Chan v. Equinix, Inc. et al

Filing 67

Order by Judge Vince Chhabria granting in part and denying in part 54 Motion to Dismiss.(vclc3, COURT STAFF) (Filed on 1/6/2025)

UNITED STATES DISTRICT COURT
NORTHERN DISTRICT OF CALIFORNIA

UNIFORMED SANITATIONMEN’S
ASSOCIATION COMPENSATION
ACCRUAL FUND,

Case No. 24-cv-02656-VC
ORDER RE MOTION TO DISMISS

Plaintiff,

Re: Dkt. No. 54

v.
EQUINIX, INC., et al.,
Defendants.

The motion to dismiss is granted in part and denied in part. This ruling assumes the
reader is familiar with the complaint’s allegations, the applicable legal standards, and the
arguments made by the parties.
I
The defendants ask the Court to take judicial notice of 26 exhibits. The plaintiff objects to
Exhibits 11-16. The request is granted as to the unopposed exhibits. As the plaintiff
acknowledges, judicial notice can be taken for SEC filings, analyst reports, earnings call and
Analyst/Investor Day transcripts, and court filings “for the purpose of determining what
information was available to the market.” In re Splunk Inc. Securities Litigation, 592 F. Supp. 3d
919, 930 (N.D. Cal. 2022); see Khoja v. Orexigen Therapeutics, Inc., 899 F.3d 988, 999 (9th Cir.
2018). The Court does not take notice of them for the truth of the statements they contain.
The request is also granted as to Exhibits 11-16. The plaintiff argues that these exhibits,
which are Equinix’s publicly available investor presentations, should not be judicially noticed for
the truth of the statements they contain because they are not referenced in the complaint and their

accuracy is questionable. As with the other exhibits, the Court takes notice of Exhibits 11-16
only for the purpose of determining the information that was available to the market, and not for
the truth of the statements they contain.
II
The complaint puts forward two claims under Section 10(b) of the Securities Exchange
Act and Rule 10b-5 based on independent theories. The first is that the defendants manipulated
Equinix’s AFFO numbers (that is, “adjusted funds from operation”) by misclassifying recurring
capital expenditures as non-recurring capital expenditures. The second is that the defendants
made misleading statements about Equinix’s power capacity and room for growth by failing to
disclose that the company was overselling its power capacity upward of 150%. The complaint’s
allegations are sufficient to support the first claim, but not the second.
A. AFFO
The complaint contains the following allegations about Equinix’s AFFO accounting,
which are assumed true for the purposes of this motion.
In 2014, Equinix converted to a real estate investment trust and adopted its methodology
for calculating AFFO. Key to this methodology was the treatment of capital expenditures, which
the defendants and analysts often referred to as “CapEx.” Specifically, the methodology
distinguished between “recurring” capital expenditures and “non-recurring” capital expenditures.
This distinction was important because AFFO represented funds from operation minus recurring
capital expenditures (among other adjustments). Non-recurring capital expenditures were not
deducted. In other words, classifying an expense as a recurring capital expenditure would
decrease AFFO; classifying it as a non-recurring capital expenditure would leave AFFO
undisturbed.
Equinix defined recurring capital expenditures to include “Maintenance & SPOF,
Reconfiguration Install, and IT/Product/HQ.” It provided examples of “Maintenance & SPOF,”
such as “end-of-life equipment replacement (i.e. chillers, strings of batteries, sprinkler system,
etc.).” Am. Compl. ¶ 53, Dkt. No. 43. Equinix also described recurring expenditures as expenses

2

“to extend the useful life of its IBX data centers or other assets that are required to support
current revenues.” Id. ¶ 67 & n.6.
The company defined non-recurring capital expenditures to include “Initial/Custom
Install; Efficiency Improvement; IT/Network Projects; Special Projects; IBX Construction; and
Capacity/Product Enhancement.” Id. Examples of non-recurring expenses included “energy
efficiency projects to reduce cooling costs and improve system reliability” and “building out new
data center and/or expand new phases.” Id.
Throughout the class period, the defendants stated several times that Equinix’s high
AFFO numbers were due to “strong operating performance.” Id. ¶¶ 94, 71, 72, 76, 80, 83, 88, 92,
94, 95, 104, 112, 117, 121, 128, 134. For instance, during the May 6, 2020, earnings call, CFO
Keith Taylor said, “Global Q1 AFFO was $535 million, above our expectations on a constant
currency basis due to strong operating performance and lower-than-planned interest expense and
income taxes” and capital “expenditures were approximately $400 million, including seasonally
low recurring CapEx of $18 million.” Id. ¶ 83. Similarly, during the April 28, 2021, earnings
call, Taylor said, “Global Q1 AFFO was $627 million, meaningfully above our expectations due
to strong operating performance and lower seasonal recurring capital expenditures.” Id. ¶ 94.
However, the complaint alleges that the defendants’ reported AFFO numbers and
statements attributing the high AFFO numbers to strong operating performance were misleading
because Equinix was in fact artificially inflating its AFFO by misclassifying recurring capital
expenditures as non-recurring. To support that theory, the complaint relies primarily on
information from a March 20, 2024, short-seller report, the Hindenburg Report. The Hindenburg
Report estimated that “Equinix’s manipulation of maintenance CapEx has resulted in a
cumulative $3 billion boost to reported AFFO since 2015,” based off its analysis and interviews
with “37 former Equinix employees.” Id. ¶¶ 153, 155. Below are some of the quotes from former
employees contained in the Hindenburg Report. It bears noting that some employees use the term
“CapEx” to refer to both recurring and nonrecurring capital expenditures. Some refer to recurring
capital expenditures as “maintenance CapEx” while referring non-recurring capital expenditures

3

as “growth CapEx” or “expansion.” Still others use “CapEx” to refer only to nonrecurring capital
expenditures, using the term “OpEx” (or, operational expenses) to refer to recurring capital
expenditures. But in all of the quotes, the employees describe efforts to classify what seem like
recurring capital expenditures as non-recurring:
•

A former executive stated that there was “pressure from management to push as much of
CapEx into expansion.” Id. ¶ 156. A former director of operations stated, “So anything,
any operation that’s completed of significance, you look for all different types of creative
accounting ways to show it as CapEx.” Hindenburg Report 24, Ex. 2, Dkt. No. 54-3. That
director also said, “I mean, it’s really about thinking out of the box about how you
approach things. And if you can, if your accounting department will agree with you.” Id.
A former data center manager said, “They would assess all our expenses versus our
recurring CapEx . . . [They were] always trying to squeeze your expenses, your OpEx
[operating expenses], because it was ‘dirty spend.’” Id. A former finance director said,
“They [operating teams] have a lot of pressure on the operating budget. So, the operating
cost and then they try to squeeze stuff that is kind of gray into the CapEx area and move
it out of the OpEx area.” Id. at 25. And a former mergers and acquisitions manager who
“directly interacted with a finance lead on projects” said that the project’s experts asked
him “to do every machination to try and make a CapEx versus OpEx.” Id.

•

A former director “described that the company would obtain new serial numbers for
refurbished chillers so it could then be accounted for as a ‘new’ item post-repair, and
sometimes recognized as growth CapEx.” Am. Compl. ¶ 157. The director said, “Chillers
stay in place. You don’t replace chillers. All you do is rebuild them. There’s been some
debate on this. What happens and what you work towards is you do a chiller overhaul.
And you’d work with your chiller vendor to give you a new serial number on the unit . . .
A new serial number with the unit, for all intents and purposes, looks, talks, walks like a
brand new installation.” Id.

•

A former operations director stated “that Equinix would classify otherwise routine battery

4

replacements as growth CapEx by characterizing this activity as replacing a ‘battery
system.’” Id. ¶ 158. The director said, “So replacing a battery system is a capital
improvement. Non-recurring . . . As long as it’s part of that, as long as you replace the
entire battery system . . . All you do is replace the batteries within the cabinets. And
that’s considered the system.” Id. Multiple former employees also stated that Equinix
improperly classified battery replacements as non-recurring capital expenditures.
•

A former operations director said they would classify replacing lightbulbs as nonrecurring expenditures. The director said, “Say you changed out fluorescents to LED light
bulbs, that’s a capital improvement. You’re not replacing lightbulbs, you’re
enhancing . . . Say you have a facility. You needed to change light bulbs . . . Or you had
to change all the ballasts in your light fixtures because they were just at age. Which is
costly. You need an electrician to come in to take it down. You need to unwire it and
rewire it. Now, if you were to go back in and now convert those florescent bulbs or
systems with the ballast and the bulbs to an LED light, now you’re performing an ‘energy
efficiency project.’” Id. ¶ 159. And that director confirmed to Hindenburg they would
classify that expenditure as non-recurring and stated, “This is one of the tricks that the
operations teams use to say, well, this is not ongoing. This is non-recurring CapEx.” Id.

•

A former director stated that Equinix would press vendors to issue new SKUs to lump
operating costs into larger purchases and then record them as non-recurring capital
expenditures. That director said, “So you can look for, even as far as buying tool sets. So
one might not qualify. But if you go over a certain threshold and you buy a set of them,
maybe five of them . . . that’ll count towards your CapEx . . . You can also work with
your vendors, your suppliers and create unique SKUs. So if I was going to buy a tool set
that was say $500, right. Had a single SKU. Initially that would be considered OpEx. But
if you went to your vendor and you said, ‘hey, why don’t you create a SKU with a five
pack of those tool kits and get me a $2,500?’ Now it’s CapEx.” Hindenburg Report 25,
Ex. 2, Dkt. No. 54-3.

5

The complaint also relies on a report from Barclays that was issued in September 2022—
roughly 18 months before the Hindenburg Report. The Barclays Report expressed concern that
Equinix had been overinflating AFFO by an average of 15% each year since 2014. That report
stated that Barclays did “not see a plausible reason why CapEx accounting changed after 2014,
other than to reflect higher AFFO,” and suggested that “some non-recurring growth capex would
be better classified as recurring.” Am. Compl. ¶¶ 108–110.
On March 25, 2024, five days after the Hindenburg Report was released, Equinix issued a
press release announcing its Audit Committee was investigating the issues referenced in the
Report and stating that Equinix had received a subpoena from the U.S. Attorney’s Office for the
Northern District of California. On May 8, 2024, Equinix announced that its Audit Committee
found no accounting errors. On a May 9, 2024, earnings call, Equinix announced it was changing
its methodology for classifying capital expenditures by adding a new category called
“redevelopment” CapEx which would represent “investments to enhance the capacity, efficiency
and operating standards of facilities” and “investments that are intended to meaningfully extend
the economic life of assets.” Id. ¶ 172. The new “redevelopment” category would be considered
non-recurring and would not be deducted in the AFFO calculation.
1. Actionable Misrepresentations and Falsity
The defendants argue that Equinix’s AFFO numbers are matters of opinion because the
definitions of recurring and non-recurring capital expenditures are often a matter of discretion.
That may be true, but the plaintiff’s theory is not that Equinix’s AFFO numbers are categorically
false or misleading. The theory is that they are misleading given the definitions of recurring and
non-recurring capital expenditures that Equinix itself articulated in its 2014 presentation rolling
out its new methodology.
The defendants next argue that the AFFO numbers were not false or misleading because
Equinix disclosed that “efficiency improvements” would be counted as non-recurring. In the
abstract, it’s certainly possible that some expenditures, like the replacement of chillers, could be
considered “efficiency improvements.” But Equinix clearly stated that “end-of-life equipment

6

replacement (i.e. chillers, strings of batteries, sprinkler system, etc.)” would be categorized as
recurring. A reasonable investor, having seen that statement, would have expected chiller and
battery replacements to be classified as recurring and deducted in the AFFO calculations.
The defendants also emphasize that the Hindenburg Report is a short-seller report that
relies on anonymously sourced allegations. It’s true that confidential witness statements must be
scrutinized carefully. And doubly so when they come from a short-seller report. See In re
DraftKings Inc. Securities Litigation, 650 F. Supp. 3d 120, 154 (S.D.N.Y. 2023). But a securities
fraud complaint is not automatically doomed just because it relies primarily or exclusively on a
short-seller report quoting anonymous sources. It depends on the amount of detail and
specificity, the corroborative nature of the allegations, the plausibility of those allegations given
the context, the number of sources quoted, and other factors. See In re QuantumScape Securities
Class Action Litigation, 580 F. Supp. 3d 714, 732 (N.D. Cal. 2022); see also Zucco Partners,
LLC v. Digimarc Corp., 552 F.3d 981, 995 (9th Cir. 2009), as amended (Feb. 10, 2009).
Here, the Report’s allegations are specific not only about the former employees’ roles and
job titles, but also about the types of expenses that were misclassified as non-recurring (e.g.,
chillers, batteries, and lightbulbs) and the method by which they were misclassified (e.g., through
asking vendors for new SKU numbers and bundling purchases as a “system” change).
The Report’s allegations are also corroborated by and consistent with other sources. For
instance, the Report contains separate allegations—discussed further below—about overselling
power. Those are directly corroborated by the complaint’s confidential witness (FE1) and by TD
Cowen’s March 21, 2024, report, which stated that TD Cowen’s “conversation with Equinix” led
them to believe there is a degree of overselling capacity in their data centers. Am. Compl. ¶ 166.
The AFFO-related allegations aren’t directly corroborated, but they are consistent with the
Barclays Report suggesting that “some non-recurring growth capex would be better classified as
recurring,” id. ¶ 109. The Hindenburg Report’s allegations are also corroborated somewhat by
the fact that Equinix changed its capital expenditure classifications after its internal audit,
creating a new non-recurring category called “Redevelopment CapEx” which was defined to

7

include “investments to enhance the capacity, efficiency and operating standards of facilities”
and expenses that “are intended to meaningfully extend the economic life of assets.” Id. ¶ 172.
Although this post-hoc change is far from dispositive, it tends to bolster the inference that
Equinix’s prior capital expenditure definitions were misleading relative to how the company had
actually been classifying its expenses.
In addition, the Report’s allegations that there was mounting pressure from management
to misclassify expenditures are consistent with Equinix’s progressively declining recurring
capital expenditure numbers from 2014 through 2023. In 2015, after adopting the new
methodology, Equinix’s recurring capital expenditures were 4.4% of total revenue, and that
number trended downward to about 2.7% by 2023. See Am. Compl. ¶ 55. That narrative is also
corroborated somewhat by Chief Accounting Officer Simon Miller’s December 6, 2023,
response to a question about how Equinix defined recurring versus non-recurring capital
expenditures, in which he stated that Equinix put “a ton of pressure on the team to just rethink
how [they] engage in replacing some of the bigger parts of that infrastructure.” Id. ¶ 140.
Each of the above facts alone would be far from sufficient to plead falsity. And there are
potential innocent explanations for the trend of declining recurring capital expenditure numbers,
for the company’s decision to add a new capital expenditure category, and for Simon Miller’s
December 2023 statement. But put together, they tend to bolster the Hindenburg Report’s
reliability. Indeed, the defendants point to no specific factual inaccuracies in the Report that
would undermine its reliability or raise doubts as to whether the plaintiff’s counsel here properly
investigated the Report’s reliability. Cf. Long Miao v. Fanhua, Inc., 442 F. Supp. 3d 774, 804
(S.D.N.Y. 2020) (finding a short-seller report unreliable when it contained factual errors that
were then imported into the complaint). In all, the complaint raises a strong inference that
Equinix misclassified routine recurring capital purchases—like chillers, batteries, and
lightbulbs—as non-recurring to artificially inflate its AFFO numbers in a way that misled
investors.

8

2. Scienter
The defendants also argue that the complaint does not “state with particularity facts
giving rise to a strong inference” that the defendants “acted with the required state of mind.” 15
U.S.C. § 78u-4(b)(2)(A). Under the PSLRA, “a defendant is liable under Section 10(b) and Rule
10b-5 when he acts with scienter, a ‘mental state that not only covers intent to deceive,
manipulate, or defraud, but also deliberate recklessness.’” City of Dearborn Heights Act 345
Police & Fire Retirement System v. Align Technologies, Inc., 856 F.3d 605, 619 (9th Cir. 2017)
(quoting Schueneman v. Arena Pharmaceuticals, Inc., 840 F.3d 698, 705 (9th Cir. 2016)).
First, the core operations inference supports scienter because AFFO was of “such
prominence that it would be ‘absurd’ to suggest that management was without knowledge of the
matter.” Reese v. Malone, 747 F.3d 557, 576 (9th Cir. 2014), overruled on other grounds by City
of Dearborn Heights, 856 F.3d 605; see also Shenwick v. Twitter, Inc., 282 F. Supp. 3d 1115,
1145 (N.D. Cal. 2017). Equinix CEO Charles Meyers and CFO Keith Taylor both repeatedly
referred to AFFO as the company’s “lighthouse metric” and “core metric.” See Am. Compl.
¶ 143 (Meyers stating that AFFO per share is their “lighthouse metric and the bedrock of longterm value creation”); id. ¶ 113 (Meyers stating that AFFO per share is “very deeply ingrained
into our thinking about how we run the business”); id. ¶ 88 (Taylor stating that AFFO per share
is their “core metric”); see also, e.g., id. ¶¶ 76, 84, 105, 106, 118, 128.
Second, the complaint contains particularized allegations that Meyers and Taylor had
“actual access to the disputed information” because they repeatedly discussed capital
expenditures and commented on the reasons why AFFO was high and recurring capital
expenditures were low. Reese, 747 F.3d at 575; see also Shenwick, 282 F. Supp. 3d at 1147. For
instance, at the March 2, 2020, Morgan Stanley conference, Taylor said that Global Q1 AFFO
was above expectations because of “strong operating performance” and that “capital
expenditures were approximately $400 million, including seasonably low recurring CapEx of
$18 million.” Am. Compl. ¶ 83. Similarly, at a May 27, 2020, RBC conference, Meyers was
asked about Equinix’s lower-than-expected maintenance costs during COVID and replied,

9

“Generally, I wouldn’t say that that was really a factor for us in terms of affecting our
maintenance costs and maintenance CapEx . . . And Q1 was low for us, but I think it was more
an artifact of one pull forward into Q4, which is a common thing for us to do at the end of the
year is to try to get those things pulled in if we can get them done in Q4 and then probably a little
bit of things pushing out into Q2.” Id. ¶ 86. These comments suggest Meyers and Taylor were
monitoring capital expenditure numbers and aware of how they affected AFFO.
Finally, the fact that Meyers’s and Taylor’s bonuses were directly tied to hitting certain
AFFO targets supports the inference of scienter, although this fact would not be sufficient on its
own. See No. 84 Employer-Teamster Joint Council Pension Trust Fund v. America West Holding
Corp., 320 F.3d 920, 944 (9th Cir. 2003) (fact that the defendants’ eligibility for stock options
and executive bonuses were tied to the company’s financial performance supported inference of
scienter). Under a holistic view of the allegations, the complaint has created a strong inference of
scienter.
3. Loss Causation
The complaint sufficiently alleges loss causation following a series of partial disclosures.
Here, the defendants primarily argue that the release of the Hindenburg Report wasn’t a
corrective disclosure. They first argue that the Report was based on publicly available
information and therefore didn’t provide any new information to the market. But the Report was
in fact largely based on non-public information from former employees. And, in any event, a
corrective disclosure can be based on solely publicly available information so long as the
complaint “plausibly” suggests “that other market participants had not done the same analysis.”
In re BofI Holding, Inc. Securities Litigation, 977 F.3d 781, 794 (9th Cir. 2020).
The defendants next argue that Hindenburg had an incentive to tank the stock price by
issuing the Report. The “relevant question for loss causation purposes is whether the market
reasonably perceived” the Report’s “allegations as true and acted upon them accordingly.” Id. at
792. Here, the complaint alleges facts suggesting the market reasonably believed that the
Report’s allegations were credible despite Hindenburg’s short position. As discussed above, the

10

allegations from anonymous former employees were highly specific and consistent with an
earlier report by Barclays. Following the Report’s release, other analyst reports discussed the
Report and downgraded Equinix, suggesting that the market was taking the AFFO allegations
seriously. See Am. Compl. ¶ 163 (HSBC Global Research); ¶ 164 (Wells Fargo); ¶ 165 (BMO
Capital Markets).
B. Power Capacity
The plaintiff alleges a second Section 10(b) claim based on a separate theory, which is
that Equinix misled investors by failing to disclose that it was overselling power, or in other
words, that it was selling contracts for use of its data centers that exceeded the data centers’
actual power capacity. Relatedly, the complaint alleges that the defendants misled investors by
saying they had room to capitalize on the power-intensive AI boom when Equinix in fact was
overselling its power capacity already.
The complaint has sufficiently alleged that Equinix oversold its power capacity and that it
never disclosed that fact to its investors. But, as discussed at the hearing, the primary problem is
that the complaint does not allege exactly how that fact was “material.” Siracusano, 563 U.S. at
38. In other words, the complaint does not explain how disclosure of the fact that Equinix was
overselling would have “significantly altered the total mix of information made available” to the
reasonable investor. Id. (quoting Basic Inc. v. Levinson, 485 U.S. 224, 231–32 (1988)). It seems
possible that the overselling put Equinix at high risk for a power outage or otherwise put Equinix
in a financially precarious position, such that a reasonable investor would have found that fact
material. But it’s also possible that Equinix oversold because it knew that its customers would
not utilize their full contractual rights and therefore overselling didn’t meaningfully increase the
risk of a power outage or constrain Equinix’s growth. In that scenario, the defendants’ failure to
disclose the fact that they oversold power and their statements that Equinix had room to grow
would not necessarily be misleading. Without more specific allegations as to the importance of
the power capacity allegations, the plaintiff’s claim based on this theory must be dismissed.

11

III
The defendants seek dismissal of the plaintiff’s Section 20(a) claims for failure to plead a
primary violation of securities law, but because the Section 10(b) claim based on the AFFO
allegations survives, the Section 20(a) claim based on those allegations survives as well. The
Section 20(a) claim based on the power capacity allegations is dismissed.
*

*

*

Dismissal of the power capacity claims is with leave to amend. The plaintiff may either
amend complaint within 21 days or proceed on this complaint and seek leave to amend later if
discovery on the surviving claims reveals additional information relevant to the dismissed
claims. Either way, discovery may begin immediately. The case management conference
scheduled for January 10, 2025, is continued to January 17, 2025.
IT IS SO ORDERED.
Dated: January 6, 2025
______________________________________
VINCE CHHABRIA
United States District Judge

12