VLSI Technology LLC v. Intel Corporation
Filing
703
Redacted Copy of 687 Sealed Order by Intel Corporation. (sm3)
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 1 of 24
UNITED STATES DISTRICT COURT
FOR THE WESTERN DISTRICT OF TEXAS
WACO DIVISION
VLSI TECHNOLOGY LLC,
Plaintiff,
§
§
§
§
§
§
§
v.
INTEL CORPORATION,
Defendant.
CIVIL NO. 6:21-CV-057-ADA
ORDER DENYING DEFENDANT INTEL’S RULE 50(B) MOTION FOR
JUDGMENT AS A MATTER OF LAW
Before the Court is Defendant Intel Corporation’s (“Intel”) Rule 50(b) Motion for
Judgement as a Matter of Law (“JMOL”). ECF No. 591. Plaintiff VLSI Technology LLC (“VLSI”)
filed its Response on May 7, 2021. ECF No. 603. Intel then filed its Reply on May 21, 2021. ECF
No. 615. Intel asks this Court for a judgment of no infringement for U.S. Patent Nos. 7,523,373
and 7,725,759, a judgment of invalidity for the asserted claims of the ’759 patent, and a judgment
of no damages. Id. After considering the parties’ briefs and relevant law, the Court DENIES
Defendant’s Motion on all counts for the reasons below.
I. BACKGROUND
VLSI filed its suit for patent infringement on April 11, 2019. ECF No. 1 at 1. In its
Complaint, VLSI accused Intel Corporation (“Intel”) of infringing U.S. Patent Nos. 7,523,373
(“the ’373 Patent”), and 7,725,759 (“the ’759 Patent”). On March 2, 2021, following a six-day
trial, the jury found that Intel literally infringed claims 1, 5, 6, 9, and 11 of the ’373 Patent, and
infringed claims 14, 17, 18, and 24 of the ’759 patent under the doctrine of equivalents. ECF No.
564 at 2–3. The jury also found that Intel had not met its burden to prove that claims 14, 17, 18,
and 24 of the ’759 patent were invalid. Id. at 5. The jury awarded VLSI a total of $2.175 billion in
1
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 2 of 24
damages: a $1.5 billion lump sum for infringement of the ’373 patent and a lump sum of $675
million for infringement of the ’759 patent. Id. at 6–7. Intel subsequently filed a Rule 50(b) Motion
for JMOL on April 9, 2021. ECF No. 591.
II. LEGAL STANDARD
A court may grant JMOL against a prevailing party only if a reasonable jury would not
have a legally sufficient evidentiary basis to find for the non-moving party on that issue. Fed. R.
Civ. P. 50(a)(1). In deciding a renewed JMOL motion, a “court must draw all reasonable inferences
in favor of the nonmoving party, and it may not make credibility determinations or weigh the
evidence.” Taylor-Travis v. Jackson State University, 984 F.3d 1107, 1112 (5th Cir. 2021). The
court must disregard all evidence favorable to the moving party that the jury is not required to
believe. Id. This is because “[c]redibility determinations, the weighing of the evidence, and the
drawing of legitimate inferences from the facts are jury functions, not those of a judge.” Wellogix,
Inc. v. Accenture, L.L.P., 716 F.3d 867, 874 (5th Cir. 2013).
Courts grant JMOL for the party bearing the burden of proof only in extreme cases, when
the party bearing the burden of proof has established its case by evidence that the jury would not
be at liberty to disbelieve, and the only reasonable conclusion is in its favor. Mentor H/S, Inc. v.
Medical Device Alliance, Inc., 244 F.3d 1365, 1375 (Fed. Cir. 2001). JMOL is inappropriate if the
record evidence is such that reasonable and fair-minded men in the exercise of impartial judgment
might reach different conclusions. Laxton v. Gap Inc., 333 F.3d 572, 579 (5th Cir. 2003).
A jury verdict must stand unless there is a lack of substantial evidence, in the light most
favorable to the successful party, to support the verdict. Am. Home Assur. Co. v. United Space
Alliance, LLC, 378 F.3d 482, 487 (5th Cir. 2004). Substantial evidence is more than a scintilla, but
less than a preponderance. Nichols v. Reliance Standard Life Ins. Co., 924 F.3d 802, 808 (5th Cir.
2
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 3 of 24
2019). Thus, JMOL must be denied if a jury's verdict is supported by legally sufficient evidence
that amounts to more than a mere scintilla. Laxton, 333 F.3d at 585.
III. DISCUSSION
I.
Substantial Evidence Supports the Jury’s Infringement Findings for the ’373 and
’759 Patents.
The jury found that the C6 SRAM power multiplexer in Intel’s accused products literally
infringed the ’373 patent. It also found that the Speed Shift feature in Intel’s accused products
infringed the ’759 patent under the doctrine of equivalents. Intel seeks JMOL on those findings.
A.
Substantial Evidence Supports the Jury’s Infringement Verdict For the ’373
Patent.
1. Substantial Evidence Supports the Jury’s Infringement Finding for The
“Minimum Operating Voltage” Limitations.
The asserted method claims of the ’373 Patent require “storing the value of the minimum
operating voltage” of a memory. ’373 Patent 13:13–14. The asserted apparatus claims require “a
memory location that stores a value representative of the minimum operating voltage.” Id. at
13:63–64. The jury found that the C6 SRAM power multiplexer in Intel’s Haswell and Broadwell
products infringe on claims 1, 5, 6, 9, and 11 of the ’373 Patent. ECF No. 591 at 2. At trial, VLSI
identified the accused “memory” as the C6 SRAM in Intel’s products and the accused “value of
the minimum operating voltage” as being stored in Intel’s RING_RETENTION_VOLTAGE fuse.
Id. Intel argues that the accused products do not store the “minimum operating voltage” and
therefore do not infringe on the ’373 Patent. Id.
The Court first looks to the evidence that VLSI presented at trial to determine whether
there is substantial evidence to support the jury’s verdict. VLSI argues that it presented substantial
evidence that the accused “memory” is the C6 SRAM, and the “minimum operating voltage of the
memory” is the RING_RETENTION_VOLTAGE. ECF No. 603 at 1–2. VLSI points to Intel
3
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 4 of 24
internal documents that indicate that the “
” Id. at 2 (citing PTX-3662.702; PTX-3851.1280). Dr. Thomas M. Conte,
VLSI’s infringement expert, testified that Intel’s use of the term “retention” is synonymous with
“memory” and that
applies to C6 SRAM memory. 2/23 Sealed Trial
Tr. 5:1-6. He then explained that the RING_RETENTION_VOLTAGE stores the minimum
retention voltage for the C6 SRAM. Id. at 75:23-76. Dr. Conte further explained that the
“defines the lowest safe voltage” and thus “the minimum operating
voltage.” Id. at 4:19-25, 9:10-22. VLSI corroborated his testimony with source code in the Haswell
and Broadwell products. 3/1 Trial Tr. 1451:4-1452:5.
The Court looks next to Intel’s arguments that the above evidence fails to show that its
products
satisfy
the
“minimum
operating
voltage” limitation.
According
to
Intel,
RING_RETENTION_VOLTAGE is merely a voltage at which the C6 SRAM can retain data, not
necessarily the minimum voltage at which data retention occurs as the asserted claims require.
ECF No. 591 at 2. Intel contends that Dr. Conte confirmed as much when he admitted that the
RING_VF_VOLTAGE_0 fuse value—which is below the RING_RETENTION_VOLTAGE
voltage value—reflects a voltage that is “actually used” in the accused products. Id. at 5. Because
the
C6
SRAM,
as
a
ring
component,
operates
at
voltages
below
the
RING_RETENTION_VOLTAGE, Intel alleges that no reasonable jury could find that the
RING_RETENTION_VOLTAGE fuse stores the C6 SRAM’s minimum operating voltage. Id.
Further, Intel contends that there is no relationship between RING_RETENTION_VOLTAGE and
the C6 SRAM specifically as opposed to the entire ring domain because the voltage applies
generally to the ring domain, which contains multiple components beyond the C6 SRAM. Id.
4
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 5 of 24
The first problem with Intel’s argument is that it misrepresents Dr. Conte’s testimony about
the RING_VF_VOLTAGE_0 fuse value. Intel argues that the RING_RETENTION_VOLTAGE
cannot store the C6 SRAM’s minimum operating voltage because, as Dr. Conte admitted, the
accused products “actually use” the RING_VF_VOLTAGE_0 value, which is a lower voltage.
ECF No. 591 at 5. But even if the RING_VF_VOLTAGE_0 fuse value is below the
RING_RETENTION_VOLTAGE voltage value, there is still a factual dispute for the jury to
resolve as to whether the ring operates at the RING_VF_VOLTAGE_0 fuse value. Dr. Conte
testified that the ring operates at a higher operating voltage level that is derived from the fuse value
after inverse temperature dependence compensations calculations are applied. 3/1 Trial Tr. 1425:11432:5, 1434:9-1437:7, 1450:8-1452:12. Dr. Conte explained that Intel’s witnesses did not
account for the inverse temperature dependence compensations when comparing the
RING_VF_VOLTAGE_0 and the RING_RETENTION_VOLTAGE. Id. His testimony thus
shows that the operating voltage from the RING_VF_VOLTAGE_0 fuse value is higher than the
RING_RETENTION_VOLTAGE. This testimony was also corroborated with multiple Intel
internal documents. One such example is an Intel specification that shows
Id. at 1425:1-1426:20. He came to this conclusion after
reviewing Intel’s source code for the accused products. Id. at 1451:4-1452:5. The jury was free to
credit that testimony over Intel’s witnesses, particularly after Intel’s expert was impeached with
Intel documents. ECF No. 603 at 6. As VLSI indicated, Intel’s internal documents described
RING_RETENTION_VOLTAGE as the
SRAM, where the
for “memory”, or the C6
is the lowest safe voltage and thus the minimum
operating voltage. ECF No. 603 at 2.
5
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 6 of 24
The second problem with Intel’s argument is that it mischaracterizes the relationship
between the RING_RETENTION_VOLTAGE and the C6 SRAM. As VLSI argues, the ’373
Patent
claims
do
not
recite
the
word
“specifically”
to
require
that
the
RING_RETENTION_VOLTAGE apply to the C6 SRAM specifically. ECF No. 603 at 2. Nor
does the ’373 Patent bar the “minimum operating voltage of the memory” from applying to parts
in addition to the memory. ECF No. 603 at 2. The patent claims draw no distinction between a
specific or general relationship connecting the minimum operating voltage to the C6 SRAM.
Instead, the ’373 Patent claims include the transition term “comprising”, and it is black-letter law
that “[t]he transitional term ‘comprising’ . . . is inclusive or open-ended and does not exclude
additional, unrecited elements.” CollegeNet, Inc. v. ApplyYourself, Inc., 418 F.3d 1225, 1235 (Fed.
Cir. 2005). Because RING_RETENTION_VOLTAGE applies to both the “LLC” and C6 SRAM
memories, VLSI concludes that RING_RETENTION_VOLTAGE is the minimum operating
voltage of the C6 SRAM. Id. VLSI presented several sources supporting its claim, including
multiple Intel documents and source code reviews. Id. at 6. The jury ultimately believed VLSI's
expert over Intel’s witnesses.
This Court agrees that VLSI has presented substantial evidence supports the jury’s
infringement finding for the “minimum operating voltage” limitations. VLSI’s presentation of Dr.
Conte’s expert testimony, Intel internal documents, and Dr. Conte’s source code analysis provided
substantial evidence to support the jury’s infringement finding. Although Intel makes several
distinct but related arguments that it cannot infringe on this limitation, the Court finds that VLSI
presented substantial evidence to the contrary. Intel’s arguments fall short of the high bar required
for JMOL, which demands a showing that the jury would not be at liberty to disbelieve Intel’s noninfringement theory considering the evidence. Instead, Intel’s internal documentation and claim
6
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 7 of 24
construction argument reinforces the jury’s infringement verdict as a reasonable one. Accordingly,
this Court denies Intel’s motion for JMOL on this ground.
2. Substantial Evidence Supports the Jury’s Infringement Finding for the “When”
Limitations.
The asserted method claims recite limitations “providing” “as the operating voltage of the
memory” (1) “the first regulated voltage . . . when the first regulated voltage is at least the value
of the minimum operating voltage,” and (2) “the second regulated voltage . . . when the first
regulated voltage is less than the value of the minimum operating voltage.” ’373 Patent 13:20–28,
41–44, 45–52. The asserted apparatus claims also require “a power supply selector that supplies”
as the operating voltage of the memory (1) “the first regulated voltage ... when the first regulated
voltage is at least the minimum operating voltage” and (2) “the second regulated voltage ... when
the first regulated voltage is below the minimum operating voltage.” Id. at 13:59–14:15, 14:20–
23. At trial, VLSI argued that the “first regulated voltage” was Intel’s VCCR, the “second regulated
voltage” is the VCCIO, and the minimum operating voltage is RING_RETENTION_VOLTAGE.
Intel alleges that VLSI introduced no evidence that the RING_RETENTION_VOLTAGE
fuse value is ever used to guide “when” VCCR and VCCIO are supplied. ECF No. 591 at 6. Intel
alleges that VLSI ignores the evidence that VCCR is supplied to the C6 SRAM even when VCCR
is below RING_RETENTION_VOLTAGE conditions that are the opposite of what the claims
require. ECF No. 615 at 4. To avoid confronting this, Intel claims VLSI improperly characterizes
the plain meaning of the “when” limitations as an untimely claim construction dispute. Id.
VLSI responds that it offered substantial evidence that these limitations are met. Dr. Conte
demonstrated that the “first regulated voltage” (VCCR) is supplied when the first regulated voltage
is at least the value of RING_RETENTION_VOLTAGE, the minimum operating voltage. 2/23
Sealed Trial Tr. 13:2-16. He explained this using a demonstrative that showed the jury how VCCR
7
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 8 of 24
is supplied when it is at least the minimum voltage. PDX4.79. He also confirmed this infringement
argument using Intel's documents. 2/23 Sealed Trial Tr. 13:23-14:11.
This Court agrees that VLSI presented sufficient evidence to support the jury’s
infringement
finding.
Intel
argues
VLSI
introduced
“no
evidence”
that
the
RING_RETENTION_VOLTAGE fuse value is ever used to guide “when” the “first regulated
voltage” and “second regulated voltage” are supplied ECF No. 591 at 6. Yet Dr. Conte
demonstrated that RING_RETENTION_VOLTAGE acts as the threshold for when the voltages
are supplied. Id. at 7. He supported his testimony with Intel documents, which together provide
substantial evidence that Intel met this limitation. Intel rebuts by pointing to an admission by
VLSI’s expert that he did not refer to RING_RETENTION_VOLTAGE in describing multiplexer
operation, but this argument merely introduces a factual dispute. The significance of this dispute
is questionable, given that the claims only require that the “first regulated voltage” rise to at least
the minimum operating voltage rather than require the minimum operating voltage “to guide.”
ECF No. 603 at 7. At best, Intel provides evidence of a legitimate factual dispute; at worst, Intel
attempts to prevail with a red herring. In either case, Intel petitions this Court to subvert the jury’s
role by weighing the evidence and making credibility determinations. That is not the Court’s task
at this stage. See Wellogix, 716 F.3d at 874. Accordingly, this Court preserves the jury’s verdict
and declines JMOL on this ground.
3. Substantial Evidence Supports the Jury’s Infringement Finding for The “First
Regulated Voltage” And “Functional Circuit” Limitations.
The asserted method claims recite that “while the second regulated voltage is provided as
the operating voltage of the memory, the first regulated voltage is provided to the functional
circuit.” ’373 Patent 13:26–28, 41–44, 45–52. The asserted apparatus claims “a power supply
8
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 9 of 24
selector” that “supplies the second regulated voltage” to the memory “while the second regulated
voltage is supplied as the operating voltage, the circuit uses the first regulated voltage.” Id. at 14:8–
15, 14:20–23. At trial, VLSI identified Intel’s VCCR as the “first regulated voltage” and the
VCCIO as the “second regulated voltage.” ECF No. 591 at 7.
Intel contends that the functional circuit in its products are not provided with and do not
use the first regulated voltage while the second regulated voltage is provided to the memory. Id. at
6. Intel’s witnesses testified the “first regulated voltage” is actually unregulated and the ring
domain components are inoperable during the state in which the “second regulated voltage”
powers the C6 SRAM bitcells. Id. Moreover, Intel contends that Dr. Conte admitted that the
Id. Thus,
Intel alleges that VLSI identifies no evidence supporting its claim that VCCR is regulated during
Package C7. Id. Intel complains that VLSI’s reliance on one Intel document does not account for
its engineer’s testimony that the “first regulated voltage” is not actually regulated during Package
C7. ECF No. 614 at 4. Consequently, Intel asserts JMOL is required because no reasonable jury
could find that the “first regulated voltage” is regulated and provided to or used by a “functional
circuit” while the “second regulated voltage” is “provided” or “supplied” to what VLSI identifies
as the claimed “memory.” ECF No. 591 at 8.
To rebut, VLSI cites Dr. Conte’s testimony that the “first regulated voltage” is regulated,
controlled, and non-zero for the ramp-down period while the “second regulated voltage” is
provided to the memory. 2/23 Sealed Trial Tr. 19:16-20:2. Dr. Conte further testified that the
circuit is using the voltage and that it is a regulated voltage for the entirety of the Package C7 state.
Id. VLSI corroborates Dr. Conte’s testimony with Intel’s internal documents that show that the
9
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 10 of 24
see also 2/25 Trial Tr. 870:20–872:25. VLSI argues that Dr. Conte’s testimony, coupled with
Intel’s documents, are substantial evidence that supported the infringement verdict. ECF No. 603
at 8.
This Court agrees with VLSI that substantial evidence supports the jury’s infringement
finding. JMOL is inappropriate if the record evidence is such that reasonable and fair-minded
people in the exercise of impartial judgment might reach different conclusions. Laxton, 333 F.3d
at 579. Here, Intel presents several points on which expert testimony conflict as to whether the
accused products supply a “first regulated voltage” and “second regulated voltage … when the
first regulated voltage is below the minimum operating voltage.” For example, Intel points out that
“the first regulated voltage” is not regulated at all, which directly contradicts Dr. Conte’s testimony
that the “first regulated voltage” is regulated, controlled, and non-zero. ECF No. 603 at 8. Intel’s
counters that VLSI lodged an unsubstantiated claim by basing it on an Intel document with no
reference to regulation. ECF No. 614 at 4.
This Court’s role, however, is not to resolve factual disputes, but to consider whether the
evidence mandates non-infringement as the only reasonable conclusion. At best, Intel has
demonstrated that there is a legitimate factual dispute over which reasonable minds could differ,
rendering JMOL improper. Intel’s core contention is that the relevant evidence failed to provide a
sound basis for the jury’s verdict because it clearly shows, among other things, that the “first
regulated voltage” is unregulated. But Intel takes for granted that the jury need only base its verdict
on legally sufficient evidence amounting to more than a mere scintilla. The conflicting expert
testimonies satisfy that evidentiary standard. Intel’s JMOL on this ground is therefore denied.
10
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 11 of 24
B.
Substantial Evidence Support’s The Jury’s Doctrine of Equivalents
Infringement Verdict for the ’759 Patent.
VLSI presented both literal and doctrine of equivalents infringement theories on the ’759
patent’s “request” limitation. The jury found that the Speed Shift feature in Intel’s products
infringes multiple claims of the ’759 patent under the doctrine of equivalents. Intel argues,
however, that JMOL of no infringement should enter because that verdict is contrary to law and
not supported by substantial evidence. ECF No. 591 at 8. Intel makes several legal arguments as
to why VLSI’s doctrine of equivalents claim should be barred. These include prosecution history
estoppel and claim vitiation. Id. at 9–10. The Court does not address those arguments in this
opinion because it already addressed them in its Order denying Intel’s Motion for Judgment of No
Infringement Under the Doctrine of Equivalents. See ECF No. 685. Those arguments are moot.
Still, Intel argues that JMOL should be granted because no reasonable juror could have found
infringement under the doctrine of equivalents.
In response, VLSI argues that there was substantial evidence sufficient for a reasonable
jury to find infringement. ECF No. 603 at 9. But the jury did not find that Intel literally infringed.
ECF No. 615 at 5. Whether there was substantial evidence supporting VLSI’s rejected claim of
literal infringement is irrelevant. To overturn the jury’s rejection of its literal infringement claim,
VLSI would have had to show that there was no substantial evidence from which a jury could
reasonably find non-infringement. Id. VLSI has not attempted to meet that burden. Id. Therefore,
this Court maintains the jury’s verdict that the accused products infringe under the doctrine of
equivalents and not under literal infringement.
11
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 12 of 24
1. Substantial Evidence Supports the Jury’s Infringement Finding for the
“Request” Limitation.
The ’759 Patent relates to the controlling clock frequency in an electronic device. The ’759
Patent discloses and claims a system in which a first master device of a plurality of master devices
provides a “request” to change a clock frequency of a high-speed clock “in response to a predefined
change in performance of the first master device.” ’759 Patent 7:66–8:15. A “programmable clock
controller” receives this “request” and provides outputs to independently control (1) a clock
frequency of a second master device coupled to a bus, and (2) a variable clock frequency of the
bus. Id. VLSI asserted that Intel’s Speed Shift feature infringed on claims 14, 17, 18, and 24 of the
’759 Patent.
The Court looks first to VLSI’s doctrine of equivalents theory and the evidence it used to
support its infringement claim. Dr. Conte testified that the combination of core 1 (“first master
device”) and core 1’s associated code in the PCU (“programmable clock controller”) provides the
claimed “request” in the ’759 Patent. Sealed Trial Tr. 55:1-11; 53:14-56:3. Specifically, he
explained that the core sends a Core_Active signal to the PCU whenever the core becomes active.
Trial Tr. 1419:2-18; 488:4-7. C0 residency counters, which are counters in the PCU, measure the
activity of the core over a predefined time interval when the core is active. Id. at 1419:12–18.
generates a
“request” for a higher or lower frequency. Sealed Trial Tr., 54:9-22; ECF No. 553-02 PDX4.205208, PDX4.210. VLSI’s literal infringement theory of “request” was based on the Core_Active
signals, but its equivalents theory was based on the output of the core in combination with the code
in the PCU.
The Court looks next to Intel’s arguments that the above evidence fails to show that its
products satisfy the “request” limitations. Intel asserts two reasons as to why its products cannot
12
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 13 of 24
meet the “request” limitations. ECF No. 591 at 10. First, the C0 residency counters in Intel’s
products are
” and are not sent “in response to a predefined change in
performance” as the claims require. Id. Intel claims that Dr. Conte admitted that C0 residency data
is “
” and “
”, thus contradicting the claims. Id. Second, Intel
alleges that VLSI’s equivalents theory is inconsistent with the claim, which requires the “request”
to be (1) “provided” or “sent” by the “first master device” and (2) “received” by the “clock
controller.” Id. Dr. Conte testified that the “request” in Intel’s products is provided by the
combination of core and its associated P-code in the PCU (“the first master device”). Id. at 11. But
Intel argues that this is improper because the claims do not allow the same component to both
provide and receive the request. Id. Thus, Intel argues that no reasonable jury could find the
“request” limitations met under the doctrine of equivalents. Id.
There are several problems with Intel’s argument. First, Intel mischaracterizes Dr. Conte’s
testimony by claiming that C0 residency information occurs “
” and “
” ECF No. 591 at 9. When asked whether his testimony
about period signals applies to the Core_Active requests, Dr. Conte responded in the negative and
further clarified that a
,
. ECF No. 603 at 10. Thus, the C0 residence
information is sent “in response to” changes in the Core_Active signals sent to the PCU. Dr. Conte
testified that the C0 residency counters
depending on
changes in Core_Active signals, which reflect changes in core performance. ECF No. 603 at 10.
Therefore, Dr. Conte’s testimony cannot be understood to vitiate the “request” limitation by
proving that C0 residency information is completely untethered to core performance, as Intel
suggests.
13
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 14 of 24
Second, Intel’s argument that VLSI’s equivalents theory is inconsistent with the claims
misconstrues Dr. Conte’s testimony. Dr. Conte did not testify that the claims require the PCU to
send and receive the claimed “request.” Id. Instead, he testified that a separated core in conjunction
with the PCU’s internal P-Code can generate the request received by the
. 2/23 Sealed Trial Tr. 55:1-19, 57:11-15. Intel understands this as suggesting that the
PCU both provides and receives the request, but VLSI’s demonstrative graphically depicts the core
and PCU as separate entities. Id. Dr. Conte distinguished the core and PCU when he testified that
“the core and Core 1’s P-Code” and “
” provided and received the
request, respectively. Id. Moreover, to the extent Intel contends the “first master device” and
“programmable clock controller” must always comprise entirely separate and distinct circuits, Intel
appears to be asking this Court for a new claim construction. Even if this Court accepts Intel’s
characterization of the claims, Intel cannot prevail because it ultimately proposes a new claim
construction, which it waived by omission. By neglecting to raise the issue to this Court in its Rule
50(a) motion, Intel has waived its right to bring a Rule 50(b) motion on this ground. And the
Federal Circuit has held that the “doctrine of equivalents does not require a one-to-one
correspondence between the accused device and that disclosed in the patent.” Intel Corp. v. ITC,
946 F.2d 821, 832 (Fed. Cir. 1991); see also Ethicon Endo-Surgery, Inc. v. U.S. Surgical Corp.,
149 F.3d 1309, 1320 (Fed. Cir. 1998) (“[o]ne-to-one correspondence of components is not
required, elements or steps may be combined without ipso facto loss of equivalency.”).
This Court agrees with VLSI that substantial evidence supported the jury’s finding that the
Speed Shift feature satisfied the “request” limitation under the doctrine of equivalents. Intel claims
JMOL of no infringement is proper for two reasons, namely that expert testimony vitiated the
“request” limitation, and that the expert proposed an equivalents theory inconsistent with the claim.
14
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 15 of 24
But as explained above, Intel’s arguments fail on both points. VLSI’s presentation of Dr. Conte’s
testimony was sufficient to support a finding that the output of the core in combination with the
code in the PCU provides the claimed “request.”
2. Substantial Evidence Supports the Jury’s Infringement Finding for the “Provide
. . . As An Output to Control” Limitations.
Claims 14 and 17 require “provid[ing] the clock frequency of the high-speed clock as an
output to control” both “a clock frequency of a second master device” and a “clock frequency of
the bus,” and claims 18 and 24 require “the clock controller configured to output a clock frequency
of a high-speed clock to control the variable clock frequency of the bus and to control a clock
frequency of a second master device coupled to the bus.” ’759 Patent at 8:63–64, 9:15–18, 9:26–
30, 10:21–24.
The Court looks first to VLSI’s doctrine of equivalents theory and the evidence it used to
support its infringement claim. Dr. Conte testified that Intel’s products include a PCU and clock
circuit that has instructions to “provide the clock frequency of the high-speed clock as an output
to control a clock frequency of a second master device” 2/23 Sealed Trial Tr. 44:9-46:25, 47:1750:17. That constitutes the programmable clock controller. Id. Dr. Conte also testified that the
programmable clock controller includes instructions to “provide the clock frequency of the highspeed clock as an output to control the variable clock frequency of the bus.” Id. at 50:18-52:20.
VLSI contends this is substantial evidence to support its theory that Intel’s products meet the
“provide . . . as an output to control” limitations. ECF No. 603 at 13.
Intel alleges that what VLSI identified as the “second master device” in the accused
products is
ECF No. 591 at 12. Because
control the clock frequency of the cores and the clock frequency of the bus, Intel argues that
15
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 16 of 24
VLSI’s infringement theory is contrary to the plain meaning of the claims, which stipulates that
the same clock frequency must be “output to control” both the second master device and the bus.
Id. at 13. According to Intel,
is not output to either the second master device or the
bus—it is output to the
respectively—and
control the second master device and the bus. Id.
Therefore, Intel concludes that no reasonable jury could find that the separate and independent
control structure of Intel’s products meets these claim limitations. Id.
VLSI responds that Intel’s non-infringement theories ignore the claim language and the
law. ECF No. 603 at 14. First, the claims do not require controlled clocks to have the same
frequency and allow for different frequencies. Id. Second, Intel says “the same clock frequency
must be ‘output to control’ both the second master device and the bus.” ECF No. 591 at 13. Dr.
Conte confirmed that the accused products meet this limitation when he explained that the same
is provided “as an output to control” both the “clock frequency of a
second master device” and the “variable clock frequency of the bus.” ECF No. 603 at 14. Third,
the claims do not prohibit the use of intermediate components in effectuating that control,
rendering Intel’s non-infringement arguments based on the
and
contrary
to law. Id. at 15. The claims use the transitional term “comprising,” which “is inclusive or openended and does not exclude additional, unrecited elements.” CollegeNet, 418 F.3d at 1235. As Dr.
Conte confirmed, “a system comprising” in the claims means that “as long as all of the elements
in the claim are present, the fact that one is adding additional features or elements would not change
the question of whether a product infringes.” ECF No. 603 at 15.
This Court agrees that VLSI presented substantial evidence supporting the jury’s
infringement finding for the “provide . . . as an output to control” limitation. Intel argues that
16
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 17 of 24
different clock frequencies separately output by the
and
to a second master device and
the bus do not satisfy the claim’s plain meaning, yet in doing so Intel draws a distinction without
a difference.
As VLSI points out, Intel’s accused products include a PCU and clock circuit with
instructions to “provide the clock frequency of the high-speed clock as an output to control a clock
frequency of a second master device” and “the variable clock frequency of the bus.” ECF No. 603
at 13. Two critical distinctions between the accused products and the claim language, according to
Intel, are that the accused products leverage intermediate hardware—that is, the
and
instead of a PCU and clock circuit directly—to provide an output signal, and that the signals are
transmitted at different frequencies. ECF No. 591 at 12. But the claims do not require that the highspeed clock transmit its output directly to the second master device or the bus. ECF No. 603 at 14.
Instead, they recite a high-speed clock that produces an output to control the second master device
or bus. Id. The claim’s plain language does not limit “control” to direct control as Intel argues and,
therefore, covers the
and
as a means for effectuating that control, especially when
considering the claim’s open-ended transitional term, “comprising.” Further, the claims do not
require the controlled clocks to have the same frequency, and if they did, expert testimony confirms
that
of the same frequency to both the “clock frequency of a second
master device” and the “variable clock frequency of the bus.” Id. at 13.
Ultimately, Intel fails to meet its burden for JMOL. Even if this Court accepts Intel’s
arguments, they fail to prove that reasonable minds could only conclude that judgment of no
infringement is proper. At best, Intel presents a legitimate factual dispute and critiques the logical
inferences underlying the jury’s conclusion. Yet JMOL is only proper when the evidence presented
shows the jury would not have been at liberty to disbelieve and pointing out legitimate factual
17
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 18 of 24
disputes over which reasonable minds could differ fails to meet this bar. The jury relied on more
than a scintilla of evidence when it weighed both parties’ expert opinions, including the factual
disputes embedded therein, and rendered its verdict. Therefore, this Court declines to grant JMOL
of no infringement on this ground.
C.
Intel’s Motion for JMOL on Claims Not Argued at Trial Is Improper.
VLSI represented that it was no longer asserting indirect infringement and claimed that it
reserved the right to assert those claims later. ECF No. 591 at 13. But VLSI did not dismiss its
inducement or contributory infringement claims, and Intel’s declaratory judgment counterclaims
on both issues are still live. Id. Thus, Intel argues that this Court should enter judgment of no
induced and no contributory infringement because VLSI did not offer any evidence at trial on
either issue. Id. However, Federal Rule of Civil Procedure 50 states that relief in this context is
appropriate only “if a party has been fully heard on an issue during a jury trial.” Rembrandt
Wireless Techs., LP v. Samsung Elecs., 2016 WL 633909, at *5 (E.D. Tex. Feb. 17, 2016). ECF
No. 603 at 15. Here, the jury could not have fully heard contributory or indirect infringement issues
because, by Intel’s own admission, VLSI did not assert evidence supporting either claim at trial.
Accordingly, this Court denies Intel’s motion on this ground as improper.
II.
Substantial Evidence Supports the Jury’s Verdict on Invalidity for the ’759 Patent.
A patent is invalid if before the patented invention, “the invention was made in this country
by another inventor who had not abandoned, suppressed, or concealed it.” 35 U.S.C. § 102(g) (preAIA). At trial, Intel asserted that claims 14, 17, 18, and 24 of the ’759 patent were invalid. The
jury found that Intel did not prove by clear and convincing evidence that those claims of the ’759
patent were invalid. ECF No. 564. Intel argues, however, that this Court should enter JMOL of
invalidity because a reasonable jury “could only conclude” that the Yonah processor anticipates
18
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 19 of 24
each asserted claim. ECF No. 591 at 14. According to Intel, VLSI based its contrary assertions
solely on its argument that Yonah did not have a hardware-based “programmable clock controller,”
but the claim language and specification foreclose VLSI’s argument. Id.
VLSI argues there was substantial evidence for a reasonable jury to conclude that Yonah
did not anticipate the ’759 Patent. VLSI alleges that Intel provided bare assertions that Yonah
purportedly had a “programmable clock controller” while failing to identify what in Yonah met
these limitations. ECF No. 603 at 16. VLSI claims that Intel’s witnesses failed to explain how any
alleged “programmable clock controller” in Yonah meets the limitations of “having an embedded
computer program therein” or being “coupled to the arbiter and coupled to the first master device.”
Id. at 17. VLSI further contends that Dr. Conte provided the jury with even more reason to reject
Intel’s flawed theory in rebuttal. Id. According to Dr. Conte, “Yonah is the old approach” to speed
changes, whereas the ’759 patent and Intel's infringing processors use “the new approach.” Id. In
the ’759 patent’s new approach, speed changes are controlled by a “computer-in-a-computer,” i.e.,
a “programmable clock controller with an embedded computer program.” Id. By contrast, the
operating system in Yonah made speed control decisions. Id. Yonah did not include a “PCU,” nor
any other “programmable clock controller” with “an embedded computer program” as required by
the ’759 patent claims. Id.
After losing at trial, Intel argued that “the claim language and specification make clear that
a hardware-based controller is not required and that software may be used.” Id. at 18. VLSI argues
that Intel ultimately seeks a new claim construction, but Intel waived that right by failing to request
a one before trial. Id. Notwithstanding Intel’s waiver, VLSI alleges that Intel’s new argument
contradicts the plain claim language. Id. Claim 14 recites a system with hardware components,
such as the programmable clock controller with an embedded computer program therein. See PTX-
19
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 20 of 24
5 (“A system comprising: a bus …; a first master device coupled to the bus …; and a programmable
clock controller having an embedded computer program therein....”). Claim 18 similarly recites a
system with hardware components that are “coupled to” each other. See PTX-5 (“A system
comprising: a bus …; a first master device coupled to the bus; an arbiter coupled to the bus and
coupled to the first master device …; and a clock controller coupled to the arbiter and coupled to
the first master device ….”). Contrary to Intel’s assertion, the claims thus do include requirements
for a hardware-based programmable clock controller. Even if Intel had presented its new theory,
VLSI suggests that a reasonable jury could reach different conclusions as to whether the “operating
system [being] executed in the cores of Yonah” is a “programmable clock controller having an
embedded computer program therein,” making JMOL improper. ECF No. 603 at 19.
This Court agrees that substantial evidence supported the jury’s invalidity verdict for two
reasons. First, Intel could not affirmatively show that Yonah anticipated each claim of the ‘759
patent, much less by clear and convincing evidence. Second, VLSI sufficiently distinguished the
’759 patent from Yonah using expert testimony.
Although Intel contends that the operating system in Yonah’s anticipates the
“programmable clock controller” with “an embedded computer program” disclosed by the ’759
patent claims, it failed to identify what they were referring to specifically. ECF No. 603 at 16.
Further, Intel could not clearly articulate what in Yonah anticipated the ’759 patent’s limitations
or how Yonah allegedly met those limitations. Id. Instead, Intel resorts to criticizing VLSI’s
distinctions between Yonah and the ’759 patent. But a party cannot meet an affirmative burden of
proof—here, clear and convincing evidence—by pointing to deficiencies in the opposing party’s
validity arguments. Indeed, Intel’s generic allegations dismissing VLSI’s expert testimony as
“conclusory” are themselves conclusory and insufficient to meet their affirmative burden of proof.
20
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 21 of 24
By contrast, VLSI presented substantial evidence supporting the jury’s verdict in two ways.
First, VLSI pointed out deficiencies in Intel witness testimony, including the absence of any
reasonable statement connecting Yonah’s “programmable clock controller” and the limitations of
“having an embedded computer program therein” or being “coupled to the arbiter and coupled to
the first master device.” ECF No. 603 at 16. Second, VLSI’s distinction between Yonah as the old
approach and the ’759 patent as the new approach carries more weight than Intel acknowledges.
Contrary to Intel’s portrayal, Dr. Conte reasonably substantiated his testimony by pointing to
specific differences between the two approaches—namely, the ’759 patent’s programmable clock
controller with an “embedded computer program” as the mechanism for triggering speed changes,
rather than Yonah’s operating system. Id. at 17. Intel believes it made clear to the jury that Yonah
disclosed a combination of hardware and software running on a chip, not the operating system, but
that provides even more reason to preserve the jury’s finding. Under Intel’s theory, the jury would
have clearly understood Intel’s arguments when deciding against them, which increases the
likelihood that they made an informed decision rather than a clearly unreasonable one warranting
reversal.
“Courts grant JMOL for the party bearing the burden of proof only in extreme cases, when
the party bearing the burden of proof has established its case by evidence that the jury would not
be at liberty to disbelieve and the only reasonable conclusion is in its favor.” Mentor H/S, Inc. v.
Medical Device Alliance, Inc., 244 F.3d 1365, 1375 (Fed. Cir. 2001). Here, both parties’ experts
provided plausible invalidity theories to the jury, which the jury was free to believe or discredit as
it saw fit. Given the technical distinctions Dr. Conte drew between the ’759 and the Yonah
processor, this Court believes that the evidence did not weigh so heavily in Intel’s favor that the
jury “could have only” rendered the ’759 patent invalid as Intel suggests. VLSI provided more
21
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 22 of 24
than a mere scintilla of legally sufficient evidence to give the jury the liberty to disbelieve either
party and, therefore, adequately supported the jury’s invalidity verdict.
III.
Substantial Evidence Supported the Jury’s Damages Award.
The jury awarded a lump sum of $1.5 billion for infringement of the ’373 Patent and a
lump sum of $675 million for infringement of the ’759 Patent. ECF No. 564. Intel argues that it is
entitled to JMOL of no damages. ECF No. 591 at 17. First, Intel alleges that no reasonable jury
could value the asserted patents at $2.175 billion considering the asserted patents’ sales price, the
lack of a formal valuation of the asserted patents, and comparable agreements for lower amounts.
Id. Second, Intel asserts that the jury’s damages award is not supported by legally sufficient
evidence because it is based on noncomparable settlement agreements. Id.at 17–18. Under Federal
Circuit precedent, those noncomparable agreements cannot support the jury’s damages award. See
Lucent Techs., Inc. v. Gateway, Inc., 580 F.3d 1301, 1332, 1340 (Fed. Cir. 2009) (reversing denial
of JMOL in part because damages expert relied on noncomparable licenses). Third, Intel argues
that the jury’s award is not supported by Dr. Sullivan’s opinions. Id. at 18–19. Accordingly, Intel
requests that this Court enter JMOL of no damages. Id.
VLSI responds that it presented substantial evidence supporting the damages award. First,
VLSI argues Intel’s patent valuation arguments rely on evidence favorable to the moving party
that the jury is not required to believe. ECF No. 603 at 19. Second, VLSI alleges that there is no
support for Intel's speculation that the jury based its damages numbers on noncomparable
agreements. Id. at 20. According to VLSI, Intel “does not really know” if the jury based its award
on any agreement, and a source matching an awarded number does not somehow “put the number
. . . off limits to the jury.” Spectralytics, Inc. v. Cordis Corp., 649 F.3d 1336, 1346-47 (Fed. Cir.
2011). Id. Third, VLSI claims that Intel rehashes arguments previously rejected by this Court in
22
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 23 of 24
Daubert motions in questioning the Dr. Sullivan’s reliability. Id. Consequently, VLSI requests that
this Court deny Intel’s motion for JMOL of no damages. Id.
This Court agrees with VLSI that the jury’s damages award was supported by sufficient
evidence. First, Intel alleges that the jury’s $2.175 billion dollar verdict contravenes
“overwhelming evidence” that the asserted patents are lower in value. But the jury considered
evidence presented by both parties. This included Intel’s “overwhelming evidence” of lower patent
family prices, no prior ownership, and non-comparable agreements. VLSI also presented
substantial evidence that Intel made over
from the nearly one billion products it sold
with the infringing performance and power saving features of the patents. 2/24 Trial Tr. 593, 65374; PDX7.74-76. The jury exercised reasonable discretion in crediting VLSI’s damages evidence
over Intel’s damages evidence. Second, Intel contends that the jury improperly based its verdict
on a non-comparable Nvidia agreement, and that Dr. Sullivan admitted as much. Id. Intel refuted
its own argument, however, when it admitted that it “does not really know” whether the jury based
its figure on any agreement. ECF No. 603 at 20. Any argument that the verdict was based on a
purportedly non-comparable agreement is entirely speculative. Third, Intel argues that VLSI’s
damages case was unreliable, in part, because VLSI rooted it in unreliable and speculative expert
testimony. ECF No. 615 at 10. But this Court has already considered and rejected these arguments
in Intel’s prior Daubert motions and its Rule 59 Motion. By asking this Court to render JMOL
with no additional compelling reasons, Intel seeks to deploy the same reliability arguments and
obtain different results. And contrary to Intel’s argument that the damages numbers were not put
into the record, Dr. Sullivan explained Intel’s revenues to the jury, and any omissions were because
he refrained from disclosing Intel’s confidential information. 2/24 Trial Tr. 651:23–658:2;
PTX3903, PTX3904.
23
Case 6:21-cv-00057-ADA Document 703 Filed 05/10/22 Page 24 of 24
Ultimately, Intel questions the degree to which the evidence justified the jury’s verdict, but
that is not the relevant question. Rather than consider how well the totality of the evidence
supported the jury’s finding, this Court must limit itself to determining whether the jury based its
verdict on legally sufficient evidence that amounts to more than a scintilla. Here, the jury satisfied
that standard by relying on the damages expert’s testimony that Intel made over
dollars
by infringing VLSI’s patents. Intel disagrees with the outcome but falls short of proving that a
verdict of no damages is the only reasonable conclusion from the evidence. Therefore, the jury’s
finding stands.
IV. CONCLUSION
For the reasons above, the Court finds that VLSI produced sufficient evidence to support
the jury’s verdict in this case. The Court therefore DENIES Intel’s Motion for Judgment as a
Matter of Law (ECF No. 591).
SIGNED this 18th day of March, 2022.
ALAN D ALBRIGHT
UNITED STATES DISTRICT JUDGE
24
Disclaimer: Justia Dockets & Filings provides public litigation records from the federal appellate and district courts. These filings and docket sheets should not be considered findings of fact or liability, nor do they necessarily reflect the view of Justia.
Why Is My Information Online?