its analytical capabilities.” SMSJ Reply at

12

15. That Teradata contends that the design of S/4HANA has no efficiency gains is irrelevant to

13

the question of whether the rule of reason applies. Instead, these “purported efficiencies suggest

14

that judicial ‘experience’ provides little basis for believing that” SAP’s S/4HANA “lacked any

15

redeeming virtue and therefore should be presumed unreasonable.” Microsoft, 253 F.3d at 90–91.

16

Rule of reason applies in this case.

17

b.

Failure to Properly Define a Tied or Tying Market

18

That said, under either test, Teradata’s tying claim fails. As established above, because

19

Teradata has failed to properly define a tied market, there is no triable issue of fact whether the

20

alleged tying arrangement harmed competition in the tied market under the rule of reason analysis.

21

Likewise, because Teradata has failed to properly define a tying market, there is no triable issue of

22

fact whether SAP has market power in a properly-defined tying market. See Truck-Rail Handling

23

Inc. v. BNSF Ry. Co., 2005 WL 8178364, at *8 (N.D. Cal. Mar. 8, 2005) (granting defendants’

24

motion for summary judgment on market definition because plaintiff’s evidence did not “assist in

25

evaluating cross-elasticity of supply and demand”).

26

Teradata contends that the issue of market definition should be decided by a jury. Opp.

27

SMSJ at 32; see High Tech. Careers v. San Jose Mercury News, 996 F.2d 987, 990 (9th Cir. 1993)

28

(“The process of defining the relevant market is a factual inquiry for the jury.”). But where there
48

1

is an absence of evidence to support Teradata’s claim that SAP competes in the purported tying or

2

tied market, summary judgment is appropriate. In Rebel Oil Co. v. Atl. Richfield Co., 51 F.3d

3

1421 (9th Cir. 1995), the Ninth Circuit acknowledged “that the definition of the relevant market is

4

a factual inquiry for the jury, and the court may not weigh evidence or judge witness credibility.”

5

Rebel Oil, 996 F.2d at 1435. It held, however, “that an issue is factual does not necessarily

6

preclude summary judgment. If the moving party shows that there is an absence of evidence to

7

support the plaintiff's case, the nonmoving party bears the burden of producing evidence sufficient

8

to sustain a jury verdict on those issues for which it bears the burden at trial.” Id. It also noted

9

that when, as here, “an expert opinion is not supported by sufficient facts to validate it in the eyes
of the law, or when indisputable record facts contradict or otherwise render the opinion

11

United States District Court
Northern District of California

10

unreasonable, it cannot support a jury's verdict” and therefore summary judgment is appropriate.

12

Id. at 1436 (quoting Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 242

13

(1993)). Accordingly, SAP’s motion for summary judgment on Teradata’s tying claim is

14

GRANTED.16

15

II.

TERADATA’S MOTION FOR SUMMARY JUDGMENT
Teradata moves for summary judgment against SAP’s counterclaims, which contend that

16
17

Teradata infringes its ’321 Patent,’179 Patent, and ’421 Patent. Dkt. No. 472 (“TMSJ”) at 1.

18

Teradata asserts that the claims of the ’321 Patent are invalid because they are directed to

19

patent-ineligible subject matter under 35 U.S.C. § 101. Id. Teradata also asserts that SAP is not

20

entitled to damages for the alleged infringement of the ’179 and ’421 Patents before May 19,

21

2019, when it first informed Teradata of its infringement allegations because SAP had failed to

22

give notice to the public that its products practice the claims of these patents prior to then. Id.; see

23

35 USC § 287. SAP does not oppose Teradata’s motion for summary judgment against an award

24

of damages for infringement of the ’179 and ’421 Patents before May 21, 2019. Dkt. No. 520

25

(“Opp. TMSJ”) at 1. SAP does, however, contend that the ’321 Patent is valid. Id.

26
27
28

16

Moreover, even if the tied market definition was proper, summary judgment would still be
appropriate because Teradata cannot show that SAP has caused actual injury to competition in a
market for “EDW products with OLAP capabilities for large enterprises.” See supra Part I.B.1.b.
49

1

A.

Legal Standard

Under Section 101 of the Patent Act, “[w]hoever invents or discovers any new and useful

2
process, machine, manufacture, or composition of matter, or any new and useful improvement
3

thereof, may obtain a patent therefor . . . . .” 35 U.S.C. § 101. The Supreme Court “has long held

4
that this provision contains an important implicit exception: Laws of nature, natural phenomena,
5
6
7

and abstract ideas are not patentable.” Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 573 U.S. 208, 216
(2014). The reason for the exception is clear enough—“such discoveries are manifestations
of . . . nature, free to all men and reserved exclusively to none.” Mayo Collaborative Servs. v.

8
Vometheus Labs., Inc., 566 U.S. 66, 71 (2012) (internal quotation marks and citations omitted).
9
The boundaries of the exception, however, are not so clear.
10

The Alice court highlighted “the concern that drives this exclusionary principle as one of

11

United States District Court
Northern District of California

pre-emption.” Alice, 573 U.S. at 216 (noting the delicate balance inherent in promoting progress,
12
the primary object of patent law, and granting a monopoly, the means for accomplishing that
13
goal). In other words, patents that seek to wholly preempt others from using a law of nature or an
14
15

abstract idea—“the basic tools of scientific and technological work”—are invalid. Id.
“Accordingly, in applying the § 101 exception, we must distinguish between patents that claim the

16
buildin[g] block[s] of human ingenuity and those that integrate the building blocks into something
17

more, thereby transform[ing] them into a patent-eligible invention.” Id. at 217 (internal quotation

18
marks and citations omitted).
19
20
21
22

In evaluating whether claims are patent-eligible, I must first “determine whether the claims
at issue are directed to one of those patent-ineligible concepts.” Alice, 573 U.S. at 217. “[T]he
‘directed to’ inquiry applies a stage-one filter to claims, considered in light of the specification,
based on whether their character as a whole is directed to excluded subject matter.” Enfish, LLC v.

23
Microsoft Corp., 822 F.3d 1327, 1335 (Fed. Cir. 2016) (internal quotation marks omitted).
24
Although there is no bright-line rule for determining whether a claim is directed to an abstract
25
idea, courts have articulated some guiding principles. When evaluating computer-related claims,
26
27

courts look to whether the claims “improve the functioning of the computer itself,” Alice, 573 U.S.
at 225, or whether “computers are invoked merely as a tool” to implement an abstract process.

28
50

1
2

Enfish, 822 F.3d at 1336.
If the claims are directed to a patent-ineligible concept, I must then “consider the elements

3

of each claim both individually and ‘as an ordered combination’ to determine whether the

4

additional elements ‘transform the nature of the claim’ into a patent-eligible application.” Id. at

5

1334 (internal citations omitted). This step entails the “search for an inventive concept—i.e., an

6

element or combination of elements that is sufficient to ensure that the patent in practice amounts

7

to significantly more than a patent upon the [ineligible concept] itself.” Alice, 573 U.S. at 217–18

8

(internal quotation marks and citations omitted). An inventive concept “cannot simply be an

9

instruction to implement or apply the abstract idea on a computer” and “must be significantly
more than the abstract idea itself.” BASCOM Glob. Internet Servs., Inc. v. AT&T Mobility LLC,

11

United States District Court
Northern District of California

10

827 F.3d 1341, 1350 (Fed. Cir. 2016).

12

“For the role of a computer in a computer-implemented invention to be deemed

13

meaningful in the context of this analysis, it must involve more than performance of

14

well-understood, routine, [and] conventional activities previously known to the industry.” Content

15

Extraction & Transmission LLC v. Wells Fargo Bank, N.A., 776 F.3d 1343, 1347–48 (Fed. Cir.

16

2014). “[T]he mere recitation of a generic computer cannot transform a patent-ineligible abstract

17

idea into a patent-eligible invention.” Id. at 1348. However, “an inventive concept can be found

18

in the non-conventional and non-generic arrangement of known, conventional pieces.” BASCOM,

19

827 F.3d at 1350.

20

B.

21

The ’321 Patent is titled “Systems and Methods for Data Processing.” Dkt. No. 124-1

Whether the ’321 Patent Is Invalid Under 35 U.S.C. § 101

22

(“’321 Patent”). SAP alleges that Teradata infringes at claims 1, 2, and 4 of the ’321 Patent. Dkt.

23

No. 461 at 1. Independent claim 1 and dependent claim 2 recite:

24
25
26
27
28

“1. A data processing method comprising:
providing a set of database tables in a data warehouse, each
database table being assigned to an entity type and storing
entities of its entity type;
providing a set of online analytical processing cubes in a data
warehouse, each online analytical processing cube specifying a
layout for transactional data storage;
providing at least one application program for processing at least
one class of database tables and at least one class of online
51

1
2
3
4
5
6
7
8
9
10

United States District Court
Northern District of California

11
12
13
14
15
16
17
18
19
20
21
22
23
24

analytical processing cubes;
mapping a sub-set of the set of database tables to the at least one
class of database tables, the sub-set of database tables
comprising database tables of one or more entity types;
mapping a sub-set of the set of online analytical processing cubes
to the at least one class of online analytical processing cubes;
invoking an online analytical processing component to fill the
online analytical processing cubes with transactional data;
processing the entities stored in the sub-set of database tables and
the transactional data stored in the sub-set of online analytical
processing cubes by the application program; and
providing analysis of the entities and the transactional data
the application program to a user.
2. The method of claim 1, comprising providing a set of application
programs, whereby each application program of the set of application
programs is adapted to process a set of classes of database tables and
online analytical processing cubes.”
’321 Patent at 7:12-42.
Independent claim 4 is a system claim that is similar to claim 1:
“4. A data processing system comprising:
a relational database of a data warehouse for storing a set of
database tables, each database table being assigned to an entity
type and storing entities of its entity type;
a relational database of a data warehouse for storing a set of online
analytical processing cubes, each online analytical processing
cube specifying a layout for transactional data storage;
at least one application program for processing at least one class of
database tables and at least one class of online analytical
processing cubes;
a mapping table for mapping a sub-set of the set of database tables
to the at least one class of database tables, the sub-set of
database tables comprising database tables of one or more entity
types;
a mapping table for mapping a sub-set of the set of online
analytical processing cubes to the at least one class of online
analytical processing cubes;
means for invoking an online analytical processing component to
fill the online analytical processing cubes with transactional
data;
means for processing the entities stored in the sub-set of database
tables and the transactional data stored in the sub-set of online
analytical processing cubes with the application program; and
means for providing analysis of the entities and the transactional
data processed by the application program to a user.”

25

Id. at 7:46–8:18. Claim 1 is representative because it is “substantially similar” to claim 4. TMSJ

26

at 5; see Content Extraction, 776 F.3d at 1348 (concluding that a claim is representative of other

27

claims when they are “substantially similar and linked to the same abstract idea”). SAP does not

28

oppose that claim 1 is representative.
52

1.

1

The ’321 Patent Is Directed to the Abstract Idea of “Organizing
Information into Logical Groups”

Teradata asserts that the ’321 Patent is directed to the abstract idea of “associating

2
3

(‘mapping’) database tables and OLAP cubes with respective classes for use with application

4

programs.”17 TMSJ at 7. When evaluating computer-related claims, the first step in the Alice

5

inquiry “asks whether the focus of the claims is on the specific asserted improvement in computer

6

capabilities” or “instead, on a process that qualifies as an ‘abstract idea’ for which computers are

7

invoked merely as a tool.” Enfish, 822 F.3d at 1335–36. Teradata argues that the “‘mapping’ to

8

classes at the heart of the ’321 patent is simply a practice of organizing information, a type of

9

activity that courts have held to be abstract and ineligible for patent protection.” TMSJ at 7. For

10

example, even SAP’s expert, Dr. David Maier, explains,
“The ’321 patent relates to ways to organize the tables and cubes used
in databases so that they can be more easily and efficiently recognized
and accessed. At a high level this organization is accomplished by
assigning a table or cube to a particular class. These classes serve to
group data structures storing related data, so an application can access
the structures together.”

United States District Court
Northern District of California

11
12
13
14

Dkt. No. 472-2 (“Maier Reb. Rep.”) ¶ 683.

15
16

SAP contends that “if there is an abstract idea, it is organizing information into logical
groups.” Opp. TMSJ at 3–6. Although its opposition assumes arguendo that the claims are

17
directed to an abstract idea, SAP does not dispute that the claims are directed to an abstract idea.
18
19
20
21

It “does not contest that the claims are directed to this idea of ‘organizing information into logical
groups’ and that it is abstract.” Id. at 4. Instead, it disputes Teradata’s assertion that the claims
are directed to the narrower abstract idea of “associating (‘mapping’) database tables and OLAP
cubes with respective classes for use with application programs.” TMSJ at 7.

22
SAP takes the unusual position as a patentee of asserting a broader definition of the
23
abstract idea in order to contend that the physical-realm claim elements—i.e., database tables,
24
OLAP cubes, application programs, and mapping—and their combination should be analyzed
25
26
27
28

In the Claim Construction Order, I rejected Teradata’s proposal to construe “mapping” as
“associating or assigning.” Claim Construction Order at 14–15. Instead I construed “mapping” as
“[c]reating and storing, in computer system memory or secondary storage for a computer system,
an association between data elements in the computer system such that a computer can locate a
data element using that association.” Id.
53
17

1

under Alice step two to determine that there is an inventive concept. See Opp. TMSJ at 6; Hearing

2

Tr. at 60–61. An inventive concept “reflects something more than the application of an abstract

3

idea using well-understood, routine, and conventional activities previously known to the industry.

4

It must be enough to transform an abstract idea into a patent-eligible invention.” Cellspin Soft,

5

Inc. v. Fitbit, Inc., 927 F.3d 1306, 1316 (Fed. Cir. 2019), cert. denied sub nom. Garmin USA, Inc.

6

v. Cellspin Soft, Inc., 140 S. Ct. 907 (2020) (internal quotation marks and citations omitted).

7

In Berkheimer v. HP Inc., 881 F.3d 1360 (Fed. Cir. 2018), the Federal Circuit held that

8

“[w]hether something is well-understood, routine, and conventional to a skilled artisan at the time

9

of the patent is a factual determination.” Berkheimer, 881 F.3d at 1369. The court held that the
claims at issue were directed to the abstract ideas of parsing, comparing, storing, and editing data.

11

United States District Court
Northern District of California

10

Id. at 1366. The patentee argued that the specification described “an inventive feature that stores

12

parsed data in a purportedly unconventional manner” which “eliminates redundancies, improves

13

system efficiency, [and] reduces storage requirements” among other things. Id. at 1369. The

14

Federal Circuit therefore held that the “improvements in the specification, to the extent they are

15

captured in the claims, create a factual dispute regarding whether the invention describes well-

16

understood, routine, and conventional activities.” Id.

17

In contrast, in BSG Tech LLC v. Buyseasons, Inc., 899 F.3d 1281 (Fed. Cir. 2018), the

18

Federal Circuit held that the dispute about whether the claims recited “unconventional features

19

that provides benefits over conventional prior art databases” was irrelevant because “a claimed

20

invention’s use of the ineligible concept to which it is directed cannot supply the inventive concept

21

that renders the invention ‘significantly more’ than that ineligible concept.” BSG, 899 F.3d at

22

1289–91. The court held that the claims at issue were directed to the abstract idea of “considering

23

historical usage information while inputting data.” Id. at 1286. The only alleged unconventional

24

feature of the claims was “the requirement that users are guided by summary comparison usage

25

information or relative historical usage information.” Id. at 1291. The Federal Circuit held that

26

“this simply restate[d]” what it had already determined was an abstract idea and therefore the

27

question about whether this requirement was non-routine or unconventional was irrelevant. Id.

28

“As a matter of law, narrowing or reformulating an abstract idea does not add ‘significantly more’
54

1

to it.” Id. The Federal Circuit affirmed the district court’s determination that the asserted claims

2

lacked an inventive concept. Id.

3

In this case, Teradata asserts that the abstract idea is “associating (‘mapping’) database

4

tables and OLAP cubes with respective classes for use with application programs” because then

5

SAP’s purported inventive concept is simply a restatement of the abstract idea and arguably fails.

6

In contrast, SAP contends that the abstract idea is “organizing information into logical groups”

7

because then the combined elements of the database, OLAP cubes, application programs, and

8

mapping arguably create an inventive concept and a genuine dispute of fact of whether the

9

combination of these elements is non-routine or unconventional. I will now address which

10

abstract idea the claims are directed towards.

United States District Court
Northern District of California

11

SAP argues that “mapping” database tables and OLAP cubes with respective classes for

12

use with application programs is not an abstract idea because database tables, OLAP cubes, and

13

application programs are computer structures, not mere concepts, and “mapping” requires the

14

creation of computer data structures. Opp. TMSJ at 4. Teradata cites no precedent holding that a

15

database, OLAP cube, or application program is an abstract idea. See id. For “mapping,”

16

however, Teradata points to Autodesk, where the district court held that “it would be difficult to

17

conceive of a more abstract concept than ‘mapping,’ when that concept is not tied to any particular

18

object or method.” East Coast Sheet Metal Fabricating Corp. v. Autodesk, Inc., 2015 WL

19

226084, at *6 (D.N.H. Jan. 15, 2015), amended in part, 2015 WL 925614 (D.N.H. Mar. 3, 2015),

20

and aff'd, 645 F. App'x 992 (Fed. Cir. 2016). Although SAP does not directly address Autodesk, it

21

contends that “mapping” as construed by the Claim Construction Order is not abstract because it

22

requires the creation of computer data structures: “[c]reating and storing, in computer system

23

memory or secondary storage for a computer system, an association between data elements in the

24

computer system such that a computer can locate a data element using that association.” Claim

25

Construction Order at 15. According to SAP, “[t]here is nothing abstract about a data structure

26

that an application program running on a computer uses to locate particular data stored in other

27

data structures in the computer system.” Opp. TMSJ at 4.

28

Teradata responds that despite the claim construction, “mapping” is an abstract idea
55

1

because “there is nothing in the claim language or specification that would materially distinguish a

2

computerized mapping table from one that could be created with a pen and paper.” TMSJ at 11. I

3

agree. In Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307 (Fed. Cir. 2016), the

4

Federal Circuit held that the patent at issue was directed to the abstract idea of “receiving e-mail

5

(and other data file) identifiers, characterizing e-mail based on the identifiers, and communicating

6

the characterization—in other words, filtering files/e-mail.” Symantec Corp., 848 F.3d at 1313. It

7

held that the patent was invalid because “with the exception of generic computer-implemented

8

steps, there is nothing in the claims themselves that foreclose them from being performed by a

9

human, mentally or with pen and paper.” Id. at 1318. In this case, SAP’s expert, Maier, opined
that “[a] folder or directory structure stored on a computer system [that] groups objects and allows

11

United States District Court
Northern District of California

10

them to be located . . . can be considered a mapping table.” Dkt. No. 472-3 (“Appendix 3 to Maier

12

Report”) at 32. Because a computer folder “originated as a metaphor for paper folders,”

13

“mapping” is an abstract idea. TMSJ at 12.

14

Moreover, the claims are not focused on how “mapping” improves computer functionality.

15

SAP’s expert explains that the “ways to organize the tables and cubes used in databases,” e.g.,

16

“assigning a table or cube to a particular class,” makes the database tables and OLAP cubes “more

17

easily and efficiently recognized and accessed.” Maier Reb. Rep. ¶ 683. But the specification

18

expressly states that the improvement is simplifying the “selection of database tables as input

19

parameters and the selection of OLAP cubes” to make it “more user friendly.” ’321 Patent at

20

5:63-65 (emphasis added); see also id. at 2:40-44; 3:61-64; 4:8-19; 5:27-30; 5:35-44; 6:11-16

21

(references to how the value from “mapping” is a result of a human’s choice to associate particular

22

tables or cubes with classes, not from a new data structure or technological improvement).

23

Further, neither the claims nor the specification recites any specific algorithms for mapping tables

24

and cubes to classes, collecting data into OLAP cubes, processing the data, or analyzing the data.

25

TMSJ at 14. Consequently, SAP only “conclusorily claims an improvement, but never identifies

26

what the specific improvement is, despite the Federal Circuit’s requirement that claims assert a

27

‘specific asserted improvement.’” MyMail, Ltd. v. OoVoo, LLC, No. 17-CV-04487-LHK, 2020

28

WL 2219036, at *15 (N.D. Cal. May 7, 2020), aff'd, 2021 WL 3671364 (Fed. Cir. Aug. 19, 2021)
56

1

(quoting McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1314–15 (Fed. Cir.

2

2016)). As a result, “mapping” is an abstraction.

3

Teradata also asserts that the presence of physical components—i.e., the computer data

4

structures composed of the database, OLAP cubes, and application programs—do not save the

5

claims from being directed to an abstract idea. Dkt. No. 559 (“Reply TMSJ”) at 5–6. For

6

example, in In re TLI Commc'ns LLC Pat. Litig., 823 F.3d 607 (Fed. Cir. 2016), the Federal

7

Circuit determined that the claim at issue was directed to an abstract idea even though the claims

8

required “concrete, tangible components such as ‘a telephone unit’ and a ’server,’” because “the

9

specification makes clear that the recited physical components merely provide a generic

10

United States District Court
Northern District of California

11

environment in which to carry out the abstract idea.” TLI, 823 F.3d at 611.
But SAP does not dispute that the claims are directed to an abstract idea; instead it disputes

12

the scope of the abstract idea. SAP persuasively contends that, contrary to Teradata’s narrow

13

characterization of multiple Federal Circuit decisions, the Federal Circuit “resists conflating a

14

claim’s abstract idea with its physical-realm elements.” Opp. TMSJ at 6. For example, Teradata

15

characterized the Federal Circuit’s conclusion in Capital One as stating that the claims were

16

directed to the abstract idea of “[s]ystems for manipulating XML documents by organizing data

17

components into data objects and records and responding to modifications of the data.” TMSJ at

18

8. Instead, the Federal Circuit concluded that “the patent claims are, at their core, directed to the

19

abstract idea of collecting, displaying, and manipulating data.” Capital One, 850 F.3d at 1340.

20

Similarly, Teradata characterized the decision in Electric Power Group to hold that the claims at

21

issue were directed to the abstract idea of “[s]ystems and methods for performing real-time

22

monitoring of an electric power grid by collecting data from multiple data sources, analyzing the

23

data, and displaying the results.” TMSJ at 8. But the Federal Circuit held that the claims were

24

focused on the following abstract idea: “a process of gathering and analyzing information of a

25

specified content, then displaying the results.” Elec. Power Grp., 830 F.3d at 1354.

26

Teradata points to Intellectual Ventures I LLC v. Erie Indemnity Co., 850 F.3d 1315 (Fed.

27

Cir. 2017) as an example of the Federal Circuit including a physical-realm element in its

28

articulation of the abstract idea. Reply TSMJ at 4. There, the Federal Circuit held that “the
57

1

invention is drawn to the abstract idea of ‘creating an index and using that index to search for and

2

retrieve data.’” Erie, 850 F.3d at 1327. An “index” was a known structure in the field of database

3

technology. The Federal Circuit, however, was not discussing the specific index in the field of

4

database technology in its definition of the abstract idea, but indexes generally. See id. (explaining

5

that “[t]his type of activity, i.e., organizing and accessing records through the creation of an

6

index-searchable database, includes longstanding conduct that existed well before the advent of

7

computers and the Internet. For example, a hardcopy-based classification system (such as

8

library-indexing system) employs a similar concept as the one recited by” the patent).
Accordingly, I agree with SAP that the claims are directed to the abstract idea of

10

“organizing information into logical groups”.18 But for the reasons explained below, the claims

11

United States District Court
Northern District of California

9

are patent-ineligible because they fail to encompass an inventive concept.

12

2.

The ’321 Patent Does Not Contain an Inventive Concept

Teradata asserts that the ’321 Patent lacks an inventive concept because it “recites

13
14

well-known, routine, and conventional database elements” and “uses these elements to perform

15

well-understood, routine, and conventional functions of collecting, organizing, processing, or

16

analyzing data.” TMSJ at 13. An inventive concept “cannot simply be an instruction to

17

implement or apply the abstract idea on a computer” and “must be significantly more than the

18

abstract idea itself.” BASCOM, 827 F.3d at 1350. “If a claim’s only ‘inventive concept’ is the

19

application of an abstract idea using conventional and well-understood techniques, the claim has

20

not been transformed into a patent-eligible application of an abstract idea.” BSG, 899 F.3d at

21

1290–91 (Fed. Cir. 2018). But “an inventive concept can be found in the non-conventional and

22

non-generic arrangement of known, conventional pieces.” BASCOM, 827 F.3d at 1350.

23

At the summary judgment stage, Teradata, as the movant, has the burden of showing that

24

there is no genuine dispute as to any material fact and that it is entitled to judgment as a matter of

25

law. FED. R. CIV. P. 56(a). “[W]hether a claim limitation or combination of limitations is

26
27
28

Teradata asserts that whether I adopt SAP’s or its articulation of the abstract idea, the Section
101 analysis does not change because its articulation “is simply a form of organizing information
into logical groups.” Reply TMSJ at 7.
18

58

1

well-understood, routine, and conventional is a factual question.” BSG, 899 F.3d at 1290.

2

Because such a fact is “pertinent to the invalidity conclusion” it “must be proven by clear and

3

convincing evidence.” Berkheimer, 881 F.3d at 1368. “When there is no genuine issue of

4

material fact regarding whether the claim element or claimed combination is well-understood,

5

routine, conventional to a skilled artisan in the relevant field, this issue can be decided on

6

summary judgment as a matter of law.” Id. And if the only alleged unconventional feature is the

7

abstract idea itself, summary judgment is appropriate. BSG, 899 F.3d at 1291.

8

a.

Inventive Concept Identified During the Hearing

Notably, SAP did not assert what the inventive concept is in its opposition. When asked

9

during the hearing, its counsel explained that the inventive concept is composed of three elements

11

United States District Court
Northern District of California

10

in claim 4:19 (1) a relational database that stores both database tables and OLAP cubes (’321

12

Patent at 7:47-53); (2) an application program that accesses and processes those database tables

13

and OLAP cubes, not individually, but as a class (’321 Patent at 7:54-56); and (3) the two

14

mapping tables, which are data structures inside the computer, which associate the database tables

15

and the OLAP cubes with a particular class (’321 Patent at 8:1-7). Hearing Tr. at 60–61. Its

16

counsel contended that combining these elements constituted an improved database technique that

17

simplified “the selection of database tables as input parameters and the selection of OLAP cubes”

18

and made it “more user friendly.” Id. at 62 (citing ’321 Patent at 5:63-67). According to the ’321

19

Patent, the improved database technique also “may enable non-expert users to perform complex

20

transactional data processing and to integrate expert knowledge in the class definitions.” Id.

21

Teradata’s counsel responded that these elements do not create an inventive concept. For

22

the first element, under the agreed claim construction, “a relational database of a data warehouse

23

for storing a set of [OLAP] cubes” is “a database that stores information in tables of rows and

24

columns of data located in a data warehouse that can store at least one [OLAP] cube.” Dkt. No.

25

206 (“Joint Claim Construction Statement”) at 2. In other words, the OLAP cube is not stored in

26

the relational database as SAP’s counsel explained, but in the data warehouse. Hearing Tr. at 64;

27
28

19

SAP only addresses claim 4 but Teradata addresses the parallel elements of claims 1 and 4
together. Reply TMSJ at 9.
59

1

see also ’321 Patent at 4:58-60, Fig. 3 (showing that the data warehouse contains a set of OLAP

2

cubes and not a relational database). And the specification admits that the storage of OLAP cubes

3

in data warehouse systems is well-understood, routine, and conventional. Id. at 1:26-27 (“An

4

OLAP cube is a multi-dimensional representation of a set of data. Such a cube is the basis for

5

transaction data storage in prior art data warehouse systems.”).
For the second element, Teradata’s counsel pointed out that SAP admitted that application

7

programs were well-known in the prior art. See Dkt. No. 211 (“SAP Opening Claim Construction

8

Brief”) at 20 (“Application programs were well-known to the POSITA at the time the ’321 patent

9

was filed.”). And for the third element, Teradata’s counsel asserted that “mapping” is “merely an

10

abstraction” for the reasons explained above, e.g., there is nothing in the specification that would

11

United States District Court
Northern District of California

6

distinguish the mapping tables from what a person could do on pen and paper. Hearing Tr. at 66;

12

see supra Part II.B.1.

13

SAP’s counsel conceded that application programs, database tables, and OLAP cubes were

14

well-known. Hearing Tr. at 68. But he argued that nothing in the specification or the record

15

suggested that any of the three elements it identified were well-known or conventional in 2003, the

16

patent’s effective filing date. Id. This does not address, however, Teradata’s argument that

17

“mapping” is an abstraction. Because “mapping” simply restates what I have determined is an

18

abstract idea, i.e., organizing information into logical groups, the question of whether the claim

19

element is well-understood, routine, and conventional, is irrelevant. BSG, 899 F.3d at 1291.

20

SAP’s response also does not address Teradata’s argument that application programs,

21

database tables, and OLAP cubes are generic software components that cannot supply an inventive

22

concept. Reply TMSJ at 11; see Content Extraction, 776 F. at 1348 (“[T]he mere recitation of a

23

generic computer cannot transform a patent-ineligible abstract idea into a patent-eligible

24

invention.”). The specification does not identify any specific advancement over prior art. Instead,

25

it explains that these physical elements were well-understood, routine, and conventional features

26

of databases. As SAP’s counsel explained, “mapping” is how the application program accesses

27

the data in the database tables and OLAP cubes as a class and therefore because “mapping” is an

28

abstraction, none of these elements can provide an inventive concept. See Hearing Tr. at 60–61.
60

1

Even though the question of conventionality is irrelevant, Teradata also points to examples

2

in the prior art to argue that “mapping” is a well-understood, routine, and conventional element.

3

Id. at 14–15. Some prior art patent applications contained the same process of classifying or

4

assigning tables or cubes to logical groupings or classes, each associated with applications for

5

processing. See, e.g., Dkt. No. 472-4 (“’061 Colossi Reference”) (U.S. Patent Application

6

Publication No. US 2004/0139061) (Figure 3 showing a grouping of tables related to the

7

measurement of sales by time, product, and region); Dkt. No. 472-5 (“Bakalash Reference”) (U.S.

8

Patent No. 6,385,604”) (Figure 4A showing a grouping of tables by supplier, time period, part, and

9

supplied parts); Dkt. No. 472-6 (“Colossi Article” or “Colossi Reference”) (Figure 1 showing a

10

United States District Court
Northern District of California

11

class of OLAP cubes related to finance, market share, employees, and customers).
SAP contends that Teradata has not shown an absence of a genuine dispute of material fact

12

because “[w]hether a particular technology is well-understood, routine, and conventional goes

13

beyond what was simply known in the prior art. The mere fact that something is disclosed in a

14

piece of prior art, for example, does not mean it was well-understood, routine, and conventional.”

15

Berkheimer, 881 F.3d at 1369. It argues that the prior art references do not support Teradata’s

16

argument because Teradata “cites no evidence that these references were widely read and

17

understood by 2003, or adopted by others so widely that they became routine and conventional.”

18

Opp. TMSJ at 10. For example, the Colossi ’061 reference was not published until 2004, after the

19

’321 Patent’s effective filing date and therefore “it cannot possibly show that others adopted its

20

teachings to such an extent that they became conventional in 2003.” Id. As for the other Colossi

21

reference, it contends that Figure 1 of the 2002 Colossi article was simply a proposal and Teradata

22

cites to no evidence showing that anyone adopted the proposal so that it became routine and

23

conventional by 2003. Id. Similarly, it argues that there is no evidence that the Bakalash

24

reference was widely read or understood by 2003. Id.

25

Further, SAP contends that Teradata has not met its burden of showing that the patent

26

claims’ combination of physical-realm elements was conventional, routine, and well-understood.

27

Id. at 11. According to SAP, the prior art references show different claimed inventions than the

28

one at issue in the ’321 Patent. Id. For example, both the Colossi ’061 reference and the Bakalash
61

1

reference describe the claimed invention as a “star schema,” a way to represent the logical

2

structure of a relational base, which is not the claimed invention at issue here. Id. at 17; see

3

Bakalash Reference at 3:54-57 (“An exemplary star schema is illustrated in FIG. 4A”); ’061

4

Colossi Reference ¶ 0075 (“FIG. 3 illustrates a sample star-join schema”). Teradata does not

5

contend that either reference shows any of the other ’321 claim elements, e.g., any system in

6

which both tables and cubes are mapped to classes, as required by the ’321 Patent. Id. at 11–12.

7

Similarly, SAP’s expert opines that Figure 1 in the Colossi Reference describes cubes that are

8

different from the construed definition of OLAP cubes in this case. Dkt. No. 520-2 (“Maier Reb.

9

Rep.”) ¶¶ 746–48.

10

These arguments are irrelevant, however, because under a Section 101 analysis, as opposed

United States District Court
Northern District of California

11

to a Section 102 or 103 analysis, Teradata does not have to compare each ’321 claim to the prior

12

art. Reply TMSJ at 10. As the Federal Circuit has explained,
“The appropriate question is not whether the entire claim as a whole
was ‘well-understood, routine [and] conventional’ to a skilled artisan
(i.e., whether it lacks novelty), but rather, there are two distinct
questions: (1) whether each of the [elements] in the claimed [product]
(apart from the natural laws themselves) involve well-understood,
routine, conventional activity previously engaged in by researchers in
the field, and (2) whether all of the steps as an ordered combination
add[ ] nothing to the laws of nature that is not already present when
the steps are considered separately.”

13
14
15
16
17
18

Chamberlain Grp., Inc. v. Techtronic Indus. Co., 935 F.3d 1341, 1348–49 (Fed. Cir. 2019), cert.

19

denied, 141 S. Ct. 241 (2020) (internal citations and quotation marks omitted). Teradata explains

20

that it was relying on the prior art references to show that “mapping” and “mapping tables,” under

21

my construction and as interpreted by SAP’s expert, were well-known in the art. Reply TMSJ at

22

10–11. SAP does not address Teradata’s arguments that the Colossi ’061 and Bakalash references

23

show mappings of tables to classes. Further, SAP’s expert undermines its argument that there is

24

no clear and convincing evidence that the Colossi and Bakalash references were widely circulated

25

or understood by 2003; Maier opines that a POSITA would have understood how to implement

26

certain claim elements at the time based on these three references. See infra Part II.B.2.b; see

27

Maier Reb. Rep. ¶¶ 704, 715. SAP’s argument that these three elements create an inventive

28

concept fails.
62

b.

1

Remaining Claim Elements

SAP’s counsel clarified that SAP was not abandoning the seven physical-realm elements

2
outlined in its opposition by focusing on the elements above. Hearing Tr. at 63. According to
3

SAP, “even if each of the [] seven claim elements individually were known, Teradata submits no

4
clear and convincing evidence that this particular combination of structural, physical-realm
5

elements was conventional by 2003”: the elements above and (1) “an OLAP component filling

6
the OLAP cubes with transactional data and [a particular] means for invoking that component to
7

perform that function; (2) “a [particular] means for processing with the application program the

8
entities stored in the sub-set of database tables and the transactional data stored in the sub-set of
9
10

OLAP cubes; and (3) “a [particular] means for providing analysis of those entities and
transactional data processed by the application program.” Opp. TMSJ at 9.

11

United States District Court
Northern District of California

But these remaining claim limitations—“filling cubes with data, processing data, and
12
providing analysis”—are also abstract and cannot provide an inventive concept. TMSJ at 14.
13

Contrary to SAP’s addition of the word “[particular]” in the elements,20 these limitations are

14
purely functional because the claims do not recite any specific algorithms for performing these
15
steps; instead they simply claim a result and reflect abstract ideas. TMSJ at 14; see Affinity Labs
16

of Texas, LLC v. Amazon.com Inc., 838 F.3d 1266, 1269 (Fed. Cir. 2016) (“The purely functional

17
nature of the claim confirms that it is directed to an abstract idea, not to a concrete embodiment of
18

that idea.”). SAP does not respond to this argument. The question of whether these elements are

19
well-understood, routine, and conventional is therefore irrelevant because these elements are
20
abstract ideas and therefore cannot supply an inventive concept. See BSG, 899 F.3d at 1290–91.
21
In any event, Teradata argues that all of these steps are also well-understood, routine, and
22

conventional, as SAP admits. Id. SAP’s expert admits that filling cubes with data, processing

23
data, and providing analysis are all well-known elements and would be familiar to a skilled
24

artisan. See, e.g., Maier Reb. Rep. ¶ 704 (arguing that the “Colossi and Colossi ’061

25
26
27
28

Teradata asserts that SAP’s rewriting of claim 4 underscores that the ordered “combination” of
elements in claim 4 cannot survive Alice step two. Reply TMSJ at 9. It argues that SAP (1)
combines the first two elements; (2) moves the “mapping” limitations to follow the filling,
processing, and providing analysis limitations; and (3) adds the word “[particular]” to make three
claim elements seem less generic. Id.
63
20

1

references . . . further demonstrate that a POSITA would have been well aware of applications that

2

utilized RDBMS components to fill OLAP cubes.”); id. ¶ 715 (arguing that the “Colossi

3

reference . . . further demonstrate[s] that a POSITA would have been well aware of applications

4

that processed the entities stored in the sub-set of database tables and the transactional data stored

5

in the sub-set of online and analytical processing cubes with the application program.”); id. ¶ 719

6

(arguing that the structures for “providing analysis of the entities and the transaction data

7

processed by the application program to a user” were “familiar to a skilled artisan”). Maier

8

pointed to business intelligence tools such as “Microsoft Excel, BusinessObjects and Tableau” as

9

examples of applications that were well-known that filled cubes with data, processed data, and

10

provided analysis of data. Id. ¶¶ 701, 712, 719.
Furthermore, the specification is silent as to any purported improvement provided by the

United States District Court
Northern District of California

11
12

claimed combination. See MyMail, 2020 WL 2219036, at *19 (invalidating claims under § 101 in

13

part because the specification was “entirely silent as to . . . how any inventive feature, alone or in

14

an ordered combination, is used in an unconventional manner.”) (internal quotation marks

15

omitted). Because the claim elements simply apply the abstract idea of organizing information

16

into logical groups using well-understood, routine, and conventional activities previously known

17

to the industry, the claims do not make the abstract idea patent eligible. Cellspin Soft, 927 F.3d at

18

1316.21 SAP’s claims are directed to the abstract idea of organizing data into logical groups.

19

There is no inventive concept that provides something more than the abstract idea itself.

20

Teradata’s motion for summary judgment is GRANTED.

21

III.

MOTION TO EXCLUDE LEONARD AND WOLFSON TESTIMONY
Teradata’s final motion to exclude expert testimony seeks to exclude portions of Dr.

22
23

Wolfson and Dr. Leonard’s reports. Dkt. No. 480 (“L&W Mot.”) at 1. Teradata asserts that I

24

should exclude Wolfson’s apportionment opinions because they do not satisfy Rule 702 and

25
26
27
28

21

SAP also contends that Teradata has not shown a lack of genuine dispute that there are no other
ways to implement the alleged abstract idea. Opp. TMSJ at 15–16. But Teradata does not have to
show that SAP has preempted an entire idea for the ’321 claims to be patent-ineligible. “While
preemption may signal patent ineligible subject matter, the absence of complete preemption does
not demonstrate patent eligibility.” FairWarning IP, LLC v. Iatric Systems, Inc., 839 F.3d 1089,
1098 (Fed. Cir. 2016) (citation omitted).
64

1

Leonard’s Profit Apportionment Method with respect to the ’321 Patent, which relies on

2

Wolfson’s apportionment factors. Id. Because I conclude that the ’321 Patent is invalid, any

3

arguments related to it are DENIED as moot. Teradata also moves to exclude certain alternative

4

calculations for reasonable royalties, such as damages before May 21, 2019 for the ’421 and ’179

5

Patents and damages related to an exhibit labeled as “Scenario 2.” Id. SAP does not dispute these

6

issues and therefore Teradata’s motion in relation to these is GRANTED. See Dkt. No. 525

7

(“L&W Opp.”) at 12, 15. As for the antitrust damages, Teradata moves to exclude Leonard’s

8

opinion that certain Teradata business decisions caused the damages. L&W Mot. at 2. Because I

9

conclude that Teradata’s tying claim fails, any argument about antitrust damages is DENIED as
moot. The remaining issues, then, are whether Wolfson’s apportionment analysis of the ’421 and

11

United States District Court
Northern District of California

10

’179 Patents is unreliable and therefore whether Leonard’s Profit Apportionment Method for the

12

’421 and ’179 Patents is unreliable.

13

Teradata asserts that Wolfson’s apportionment analysis is flawed. Id. at 5. SAP explains

14

that Wolfson “was tasked with estimating the value of Teradata’s infringing technology by

15

drawing on his more than 35 years of experience as a computer science professor and the president

16

of a startup company in the data science field.” L&W Opp. at 1 (citing Dkt. No. 488-6 (“Wolfson

17

Rep.”) ¶¶ 4–18). Teradata points out that Wolfson admits that he has never conducted an

18

apportionment analysis before, and he could not provide examples of others apportioning revenues

19

in the manner he did or any third-party resources that could guide his efforts. Dkt. No. 488

20

(“Wolfson Depo.”) at 42–43, 46, 77–78.

21

Wolfson’s analysis follows the same methodology affirmed by the Federal Circuit in

22

Summit 6, LLC v. Samsung Elecs. Co. Ltd., 802 F.3d 1283, 1296–98 (Fed. Cir. 2015). Id. at 2–3.

23

There, the expert determined the smallest salable patent-practicing unit for which revenue data is

24

available and further apportioned the value of the claimed invention to take into account only

25

those features that infringed. Summit 6, 802 F. 3d at 1297. In this case, Wolfson identified the

26

smallest salable unit for each patent, e.g., the Teradata Columnar feature for the ’421 Patent and

27

the Teradata Database for the ’179 Patent, and then undertook a “multi-step, quantitative

28

apportionment specific to each patent.” Id. at 1297. Teradata asserts that Summit 6 is
65

1

distinguishable because there, an economist, not a technical expert, made opinions based on his

2

careful quantitative review of objective financial and customer usage data. Dkt No. 561 (“L&W

3

Reply”) at 4. Here, Wolfson repeatedly confirmed that he had no data on customers’ usage of the

4

specific features. See Dkt. No. 560-6 (“Wolfson Depo.”) at 98–110.

5

But Wolfson explains that he does not use such data because Teradata claims that it does

6

not possess or maintain information about how its customers deploy or configure features such as

7

the Teradata Columnar, for example. Wolfson Rep. ¶ 49. Instead, Wolfson “had information

8

about how certain Teradata employees who are in direct contact with customers value various

9

features that are related to ’421 . . . .” Wolfson Depo. at 110. Like the expert in Summit 6, for the
’421 Patent he determined the proportion of customers who would be expected to configure

11

United States District Court
Northern District of California

10

Teradata Columnar in an infringing manner based on Teradata’s documentation and publications

12

and then subtracted non-infringing configuration options. Wolfson Rep. ¶¶ 48–69. For the ’179

13

Patent, he approximated the value of Teradata’s “complex query” processing components of the

14

Teradata Database based on Teradata’s internal spreadsheets. Id. ¶¶ 97–108. Wolfson then

15

excluded use cases that do not involve the infringing subquery processing. Id. ¶¶119–22. His

16

methodology is proper.

17

Teradata asserts that when asked how he came to determine or know what the alleged

18

infringing conduct was, Wolfson responded that he did not “exactly recall how” he identified the

19

exact piece that infringes and that “some of it is a hunch.” Wolfson Depo. at 77–78. But the full

20

context of his statement was that apportionment is not an “exact science,” which has been

21

acknowledged by courts. L&W Opp. at 4; see Apple Inc. v. Motorola, Inc., 757 F.3d 1286, 1315,

22

1319 (Fed. Cir. 2014), overruled on other grounds by Williamson v. Citrix Online, LLC, 792 F.3d

23

1339 (Fed. Cir. 2015) (recognizing that estimating a “reasonable royalty” for example “is not an

24

exact science” and holding that an expert’s method of apportionment was admissible even if other

25

reliable methods of estimating a reasonable royalty existed). Wolfson testified that he did not

26

analyze the infringement, or the claim construction order himself and he never spoke to SAP’s

27

expert on infringement, Dr. Maier. Id. at 34, 84. But his analysis of the relative value of the

28

infringing technology relies on Maier’s opinions on patent infringement, which is common and
66

1

appropriate. Fujifilm Corp. v. Motorola Mobility LLC, No. 12-cv-03587-WHO, 2015 WL

2

1737951, at *4 (N.D. Cal. Apr. 8, 2015) (It is “reasonable to expect that experts will rely on the

3

opinion of experts in other fields as background material for arriving at an opinion.”). When

4

asked when he received Maier’s report, he stated that the earliest copy he had was from the same

5

day that he signed his own report, but he testified that it was his impression that he had received

6

something similar beforehand. Id. at 162–63. In addition, Wolfson testified that he relied upon

7

SAP’s infringement contentions, which were the basis of Maier’s report, before finalizing his

8

report. See Wolfson Depo. at 67–68, 157–61.
Teradata also emphasizes that “Wolfson’s views on the patents are confused and plainly

10

omit key elements of the claimed invention.” L&W Mot. at 7. According to Teradata, he ignored

11

United States District Court
Northern District of California

9

the existence of a limitation in the ’179 Patent that subqueries be optimized “without

12

transformation” and only focused on whether there were subqueries to the queries being processed.

13

See Dkt. No. 491-15 (“’179 Patent”) at 39:3–5; see Wolfson Depo. at 126–27 (stating that he

14

believed the question of “transformation” was “completely tangential” to his report and confirming

15

that the word “transformation” was “not even in” his report). Because Wolfson “values only the

16

prevalence of subqueries rather than subqueries that are optimized ‘without transformation,’”

17

Teradata asserts that his approach is inappropriate. L&W Mot. at 7. SAP responds that it is

18

appropriate that Wolfson only considered the prevalence of subqueries because according to Maier,

19

the “without transformation” limitation is met when a query contains a subquery. L&W Opp. at 5

20

(citing Dkt. No. 524-11 (“Appendix 4 of the Maier Expert Rep.”) at 82–84). It contends that

21

Teradata has cited no evidence otherwise and therefore Wolfson did not need to account for the

22

“without transformation” limitation. Id. These arguments go to the weight and not the admissibility

23

of Wolfson’s opinions.

24

Finally, Teradata asserts that Wolfson’s calculations “reflect a host of allegedly quantitative

25

assumptions that do not connect to the qualitative documents on which he relies.” L&W Mot. at 8.

26

For example, Teradata argues that he assumes that 50–100% of Teradata’s customers use a certain

27

configuration of Columnar solely because Teradata “recommends” the configuration, but he has no

28

basis for this assumption. Id. SAP points out that this is a rational analysis given that some
67

1

customers may not follow Teradata’s recommendation and therefore would be expected to choose

2

both options equally (the 50% endpoint of the range). Id. at 5. And because customers are likely to

3

follow a manufacturer’s recommendation, this would result in the upper endpoint of the range

4

(100%). Id. Teradata also argues that with the ’179 Patent, Wolfson relied on a 50% estimate for

5

the prevalence of queries containing subqueries “solely on a third-party paper,” but SAP contends

6

that the paper “analyzed a well-established industry performance benchmark, was peer-reviewed, []

7

was presented at a prestigious computer science conference,” and Wolfson analyzed the benchmark

8

independently. L&W Mot. at 8–9; L&W Opp. at 5–6; see Wolfson Rep. ¶ 121. These arguments

9

go to the weight and not the admissibility of Wolfson’s opinions.
Wolfson’s apportionment analysis was based on reliable principles and guided by Federal

11

United States District Court
Northern District of California

10

Circuit case law. Leonard’s Profit Apportionment Approach for the ’421 and ’179 Patents should

12

not be excluded either.
CONCLUSION

13
14

For the reasons above, SAP’s motion for summary judgment on Teradata’s trade secret

15

claims is GRANTED. Its motion related to Teradata’s business trade secret claims under the

16

DTSA is DENIED as moot. Its motion related to Teradata’s tying claim is GRANTED.

17

Teradata’s motion for summary judgment on the invalidity of the ’321 Patent is GRANTED. Its

18

motion for partial summary judgment against an award of damages for infringement of the ’179

19

and ’421 Patents before May 21, 2019, is GRANTED. Its motion to exclude portions of Kraska’s

20

expert report is DENIED as moot. Its motion to exclude portions of Horn’s report is GRANTED

21

in part and DENIED in part. Its motion to exclude portions of the Leonard and Wolfson reports is

22

DENIED in part as moot and DENIED in part on the merits. Its motion to exclude portions of

23

Mehrotra’s report is DENIED. SAP’s motion to exclude portions of Asker’s report is GRANTED

24

in part and DENIED in part.

25
26

IT IS SO ORDERED.
Dated: November 8, 2021

27
William H. Orrick
United States District Judge

28
68