In re NVIDIA Corporation Securities Litigation
Pending before the Court are Defendants’ motion to exclude the damages-related opinions of Dr. Joseph R. Mason, Dkt. No. 257, and Plaintiffs’ motion for class certification, Dkt. No. 237. The Court DENIES Defendants’ motion to exclude and GRANTS Plaintiffs’ motion for class certification.
I. BACKGROUND
This is a consolidated securities class action brought by Plaintiffs Lannebo Kapitalförvaltning AB (“Lannebo”1) and Stichting Pensioenfonds PGB (“PGB”) (collectively, “Plaintiffs”) against Defendants NVIDIA Corporation (“NVIDIA”) and Jensen Huang, co-founder and Chief Executive Officer (collectively, “Defendants”).2 Plaintiffs allege violations of Sections 10(b) and 20(a) of the Securities Exchange Act of 1934 (the “Exchange Act”) and Rule 10b-5. Dkt. No. 149 (“FAC”) ¶¶ 174–75. Plaintiffs bring this securities action individually and “on
A. Factual Background
The Court summarizes the relevant allegations below. A more complete description of the allegations—particularly of the alleged misstatements—is included in the Court‘s prior order on Defendants’ motion to dismiss. See Dkt. No. 174 at 2–8.
a. NVIDIA
NVIDIA “is a multinational technology company” that produces graphics processing units (“GPUs”), processors that are used for rendering computer graphics. FAC ¶¶ 1, 38. NVIDIA‘s GPU business is divided into five market segments, two of which are most relevant to this case. Id. ¶ 39. The first is the Gaming segment, which includes chips designed for video games and is comprised primarily of the “GeForce” GPU product line. Id. ¶¶ 39–40. The second is the Original Equipment Manufacturer & IP (“OEM”) segment, which includes chips designed for devices such as tablets and phones. Id. ¶ 39. “In every quarter of the Class Period, Gaming revenues exceeded those of the four other segments combined.” Id. ¶ 40.
b. Crypto-Mining and GPUs
Generally, cryptocurrencies (or “crypto”) are digital tokens exchanged through peer-to-peer Internet transactions. Id. ¶¶ 44, 47. These transactions are verified on a “decentralized, immutable ledger,” through a process referred to as “crypto-mining,” or simply “mining.” Id. ¶¶ 45–47. Miners eventually turned to “GPUs, which could execute the computationally intensive work of crypto-mining hundreds of times faster” than CPUs in home computers. Id. ¶ 52. Due to the significant costs, crypto-mining is only profitable when prices for cryptocurrencies are above a certain level. Id. ¶ 54. Thus, “[b]ecause cryptocurrency prices have swung wildly over their short history, . . . the demand for mining hardware—including GPUs—has proven extremely volatile.” Id. ¶ 55.
In 2017, prices in the cryptocurrency market began to climb, creating new demand for GPU processing power. Id. ¶¶ 52, 60, 62. The Ethereum network and its Ether token, “[t]he most
In May 2017, NVIDIA launched a special GPU designed specifically for cryptocurrency mining (“Crypto SKU”). Id. ¶ 6. Revenues from Crypto SKU sales were reported in NVIDIA‘s OEM segment, not the Gaming segment. Id. Plaintiffs allege that “[l]aunching the Crypto SKU and reporting its sales in the OEM segment thus allowed Defendants to claim that any mining-related revenues were cordoned off in OEM, creating the impression that NVIDIA‘s crown jewel Gaming business was insulated from crypto-related volatility (and the crash in demand that would follow the cryptocurrency markets’ inevitable bust).” Id.
c. Summary of Alleged False and Misleading Statements
“Throughout the Class Period, NVIDIA reported skyrocketing revenues in its core Gaming segment.” Id. ¶ 63. Plaintiffs allege that “investors and analysts alike questioned whether those revenues truly derived from GeForce GPU sales to gamers or, rather, were from sales of GeForce GPUs to cryptocurrency miners, whose demand was at risk of disappearing if the economics of mining turned negative.” Id. ¶ 64. Plaintiffs allege that three categories of representations in Defendants’ public statements were materially false and misleading “and concealed from investors the enormous risk to NVIDIA‘s financial results posed by the Company‘s outsized exposure to crypto-mining”:
First, Defendants represented to investors that revenues from sales of its products to cryptocurrency miners were insignificant overall. Second, Defendants asserted that NVIDIA‘s soaring Gaming revenues indeed resulted from sales “for gaming”—not cryptocurrency mining. And third, Defendants represented that NVIDIA‘s cryptocurrency-related revenues were contained primarily in the Company‘s OEM reporting segment, when in fact, almost two-thirds of such revenue came from GeForce sales recorded in its Gaming segment.
Id. ¶ 62 (emphasis omitted).
Plaintiffs specifically identify eight related sets of misrepresentations:
On August 10, 2017, NVIDIA held its second-quarter fiscal year 2018 earnings call.3 FAC ¶ 179. A Goldman Sachs analyst asked, “So Q2 revenue came in roughly about $250 million above your guide. Can you confirm what some of the drivers were to the upside relative to your guidance? Was it all cryptocurrency or was it a combination of multiple things?” Dkt. No. 254-7, Ex. G at 8.4 5 Defendant Huang responded:
Id.; see also FAC ¶ 179.6[T]he $250 million, you could see in our – what we categorized under the OEM SKUs, basically the cryptocurrency SKUs. And that, if you reverse-engineered it out, I think, is approximately $150 million. And I – and we serve the vast – I would say, the large majority of the cryptocurrency demand out of that specialized product[]. There‘re still small miners that buy GeForces here and there, and that probably also increased the demand of GeForces.
- On August 12, 2017, VentureBeat published an article that included a transcript of an
interview with Defendant Huang. FAC ¶ 183. The interviewer asked if Defendant Huang “sa[id] a hallelujah for cryptocurrency.” Id. Huang responded: “No? Cryptocurrency is around. But it represented only a couple hundred million dollars, maybe $150 million or so. [Crypto mining] comes and goes. . . . We‘re not opposed to it. But our core business is elsewhere.” Dkt. No. 254-8, Ex. H at 4; see also FAC ¶ 183. - On August 23, 2017, Defendant NVIDIA filed its Q2 FY18 Form 10-Q, which stated that “GPU business revenue increased by 59%” year-over-year, which “was due primarily to increased revenue from sales of GeForce GPU products for gaming.” Dkt. No. 254-9, Ex. I at 12; see also FAC ¶ 187.
- On November 10, 2017, VentureBeat published another article in which Defendant Huang reportedly stated that cryptocurrency “is small but not zero. For us it is small because our overall GPU business is so large.” Dkt. No. 254-12, Ex. L at 4; see also FAC ¶ 196. Defendant Huang also said that “[crypto is] going to remain small for us. Maybe $70 million. It‘s large for somebody else. But it is small for us.” Ex. L at 5; see also FAC ¶ 196.
- On November 21, 2017, Defendant NVIDIA filed its Q3 FY18 Form 10-Q which stated that “GPU business revenue increased by 31%” year-over-year, which “was due primarily to increased revenue from sales of GeForce GPU products for gaming.” Dkt. No. 254-13, Ex. M at 12; see also FAC ¶ 200.
- On February 9, 2018, in an article published by Barron‘s, Defendant Huang was reported to have said that “crypto was a real part of our business this past quarter, even though small, overall.” Dkt. No. 254-19, Ex. S at 2; see also FAC ¶ 207.
- On March 26, 2018, in an article published by TechCrunch, Defendant Huang was reported to have said that “he still attributes crypto‘s demands as a small percentage of [NVIDIA‘s] overall business.” Dkt. No. 254-21, Ex. U at 5; see also FAC ¶ 210.
- On March 29, 2018, Defendant Huang appeared on the CNBC show Mad Money. FAC ¶ 213. When asked about the growth of cryptocurrency risks, Defendant Huang stated:
[O]ur core growth drivers come from video games. It comes from professional graphics visualization . . . [and] from our data center business, which is now a multi-billion dollar business doubling each year, as well as in several years our autonomous vehicle business. So, those are our primary growth drivers. . . . Cryptocurrency just gave it that extra bit of juice that caused all of our GPUs to be in such great demand.
Dkt. No. 254-22, Ex. V at 4; see also FAC ¶ 213. He also agreed with the host that “if people think [crypto] is that important, then they‘re going to miss the bigger picture.” Ex. V at 4.
In addition, Plaintiffs allege that Defendant Huang‘s answers during the Q2 FY19 earnings call on August 16, 2018, were materially false and misleading. FAC ¶ 216. When asked about excess inventory, Huang responded, “We‘re expecting the channel inventory to work itself out. We are the masters at managing our channel, and we understand the channel very well. . . . [W]e have plenty of opportunities as the – as we go back to the back-to-school and the gaming cycle to manage the inventory, so we feel pretty good about that.” Dkt. No. 254-31, Ex. EE at 12; see also FAC ¶ 216. Plaintiffs allege that these statements were materially false and misleading because (1) “throughout the Class Period, the overwhelming majority of NVIDIA‘s cryptocurrency-related revenues . . . [were] made through the Gaming segment, not through the OEM segment‘s Crypto SKU”; and (2) “[NVIDIA] had a massive glut of unsold GeForce GPUs that NVIDIA had amassed to satisfy the anticipated demand, which no longer existed, from crypto-miners.” FAC ¶ 217.
d. Corrective Disclosures
“In the spring of 2018, the cryptocurrency markets started to weaken,” and “[t]he façade of the Gaming segment‘s invulnerable growth began to crumble.” Id. ¶ 16. Plaintiffs claim that Defendants were forced to partially reveal the truth on August 16, 2018, when they lowered their revenue guidance and acknowledged that “a great deal” of cryptocurrency miners had bought GeForce Gaming GPUs and that GeForce inventories “had ballooned more than 36% . . . , reflecting the glut of supply that followed the end of crypto-related demand.” Id. ¶¶ 17, 155; Dkt. No. 254-30, Ex. DD at 18; Ex. EE at 8, 10, 13. Defendants also stated that they were “now projecting no [cryptocurrency] contributions going forward.” FAC ¶ 155; Ex. EE at 5, 8, 13, 16.
Plaintiffs then claim that “it became fully apparent to the market [on November 15, 2018] that, contrary to Defendants’ earlier representations, NVIDIA‘s revenues were unduly dependent on cryptocurrency mining,” as NVIDIA cut its revenue guidance for the fourth quarter, “[a]ttributing the reversal to a ‘sharp falloff in crypto demand.’” Id. ¶ 18; see also Dkt. No. 254-39, Ex. MM at 17, 19; Dkt. No. 254-40, Ex. NN at 5. CFO Kress stated that “Gaming was short of expectations as post crypto channel inventory took longer than expected to sell through,” and “Gaming card prices, which were elevated following the sharp crypto falloff, took longer than expected to normalize.” FAC ¶ 161 (emphasis omitted); Ex. NN at 5. Kress also stated that gross margin results were “below our outlook due to the $57 million charge for prior architecture components and chips following the sharp falloff in crypto demand.” Ex. NN at 7; FAC ¶ 161.
Plaintiffs claim that this also revealed that “Defendants’ purported ‘mastery of the channel’ had been a fiction; to the contrary, they had flooded the channel with GeForce inventory to meet crypto-miners’ demand just as that demand began its inevitable decline.” FAC ¶ 162. Following these alleged disclosures, NVIDIA stock price “plummeted 28.5% over two trading sessions.” Id. ¶ 171.
B. Procedural Background
In March 2021, the Court dismissed Plaintiffs’ complaint for a second time for failure to adequately plead scienter. Dkt. No. 174 at 22. The Ninth Circuit affirmed with respect to Defendants Kress and Fisher, but reversed with respect to Defendants Huang and NVIDIA. E. Ohman J:or Fonder AB v. NVIDIA Corp., 81 F.4th 918, 947 (9th Cir. 2023). Plaintiffs moved for certification of a 23(b)(3) class in July 2025. See generally Mot.
II. MOTION TO EXCLUDE
Pending before the Court is Defendants’ motion to exclude the damages-related opinions of Dr. Joseph R. Mason. Dkt. No. 257 (“Mot.”); Dkt. No. 263 (“Opp.”); Dkt. No. 264 (“Reply”). The Court DENIES the motion.
A. Legal Standard
(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;
(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‘s opinion reflects a reliable application of the principles and methods to the facts of the case.
B. Discussion
Plaintiffs’ damages expert, Dr. Mason, proposes using “a well-established and widely accepted approach based on ‘out-of-pocket’ losses” to calculate class-wide damages, which takes “the difference between: (a) price inflation at the time the securities were purchased (or acquired); and, if necessary, (b) price inflation at the time the securities were sold after one or more curative events.” Dkt. No. 237-2 (“Mason Rpt.”) ¶¶ 99, 106. Dr. Mason claims he can “measure the impact of a curative event on NVIDIA‘s common stock price using the abnormal returns, if any, that followed such disclosure, as determined through an event study approach.” Id. ¶ 102. “If information unrelated to the alleged fraud (i.e., ‘confounding information’) was also released at or about the same time as the curative events,” then Dr. Mason “would evaluate whether such confounding information is responsible for all or some of the abnormal stock price return.” Id. ¶ 103. To do so, Dr. Mason would “isolate the portion of such price decline caused by the
Defendants argue that Dr. Mason‘s damages-related opinions, id. ¶¶ 11, 97–108, should be excluded because (1) he “does not account for factors outside of NVIDIA‘s control that he admits would have contributed to the alleged artificial inflation of NVIDIA‘s stock price”; (2) he “does not explain how he would disaggregate damages attributable to the alleged misstatements that survived the Ninth Circuit appeal from those that did not”; and (3) his event study is not grounded in Plaintiffs’ theory of liability and “fails to account for the portion of the drop in share price attributable to the materialization of risks the market already knew existed.” Mot. at 6–7.
“Plaintiffs may rely on an unexecuted damages model to demonstrate that damages are susceptible to common proof.” Lytle v. Nutramax Lab‘ys, Inc., 114 F.4th 1011, 1032 (9th Cir. 2024). “[W]here an expert‘s model has yet to be fully developed, a district court is limited at class certification to making a predictive judgment about how likely it is the expert‘s analysis will eventually bear fruit. This still requires determining whether the expert‘s methodology is reliable, so that a limited Daubert analysis may be necessary, but the more full-blown Daubert assessment of the results of the application of the model would be premature.” Id. at 1031. “The fact that a model is underdeveloped may weigh against a finding that it will provide a reliable form of proof. Merely gesturing at a model or describing a general method will not suffice to meet this standard. Rather, plaintiffs—or their expert—must chart out a path to obtain all necessary data and demonstrate that the proposed method will be viable as applied to the facts of a given case.” Id. at 1032.
Plaintiffs have met this burden. This evidence is clearly relevant to the
“[H]ere there is no dispute that [Dr. Mason‘s] analysis is capable of measuring class[-]wide damages, at least in the abstract, and the only real question at this stage is whether Dr. [Mason] will properly apply the method to the facts of the case.” Lytle, 114 F.4th at 1033. Dr. Mason has explained that he will use “standard financial analysis and valuation tools” to remove confounding price impact and will use the facts of the case to address differential inflation over the Class Period. Mason Rpt. ¶¶ 103, 105; see also Dkt. No. 263-8, Ex. G at 50:21–51:25 (stating in deposition that “standard financial tools” like “discounted cash flow techniques, potential comparables techniques, [and] multiples techniques” could be used to “disaggregate the impact of confounding information”). The lack of specificity here may undercut the weight of the testimony somewhat, but it is sufficient to “chart out a path to obtain all necessary data and demonstrate that the proposed method will be viable.” Lytle, 114 F.4th at 1032.7 The Court will consider Defendants’ challenges when evaluating the weight of this evidence in the predominance inquiry, and Defendants may challenge Dr. Mason‘s executed model at a later stage.8 For now, the Court DENIES the motion.
III. MOTION FOR CLASS CERTIFICATION
Pending before the Court is Plaintiffs’ motion for class certification. Dkt. No. 287-2
A. Basic Legal Standard
If the four prerequisites of Rule 23(a) are met, a court also must find that the plaintiff “satisf[ies] through evidentiary proof” one of the three subsections of
“Rule 23 grants courts no license to engage in free-ranging merits inquiries at the certification stage.” Amgen Inc. v. Conn. Ret. Plans & Tr. Funds, 568 U.S. 455, 466 (2013). “Merits questions may be considered to the extent—but only to the extent—that they are relevant to determining whether the Rule 23 prerequisites for class certification are satisfied.” Id. Courts “must take the substantive allegations of the complaint as true” but “need not accept conclusory or generic allegations regarding the suitability of the litigation for resolution through class action.” Keilholtz v. Lennox Hearth Prods. Inc., 268 F.R.D. 330, 335 (N.D. Cal. 2010).
B. Statute of Repose
Defendants briefly argue that because the unnamed class members are not yet parties, “[w]ere the proposed class to be certified now, Defendants would face new claims from new parties regarding events that occurred over seven years ago, in plain violation of the [Exchange Act‘s] five-year statute of repose and the Rules Enabling Act.” Opp. at 30.
The Court disagrees. The complaint was filed well within the necessary five-year period. “[T]he filing of a timely class action complaint commences the action for all members of the class as subsequently determined.” Am. Pipe & Const. Co. v. Utah, 414 U.S. 538, 550 (1974) (“American Pipe”). Several courts have already rejected Defendants’ argument for this reason. See In re SunEdison, Inc. Sec. Litig., 329 F.R.D. 124, 145 (S.D.N.Y. 2019) (“Upon certification of the class, the claims of the class members are deemed to have been asserted on the date that the individual plaintiff commenced the action. That the timeliness of an absent class member‘s claim is measured from the date that the class representative filed suit, rather than, for example, the date of the Court‘s certification order, does no offense to the . . . repose period.”); In re NIO, Inc. Sec. Litig., No. 19-CV-1424 (NGG) (JRC), 2023 WL 5048615, at *3–*5 (E.D.N.Y. Aug. 8, 2023) (collecting cases).
Defendants’ cited case is distinguishable, as it barred an individual who had opted out of a securities class from subsequently filing a new individual action outside of the statute of repose‘s
C. Rule 23(a) Factors
The parties do not dispute that numerosity, commonality, typicality, and adequacy are satisfied, and the Court finds that these uncontested Rule 23(a) requirements are met. NVIDIA had at least 595 million outstanding shares and an average weekly trading volume of 66.1 million shares during the Class Period, which “support[s] a ready inference that thousands of geographically dispersed investors comprise the proposed Class.” Mot. at 17 (citing Mason Rpt. ¶¶ 39, 80); cf. Pardi v. Tricida, Inc., No. 21-CV-00076-HSG, 2024 WL 4336627, at *4 (N.D. Cal. Sept. 27, 2024). There is at least one common question in this case, including (1) whether Defendants made material misrepresentations or omissions to investors during the Class Period; (2) whether Defendants acted with scienter; and (3) whether Defendant Huang exercised control over NVIDIA. Mot. at 18. Plaintiffs are typical because they, like all class members, purportedly “purchased NVIDIA stock at artificially inflated prices during the Class Period.” Id. (citing Dkt. No. 42-3). Finally, Plaintiffs contend that they have and will work diligently to protect class members’ interests, there are no apparent conflicts of interest, and counsel has substantial experience in litigating securities class actions. See Dkt. No. 237-7 (Lannebo declaration); Dkt. No. 237-8 (PGB declaration); Dkt. No. 237-9 (KTMC resume); Dkt No. 237-10 (BLBG resume).
D. Superiority
The parties similarly do not dispute superiority. Courts examine four factors to determine
E. Predominance
The central dispute in this case concerns predominance. For violations of Section 10(b) and Rule 10b-5 (and thus Section 20(a)), a plaintiff must show: “(1) a material misrepresentation or omission by the defendant; (2) scienter; (3) a connection between the misrepresentation or omission and the purchase or sale of a security; (4) reliance upon the misrepresentation or omission; (5) economic loss; and (6) loss causation.” See Amgen, 568 U.S. at 460–61 (quotation omitted); see also Howard v. Everex Sys., Inc., 228 F.3d 1057, 1065 (9th Cir. 2000) (requiring a predicate violation of federal securities laws for Section 20(a) claim). Whether the predominance requirement of Rule 23(b)(3) is satisfied hinges on whether reliance can be resolved on a class-wide basis. Plaintiffs argue that they have shown predominance under either the fraud-on-the-market presumption discussed in Basic Inc. v. Levinson, 485 U.S. 224 (1988), or the presumption discussed in Affiliated Ute Citizens of Utah v. United States, 406 U.S. 128 (1972). Mot. at 21. The Court finds that Plaintiffs are entitled to the Basic presumption, which Defendants have failed to rebut.10
i. Legal Standard
The Basic fraud-on-the-market presumption arose “as a practical response to the difficulties of proving direct reliance in the context of modern securities markets, where impersonal trading rather than a face-to-face transaction is the norm.” SEB Inv. Mgmt. AB v. Symantec Corp., 335 F.R.D. 276, 283 (N.D. Cal. 2020). The presumption is based on the well-founded principle that “a public, material misrepresentation will distort the price of stock traded in an efficient market, and that anyone who purchases the stock at the market price may be considered to have done so in reliance on the misrepresentation.” Halliburton Co. v. Erica P. John Fund, Inc., 573 U.S. 258, 283–84 (2014) (“Halliburton II”). To invoke the Basic presumption of class-wide reliance at the class certification stage, a plaintiff must show that (1) the alleged misrepresentations were publicly known; (2) the stock traded in an efficient market; and (3) the plaintiff traded the stock between the time when the misrepresentations were made and when the truth was revealed. Goldman Sachs Grp., Inc. v. Arkansas Tchr. Ret. Sys., 594 U.S. 113, 118–19 (2021).11
Once a plaintiff has established the Basic presumption, a defendant can rebut it by demonstrating that the misrepresentations did not actually affect the market price of the stock. Id. at 119. Defendants “bear the burden of persuasion to prove a lack of price impact by a preponderance of the evidence.” Id. at 117. “[I]t is Defendants’ burden to show the absence of price impact – not merely to challenge Plaintiff[s] on the persuasiveness of [their] own price impact claim – once Basic‘s presumption of reliance attaches.” In re Allergan PLC Sec. Litig., 2021 WL 4077942, at *10 (S.D.N.Y. Sept. 8, 2021) (quotation omitted) (emphasis in original). “The defendant must ‘in fact’ ‘sever the link’ between a misrepresentation and the price paid by the plaintiff—and a defendant‘s mere production of some evidence relevant to price impact would rarely accomplish that feat.” Goldman, 594 U.S. at 125 (cleaned up and quotation omitted)
In these circumstances, “[t]he district court‘s task is simply to assess all the evidence of price impact—direct and indirect—and determine whether it is more likely than not that the alleged misrepresentations had a price impact.” Id. at 126–27. Courts must “take into account all record evidence relevant to price impact, regardless [of] whether that evidence overlaps with materiality or any other merits issue,” and they should do so “aided by a good dose of common sense.” Id. at 122, 124 (quotation omitted) (emphasis in original). The Supreme Court also explained that “the district court must use the evidence to decide the price impact issue while resisting the temptation to draw what may be obvious inferences for the closely related issues that must be left for the merits.” Id. at 122 n.2 (quotation omitted).13 In short, Defendants have the burden of showing that “alleged misrepresentation[s] did not actually affect the market price of the stock,” with “the district court [serving] as the fact finder.” Jaeger v. Zillow Grp., Inc., No. 24-6605, 2025 WL 2741642, at *1 (9th Cir. Sept. 26, 2025) (quotation omitted).14
ii. Applying Goldman
Plaintiffs argue that they have established the Basic presumption and introduce evidence showing that (1) the alleged misrepresentations were made publicly; (2) Plaintiffs and other class members purchased NVIDIA stock after the misrepresentations but before the November 2018 disclosure; and (3) NVIDIA common stock traded in an efficient market. See Mot. at 21 (citing Mason Report ¶¶ 10, 95–96; and FAC ¶¶ 179, 183, 187, 196, 200, 207, 210, 213, 216); see also Mot. at 7 (limiting class to investors who purchased after first misrepresentation and before corrective disclosure), 22–26 (assessing factors supporting market efficiency). Defendants do not dispute Plaintiffs’ entitlement to the presumption. The Court finds that Plaintiffs have put forth sufficient evidence to trigger the presumption, such that the burden shifts to Defendants to rebut it.
Here, Defendants argue that “(1) none of the challenged statements increased NVIDIA‘s stock price on the ‘front end,’ and (2) the alleged August and November 2018 stock drops (the ‘back end’) are not evidence that the challenged statements impacted the stock price.” Opp. at 17. As discussed in detail below, the Court finds that, “it is more likely than not that the alleged misrepresentations had [at least some] price impact.” Goldman, 594 U.S. at 127.
a. Front-End Price Impacts
Defendants first argue that there was no front-end price impact “because the alleged misstatements did not increase the stock price.” Opp. at 17–18. They rely on an expert report from Dr. Paul Zurek, who used an event study model, analyst reports, and prior publicly available information to analyze NVIDIA‘s stock price movements during the Class Period. Dkt. No. 261-1
Dr. Zurek also found that the statistically significant price increases following the remaining two misstatements were not the result of the alleged misstatements. Id. ¶¶ 61, 73. The first alleged misstatement was Defendant Huang‘s observation in the August 12, 2017 VentureBeat article that cryptocurrency “represented only a couple hundred million dollars, maybe $150 million or so” and “our core business is elsewhere. Id. ¶ 57 (emphasis omitted); see FAC ¶ 183; Ex. H at 4. Dr. Zurek concluded that this information was already publicly available, since Defendants had stated on August 10, 2017, that OEM cryptocurrency revenue was approximately $150 million in Q2 FY18, NVIDIA served some portion of cryptocurrency demand with GeForce GPUs, and cryptocurrency revenue was less than 10% of NVIDIA‘s quarterly revenue. Zurek Rpt. ¶ 60; see Ex. G at 4, 8. NVIDIA‘s stock price apparently went down following that August 10 disclosure. Zurek Rpt. ¶ 55. In addition, Dr. Zurek found that none of the analyst reports published in the subsequent seven calendar days attributed a positive reaction in NVIDIA‘s stock price to the alleged misrepresentation or mentioned the VentureBeat article at all. Id. ¶ 59. Based on this, Dr. Zurek concluded that no evidence supports the claim that NVIDIA‘s stock price increased because of the misrepresentation. Id. ¶ 61.
The second alleged misstatement was Defendant Huang‘s claim in the November 10, 2017 VentureBeat article that cryptocurrency revenue “is small but not zero” and amounted to “[m]aybe $70 million. It‘s large for somebody else. But it is small for us.” Id. ¶ 69 (emphasis omitted); FAC ¶ 196; Ex. L at 4–5. Dr. Zurek found that this information also was already publicly available, since Kress stated the day before that NVIDIA‘s “specific crypto [boards] equated to about $70 million of revenue,” and Defendant Huang stated that “crypto will be a small but not [zero] part of [NVIDIA‘s] business.” Zurek Rpt. ¶ 72; see Dkt. No. 254-10, Ex. J at 5, 12 (noting
Plaintiffs argue that the statistically significant stock price increases following the August 12 and November 10 misrepresentations are evidence of front-end price impact. Reply at 16. But Plaintiffs do not adequately rebut Dr. Zurek‘s persuasive explanation that (1) the alleged misstatements on these two days reiterated existing public information; and (2) analysts did not mention the August 12 article at all in the week following its publication.15 Cf. In re Apache Corp. Sec. Litig., No. 4:21-CV-00575, 2024 WL 532315, at *6–*7 (S.D. Tex. Feb. 9, 2024) (finding no evidence of front-end impact despite a statistically significant increase after one statement where “all sources attributed the price increase to an EBITDAX and production beat, not to the alleged misrepresentation”); IBEW Loc. 98 Pension Fund v. Best Buy Co., 818 F.3d 775, 782 (8th Cir. 2016) (considering evidence that an alleged misstatement had already been made public two hours earlier when assessing price impact).
Plaintiffs and their expert, Dr. Joseph Mason, argue that Defendants “do not offer any explanation for these price increases or attribute them to something other than Huang‘s statements.” Reply at 16; see also Dkt. No. 270-3 (“Mason Reply”) ¶ 36. Such evidence could have strengthened Zurek‘s conclusion, but Goldman does not narrowly proscribe what evidence is probative, and Defendants are not required to provide an alternative theory in order to discredit price impact. Dr. Mason also argues that “it is premature to conclude that the market‘s reaction to the August 10 [misstatement] would have been fully reflected within a single trading day” and by
However, the persuasive value of Defendants’ evidence is limited. First, while event studies can be used to assess price impact, In re Waste Mgmt. Sec. Litig., 775 F. Supp. 3d 742, 763 (S.D.N.Y. 2025), “the lack of price movement on the dates of the alleged misrepresentations does not [fully] rebut the Basic presumption,” Waggoner v. Barclays PLC, 875 F.3d 79, 104 (2d Cir. 2017); see also Carpenters Pension Tr. Fund of St. Louis v. Barclays PLC, 310 F.R.D. 69, 95 (S.D.N.Y. 2015) (“The failure of an event study to find price movement does not prove lack of price impact with scientific certainty.”).
This is partially because the lack of a statistically significant price increase does not eliminate “the possibility that the Alleged Misstatements prevented NVIDIA‘s share price from declining had the allegedly concealed truth been revealed.” Mason Reply ¶ 15. According to Dr. Mason, “Plaintiffs allege that Defendants[‘] Alleged Misstatements continued to cause NVIDIA‘s common stock to trade at artificially inflated prices until the truth was revealed through the Corrective Disclosures.” Id. Outside of Dr. Mason‘s reports, Plaintiffs do not clearly spell out whether they are pursuing this kind of “inflation-maintenance” theory, as opposed to one where the misstatements caused the stock price to rise. See Zurek Rpt. ¶ 196 (noting the same). The Court understands Plaintiffs’ theory to be that NVIDIA‘s stock was doing well in 2017 (in part because of “flourish[ing]” Gaming revenue), and Defendants’ misstatements prevented the stock
As noted earlier, the fact that Plaintiffs are apparently pursuing this kind of “inflation maintenance” theory does not mean that this evidence shouldn‘t be considered, just that it cannot entirely rebut the Basic presumption. In addition, courts analyzing this kind of theory have suggested that Defendants need not refute front-end price impact and can rebut the Basic presumption by attacking back-end price impact. See Crews v. Rivian Auto., Inc., No. 2:22-CV-01524-JLS-E, 2024 WL 3447988, at *13 n.8 (C.D. Cal. July 17, 2024) (disagreeing that “Defendants failure to analyze [ ] price increases is fatal to their price impact challenge” and noting that a lack of back-end price impact would “be strong and persuasive evidence that there was no price impact at all“).
Next, Dr. Mason broadly critiques Dr. Zurek for “fail[ing] to address any confounding information on days when the Alleged Misstatements were made, and in particular, days when NVIDIA announced its earnings results and released SEC filings, which would have included financial performance results related to all of NVIDIA‘s business segments.” Mason Reply ¶ 3(b) (emphasis omitted); see also id. ¶¶ 20-21. For example, Dr. Mason claims that Dr. Zurek cannot
The Court agrees that this weakens some of Dr. Zurek‘s findings. See In re Petrobras Sec., 862 F.3d 250, 279 (2d Cir. 2017) (noting that “it can be extremely difficult to isolate the price impact of any one piece of information in the presence of confounding factors, such as other simultaneously released news about the company, the industry, or the geographic region“); see also In re Allstate Corp. Sec. Litig., 966 F.3d 595, 613 n.6 (7th Cir. 2020) (“Econometrics, finance, and securities law experts have criticized the methods used in event studies prepared for litigation, and they caution courts to think carefully about how such study designs and findings often do a poor job of answering the legal questions at stake.“). But this argument does not fully defeat Dr. Zurek‘s analysis, since Plaintiffs only identify specific examples of confounding information for a handful of the misrepresentations, see, e.g., Mason Reply ¶¶ 28, 70, and Dr. Zurek paired his event study with supplementary review of analyst reports.
Given this, the Court agrees that Defendants have not satisfied their burden simply by attacking front-end price impact. But the Court is also persuaded that this evidence is consistent with Defendants’ theory that there was no front-end price impact in NVIDIA‘s stock attributable to the alleged misrepresentations.
b. Back-End Price Impacts
The Court next considers the evidence regarding back-end price impact. Defendants argue that the stock price drops following the August and November 2018 disclosures are not evidence of any such impact. Opp. at 19. Goldman addressed price impact cases like this one where a misrepresentation allegedly maintained inflation that was already built into the stock price. 594 U.S. at 123. Courts measure this inflationary effect by looking to changes in the stock price after the truth is finally disclosed: if the stock price drops, this “back-end” price drop serves as indirect evidence of “front-end” inflation. Id. But Goldman explained that the “inference . . . that the back-end price drop equals front-end inflation . . . starts to break down when there is a mismatch
Avoiding a “mismatch” “requires a closer fit (even if not precise) between the front- and back-end statements than courts have required when analyzing the loss causation element of securities fraud.” Id. at *6 (quotation omitted). However, a corrective disclosure “need not precisely mirror the earlier misrepresentation.” In re BofI Holding, Inc. Sec. Litig., 977 F.3d 781, 790 (9th Cir. 2020) (quotation omitted). “It is enough if the disclosure reveals new facts that, taken as true, render some aspect of the defendant‘s prior statements false or misleading.” Id. Further, “the true facts concealed by the defendant‘s misstatements may be revealed over time through a series of partial disclosures.” Id. “An event can [also] be corrective when it merely discloses a consequence of concealed information (for example, an earnings miss), even if the connection between the cause and the consequence is not made explicitly in the disclosure.” Jaeger v. Zillow Grp., Inc., 746 F. Supp. 3d 1025, 1036 (W.D. Wash. 2024), aff‘d, No. 24-6605, 2025 WL 2741642 (9th Cir. Sept. 26, 2025).
“[R]evelations that are not ‘corrective’ cannot form the basis for a corrective disclosure.” In re FibroGen Sec. Litig., No. 21-CV-02623-EMC, 2024 WL 1064665, at *12 (N.D. Cal. Mar. 11, 2024). “A finding of ‘back-end’ price impact requires proof that the information disclosed . . . was (i) corrective of one or more prior false statements or omissions, (ii) new (unknown to the market prior to [the purported corrective disclosure date]), and (iii) ‘value-relevant’ (i.e., caused at least some of the stock price decline).” Id. Defendants may also show that a challenged statement ceased to impact stock price at a particular point in a putative class period. See Ferris v. Wynn Resorts Ltd., 2023 WL 2337364, at *11 (D. Nev. Mar. 1, 2023) (under the “mismatch” analysis, class “may not proceed” as to one purported corrective disclosure).
i. August 2018 Partial Disclosure
Plaintiffs argue that NVIDIA issued a partially corrective disclosure on August 16, 2018, in its Q2 FY19 earnings call and Form 8-K. Mot. at 12-13; FAC ¶ 155. On that date, (1) Defendant Huang stated that “‘probably . . . a great deal’ of cryptocurrency miners had bought NVIDIA‘s GeForce gaming cards“; (2) NVIDIA “lowered its revenue guidance by approximately 2.2% for 3Q18” and issued a revised earnings forecast that was significantly lower than expected and which “include[d] no contribution from crypto“; (3) CFO Kress stated that, while NVIDIA “had previously anticipated cryptocurrency to be meaningful for the year, we are now projecting no contributions going forward,” and “we have reached a normal period as we‘re looking forward to essentially no cryptocurrency as we go forward“; (4) Kress stated that “over the last several quarters, we have seen the impacts of crypto and what that can do to elevate our overall gross margins“; and (5) NVIDIA “revealed that inventories had ballooned more than 36% from $797 million in 2Q19, to $1.09 billion in 3Q19.” FAC ¶ 155 (emphasis omitted); Ex. DD (Form 8-K); Ex. EE (Earnings Call). Plaintiffs allege that, as a result of these disclosures, NVIDIA‘s stock price fell by 4.9%. FAC ¶ 156. Plaintiffs claim that these disclosures “corrected Defendants’ prior materially misleading misstatements and omissions . . . by demonstrating that cryptocurrency-related sales were in fact a substantial and significant driver of the Company‘s revenues and, specifically, its Gaming segment revenues.” Id. ¶ 158.19
The “Great Deal” Comment. Defendants first argue that Huang‘s “great deal” comment cannot have been corrective because it “did not disclose anything new about [the referenced quarters] because the market knew long before August 2018 that miners had made many purchases of Gaming GPUs during that crypto spike.” Opp. at 19. The “great deal” comment was made by Defendant Huang during the question-and-answer period of the Q2 FY19 earnings call:
[Morgan Stanley Analyst]: I wonder if you could talk about cryptocurrency now that the dust has settled. You guys have done a good job of kind of laying out exactly how much of the OEM business has been driven by that. But there‘s also been, I think, some sense of
some of the GeForce business was being drive by crypto. Can you - looking backwards, can you size that for us? And I guess, I‘m trying to understand the impact that crypto would have on the guidance for October, given that it seems like it was very small in the July quarter. Huang: Well, I think - I mean, the second question is easier to answer, and the reason - the first one is just - it‘s ambiguous and hard to predict anyway. It‘s hard to estimate no matter what. But the second question, the answer is we‘re expecting - we‘re projecting 0 basically. And for the first question, how much of GeForce could have been used for crypto, a lot of gamers at night, they could - while they‘re sleeping, they could do some mining. And so whether they buy it for mining or do they but it for gaming, it‘s kind of hard to say. And some miners were unable to buy our OEM products, and so they jumped onto the market to buy it from retail. And that probably happened a great deal as well. And that all happened in the last - the previous several quarters, probably starting from Q - late Q3, Q4, Q1 and very little last quarter, and we‘re projecting no crypto mining going forward.
See Ex. EE at 14 (emphasis added).
Defendants argue that this did not disclose any new information because disclosures in November 2017, February 2018, and May 2018 had already made clear that miners likely bought a great deal of GPUs in Q3 FY18, Q4 FY18, and Q1 FY19. Opp. at 19-21. First, during the Q3 FY18 earnings call in November 2017, Kress stated that “GPU sales also benefited from continued cryptocurrency mining. We met some of this demand with a dedicated board in our OEM business and a portion with GeForce GTX boards, though it‘s difficult to quantify.” Dkt. No. 254-10, Ex. J at 5. Second, during the February 2018 Q4 FY18 earnings call, Kress stated:
Strong demand in the cryptocurrency market exceeded our expectations. We met some of this demand with a dedicated board in our OEM business, and some was met with our gaming GPUs. This contributed to lower than historical channel inventory levels of our gaming GPUs throughout the quarter. While the overall contribution of cryptocurrency to our business remains difficult to quantify, we believe it was a higher percentage of revenue than the prior quarter. That said, our main focus remains on our core gaming market as cryptocurrency trends will likely remain volatile.
Dkt. No. 254-18, Ex. R at 5; see also Dkt. No. 254-20, Ex. T at 18 (“Strong growth across our Pascal-based GeForce gaming GPUs was driven by growth associated with GPU refreshes/upgrades, new gamers, new games, eSports, and cryptocurrency mining.“). Third, during the May 2018 Q1 FY19 earnings call, Defendant Huang stated that “[c]rypto miners bought a lot of our GPUs during the quarter, and it drove prices up. And I think that a lot of gamers
In addition, Dr. Zurek concluded that “[a]t least as early as August 10, 2017 (the start of the Proposed Class Period), analyst commentary included assessments that the contribution of cryptocurrency mining to Gaming revenue was substantial.” Zurek Rpt. ¶ 122 (citing 19 analyst examples between August 2017 and May 2018); see also Dkt. No. 254-3 (“Wood Rpt.“) ¶¶ 77-85 (summarizing pre-August 2018 statements, news articles, and analyst reports to determine that “by the time of the first alleged corrective disclosure on August 16, 2018, investors following NVIDIA likely would have understood that cryptocurrency-related demand accounted for a substantial portion of NVIDIA‘s Gaming revenue“).21 Defendants state that, as a result of these prior disclosures, “none of the 57 analyst reports covering NVIDIA in August 2018 expressed any
This is a close call, but the Court finds that the “great deal” comment most likely did not disclose any information that investors did not already know by May 2018. The Court begins with the May 2018 disclosure, which is the most persuasive evidence here. The Court sees no convincing difference in magnitude between the May 2018 statement that miners “bought a lot of our GPUs,” Ex. W at 9, and the August 2018 statement that “miners . . . jumped onto the market to buy [GPUs] from retail,” and “that probably happened a great deal,” Ex EE at 13. Plaintiffs claim that “a lot” is a vague statement that did not convey the extent to which gaming revenue had been driven by mining, Reply at 12, but the same is true for the “great deal” statement.22
Plaintiffs argued at the motions hearing that even if the May 2018 disclosure was partially corrective of prior misrepresentations, that statement was only referring to Q1 FY19 and did not correct misrepresentations about gaming revenue in Q3 FY18 and Q4 FY18. Tr. 6:6-15, 13:16-21. It‘s true that the May 2018 disclosure explicitly referred to GPU sales “during the quarter,” Ex. W at 9, while the August 2018 disclosure referred to “late Q3, Q4, Q1, and very little [of Q2],” Ex EE at 13.23
Defendants countered that the February 8, 2018 disclosures revealed the same information for Q3 FY18 and Q4 FY18. Tr. 14:24-16:16. In the February 2018 disclosure, Kress and NVIDIA stated that “[s]trong demand in the cryptocurrency market exceeded [their] expectations” and “contributed to lower than historical channel inventory levels” for Gaming GPUs, Ex. R at 5, and that “[s]trong growth across . . . Pascal-based GeForce gaming GPUs was driven by” cryptocurrency, among other things, Ex. T at 18. Kress also stated that “[w]e indicated how much [crypto] we had in OEM boards, and we also indicated that there was definitely some also in our GTX business. . . . It‘s up [in Q4] a bit from what we saw in Q3, and we do again expect probably going forward.” Ex. R at 11. The match here is not as exact, but the Court is persuaded that Defendants had already disclosed strong demand for cryptocurrency in Q4 FY18 that was “definitely” drawing from the GeForce stock in a significant way and was increasing from Q3 FY18.24 And the conclusion that this was not new information is further strengthened by the fact that analysts later did not mention the “great deal” comment at all.25 26
Plaintiffs claim that this information was nevertheless new because “Defendants
The Court disagrees. First, Plaintiffs note that the day after the February 8, 2018 statements regarding strong demand from cryptocurrency exceeding expectation, Barron‘s published the February 9 article where Defendant Huang was reported to have said that crypto was a “small, overall part of our business . . . despite substantial internal evidence to the contrary.” Reply at 11 (quoting FAC ¶ 207 and Dkt. No. 268-5); see Ex. S at 2 (“And crypto was a real part of our business this past quarter, even though small, overall.“). In that article, Huang reportedly clarified that “[c]rypto is still just a small portion because ‘data centers is growing so fast.‘” Ex. S. at 2 (emphasis omitted). Applying a “good dose of common sense” here, Goldman, 594 U.S. at 122 (quotation omitted), it‘s hard to imagine that this single, partially qualified comment published a day later obscured the substance of the February 8 statement. Cf. Arkansas Tchr. Ret. Sys. v. Goldman Sachs Grp., Inc., 77 F.4th 74, 94 (2d Cir. 2023) (”ATRS“) (“Case law does not suggest, however, that investors read one statement in conjunction with separately disseminated statements, at least where, as here, those statements do not obviously build off one and other.“).28
Second, Plaintiffs argue that Defendant Huang‘s May 2018 statement that miners “bought a lot of our GPUs during the quarter” was surrounded by Kress’ misrepresentation that NVIDIA was (1) “able to fulfill most of [the crypto demand] with crypto-specific GPUs, which are included in [NVIDIA‘s] OEM business, at $289 million“; and (2) “[a]s a result, [NVIDIA] could protect the
Finally, the Court briefly observes that the analyst evidence here is mostly unpersuasive. Plaintiffs argue that “[d]uring the Class Period, analysts offered mixed, and often conflicting, views of NVIDIA‘s mining exposure based on their interpretation of public market data,” and thus Zurek‘s evaluation of analyst estimates cannot “show that the allegedly concealed facts were publicly known.” Reply at 13. The Court agrees that the analyst revenue estimates are mixed, see Zurek Rpt., App‘x C, and Defendants cannot meet their burden of demonstrating a lack of any price impact by cherry-picking analyst reports, see Rivian, 2024 WL 3447988, at *13 (rejecting truth-on-the-market where analysts had differing expectations). But the Court finds Defendants’ unrefuted evidence that none of the analyst reports made note of Huang‘s “great deal” comment particularly significant. The absence of any reaction here tellingly contrasts with Plaintiffs’ contention that “analysts that provided estimates of crypto revenue during the Class Period were surprised by the magnitude of NVIDIA‘s true crypto dependence revealed in November 2018.” Reply at 13 (emphasis added). As a result, the Court finds that the analyst reports here slightly support an inference that the August 2018 “great deal” comment was not new information (and, if it was, was not value-relevant).34
[of] the pricing for [the] vehicle,” despite the fact that a price increase was inevitable. Id. at *12. As discussed, the Court is not persuaded that the February 2018 statements were surrounded by a sufficiently distracting comment, and it is not persuaded that the May 2018 statements were surrounded by a misrepresentation.
Forward-Looking Guidance. Defendants argue that guidance about “then-current expectations for a future period . . . did not say anything, let alone correct, revenues for past periods that were the subject of the challenged statements.” Opp. at 21 (emphasis in original); see also Zurek Rpt. ¶¶ 133-40. They also argue that NVIDIA‘s statements that they expected a “negligible contribution from crypto going forward was a new development based on new crypto prices that could not have been disclosed as of the dates of the alleged misstatements.” Opp. at 21.
There is a mismatch here between the subject matter of the alleged misrepresentations - namely, whether the magnitude of miner revenue in Gaming was large and if demand was mostly contained in the OEM segment - and the alleged disclosures - a lack of cryptocurrency revenue in the future earnings forecast. And in some circumstances, courts have found that future financial guidance did not correct prior challenged statements. See In re Concho Res., Inc., No. 4:21-CV-2473, 2025 WL 1040379, at *2, *16 (S.D. Tex. Apr. 7, 2025) (finding that 2019 financial results and guidance did not “actually correct” some challenged statements about 2017 and 2018 results, and finding correction with respect to other statements because the purported disclosure specifically mentioned the efficiency of a manufacturing mode that had previously been discussed); Plumbers & Steamfitters Loc. 60 Pension Tr. v. Meta Platforms, Inc., No. 4:22-CV-01470-YGR, 2024 WL 4251896, at *11 (N.D. Cal. Sept. 17, 2024) (finding that prediction of 2022 financial condition did not correct representations regarding past performance, and stating that “[i]t strains credulity to reverse engineer this forward-looking observation into a retrospective analysis,” particularly where experts were surprised by the predictions rather than any past performance).
Plaintiffs and their expert assert that “lower-than-expected Q3 FY19 guidance, coupled with the information that NVIDIA guided cryptocurrency sales estimates within its OEM segment to zero, that its gross margins were expected to be negatively impacted, that its balance sheet
Plaintiffs’ further claim that this guidance revealed that “crypto was in fact large enough to impact the quarterly earnings guidance.” Reply at 11. But even assuming all of Plaintiffs’ contentions are true, none of the alleged misstatements suggest that crypto couldn‘t have an impact on the quarterly earnings guidance - it was just a question of how much. For example, it‘s
None of this is to say that a statement can never be corrective just because it is forward looking or because it addresses conditions that weren‘t known at the time of the misstatement. Cf. Baker v. SeaWorld Ent., Inc., 423 F. Supp. 3d 878, 929-32 (S.D. Cal. 2019) (rejecting similar argument). “An event can be corrective when it merely discloses a consequence of concealed information (for example, an earnings miss), even if the connection between the cause and the consequence is not made explicitly in the disclosure.” Jaeger, 746 F. Supp. 3d at 1036. But the weight of the evidence as to the August 2018 disclosure suggests that the analysts were reacting to (and the probable cause of the adjusted Q3 FY19 forecast was) the rapid drop-off in cryptocurrency demand, not the past guidance about demand in the Gaming segment. See Ex. EE at 11 (“In the beginning of the year, we thought and we projected that crypto would be a larger contribution through the rest of the year, but at this time, we consider it to be immaterial for the second half.“).39
Gross Margins. Defendants similarly argue that the Q3 FY19 gross margin guidance was not corrective and was a materialization of a known risk. Opp. at 21 n.8 (citing Zurek Rpt. ¶¶ 129-32). Kress explained in the earnings call:
We have essentially reached, as we move into Q3, a normalization of
our gross margins. I believe over the last several quarters, we have seen the impacts of crypto and what that can do to elevate our overall gross margins. We believe we ha[ve] reached a normal period as we‘re looking forward to essentially no cryptocurrency as we move forward.
Ex EE at 10. As elsewhere, Kress directly tied this change in guidance to the drop-off in cryptocurrency demand. Nothing about this disclosure appears to correct information about the magnitude of the cryptocurrency revenue in the past. See also Zurek Rpt. ¶ 131 (suggesting analysts already knew that gross margins were impacted generally by cryptocurrency demand). Plaintiffs raise the same arguments that the Court has already considered. See, e.g., Mason Reply ¶ 124.40
Inventory. Defendants similarly argue that the disclosure about increased inventory “was based on information that did not exist when the last of the challenged statements was made in March 2018, five months earlier, and therefore could not have been corrective.” Opp. at 21. Defendants also argue that they had told the market they would increase inventory to prepare for the launch of their new Turing architecture and to increase supply for gamers. Id. at 22. Plaintiffs do not advance any new argument specific to the inventory statements. Of course, Defendants cannot fully demonstrate a lack of any price impact just by pointing to other disclosed explanations for the inventory increase, see Zurek Rpt. ¶ 145 (discussing Ex. EE at 10), and the Court does not agree that an inventory increase after a particular event like the crypto fall-off could never suggest that past statements about crypto demand for that inventory were false. But, as with the forward-looking guidance and gross margin statements, the Court agrees with Defendants that there is a mismatch between this corrective disclosure and the subject matter of the alleged misstatements. That mismatch is substantially strengthened by the fact that not a
Conclusion. The Court finds that Defendants have proven that the August 2018 statements do not establish back-end price impact. The Court is particularly persuaded that (1) the substance of the “great deal” comment that allegedly partially corrected the market‘s misapprehension about the magnitude of cryptocurrency demand outside of the OEM segment had already been disclosed by May 2018; (2) there was no statistically significant price impact when that information was revealed in February and May 2018; (3) analysts did not draw attention to the “great deal” comment or otherwise express surprise about the magnitude of cryptocurrency revenue after the August 2018 disclosure; and (4) the remaining August 2018 statements were not obviously corrective and are mismatched with the subject matter of the alleged misrepresentations.41 42
However, this is a narrow finding that Defendants have severed the link between the alleged misstatements and the stock price decrease on August 16, 2018, and it does not resolve the question of whether the full November 2018 disclosure can serve as back-end evidence of front-end price impact. Even if investors understood that a “great deal” or “a lot” of GeForce GPUs had been purchased by crypto-miners by August 2018, they may not have understood the true magnitude of those purchases. As a result, the Court turns to consider the November 2018 disclosure.
ii. November 2018 Disclosures
Plaintiffs also argue that NVIDIA issued a full corrective disclosure on November 15, 2018, in its Q3 FY19 earnings call and Form 8-K. Mot. at 14-15; FAC ¶¶ 160-62. On that day, Defendants announced “a nearly 2% revenue miss” and “that it was expecting revenues of only $2.7 billion in 4Q19, a 7% decline year-over-year.” FAC ¶ 160 (emphasis omitted); see Ex. MM at 16, 19. Kress stated that “Gaming was short of expectations as post crypto channel inventory took longer than expected to sell through,” and “Gaming card prices, which were elevated following the sharp crypto falloff, took longer than expected to normalize.” FAC ¶ 161 (emphasis omitted); Ex. NN at 5. Kress further stated that “[a]lthough the cryptocurrency wave has ended, the channel has taken longer than expected to normalize,” “channel prices and inventory levels remained higher than expected” on GeForce gaming cards, and gross margin results were “below our outlook . . . following the sharp falloff in crypto demand.” FAC ¶ 161; Ex. NN at 6-7.
Defendants also disclosed that “excess Gaming GPU inventory following the disappearance of crypto-miners . . . would persist for at least 12 weeks” and “amount to $600 million in lost revenue.” FAC ¶ 162. In their 8-K, Defendants stated that the “gaming revenue outlook for the fourth quarter of fiscal 2019 [was] impacted by the expected work-down of Pascal [GeForce] mid-range gaming card inventory in the channel, and a decline in gaming console reflecting seasonal build patterns [and] assumes no meaningful shipments of mid-range Pascal GPUs during the quarter.” Ex. MM at 19. During Q&A, Defendant Huang stated that NVIDIA “came into Q3 with excess channel inventory post the crypto hangover,” and “the crypto hangover lasted longer than we expected.” Ex. NN at 11-12; FAC ¶¶ 163, 165. As a result of this news, NVIDIA‘s stock fell 28.5% over two trading sessions. FAC ¶ 171.
Plaintiffs argue that this disclosure revealed “the degree to which [NVIDIA‘s] revenues during the Class Period had depended on sales to crypto miners,” and corrected Defendant Huang‘s August 16, 2018 statements about “NVIDIA‘s purported mastery of its sales channel.” Reply at 14 (quotation omitted). While the content of this disclosure is similar to the August 2018 disclosure, the Court concludes that Defendants have not carried their burden to show a lack of any price impact here, in significant part due to much clearer analyst reports that linked the
Correction of Reliance on Crypto. Defendants argue that the November 2018 statements did not correct the eight alleged misstatements regarding NVIDIA‘s dependence on cryptocurrency because the results “refer[red] to new information about time periods after the relevant challenged statements,” and “the stock dropped because of disappointing revenue guidance regarding future sales during a time of industry volatility and expected product launches.” Opp. at 22 (emphasis omitted); see Ex. NN at 5, 12-13 (noting that “gaming was short of expectations as post crypto channel inventory took longer than expected to sell through,” “prices, which were elevated following the sharp crypto falloff, took longer than expected to normalize,” and “demand was uncertain about when the price will be stabilized,” which “froze the market a little longer than [NVIDIA] expected“).
The parties dispute what analyst evidence reveals about the corrective nature of this disclosure. Defendants argue that “[a]lthough market analysts were disappointed by NVIDIA‘s Q4 guidance, they generally agreed that excess channel inventory arose from channel pricing dynamics that had manifested during the most recent quarter and characterized the issue as transitory.” Opp. at 23-24 (citing Zurek Rpt. ¶ 184; Dkt. No. 254-41, Ex. OO at 5; and Wood ¶¶ 24, 67-68, 71-73); cf. Apache Corp. v. Castaldi, 2024 WL 532315, at *10 (finding lack of price impact where it was more likely than not that concerns unrelated to misstatements were what moved the market). According to Defendants, to the extent that “[a] handful of analysts speculated about whether crypto might have contributed to Gaming revenues in prior quarters more than they had previously guessed,” “nothing in that hindsight speculation purported to ‘correct’ any of the challenged NVIDIA statements,” particularly where NVIDIA had previously disclosed the challenges of quantifying crypto demand in Gaming. Opp. at 24 (citing Zurek Rpt. ¶¶ 115-16, 158-59).
In contrast, Plaintiffs argue that “[a]nalysts immediately recognized that NVIDIA‘s November 2018 revelations contradicted Defendants’ Class Period assurances that crypto contributed only a small part of revenue and was not driving Gaming segment results.” Reply at 14 (citing Mason Reply ¶¶ 142-56). They emphasize that “this [analyst] commentary directly tied the disclosure to Defendants’ prior false statements minimizing crypto demand in 2017 and 2018,”
For example, one analyst report from Morgan Stanley stated that “[t]he implication of [NVIDIA‘s] commentary is that a larger portion of demand in late 2017/early 2018 was for crypto than they had initially indicated, and that an end to the crypto bubble caused a channel refill which overshot.” Mason Reply ¶ 142(d). Another report from Macquarie Research stated that “[NVIDIA‘s] weaker outlook is not a surprise, but the magnitude is. . . . Magnitude of the weakness suggests that Crypto benefit was much higher than previously thought, . . . and appears to have resulted in excess channel inventory.” Id. ¶ 142(c); see also id. ¶ 144(p) (VentureBeat interviewer stating that “it seemed, in the past, that [crypto] was described as a small part of revenue, and now it‘s something that can affect one or two quarters worth of inventory“); id. ¶ 144(g) (noting surprise at “the magnitude of the gaming miss driven by crypto currencies“). RBC revised its estimates in January 2019, positing that ” [NVIDIA] generated $1.95B in total revenue related to crypto/blockchain. This compares to company‘s statement that it generated ~$602M over the same time period.” Ex. 20 at 2.
On one hand, as with the August 2018 disclosures, it‘s not explicit what in these disclosures “reveal[ed] new facts that, taken as true, render[ed] some aspect of [Defendants‘] prior statements false or misleading.” In re Bofl Holding, 977 F.3d at 790. Investors already knew at least by August 2018 that a ” great deal” of Gaming GPUs had been purchased by crypto-miners during the previous four quarters. Investors also already knew that cryptocurrency demand had fallen off by the August 2018 earnings call. Defendants explained in the November 2018 disclosure that the failure to normalize inventory related to unexpectedly slow demand reaction times, not (for example) a higher-than-disclosed level of past reliance on cryptocurrency revenue. See also Opp. at 14 (summarizing comments from Kress and Huang that “channel participants appeared to hold out hope for another crypto boom and buyers waited for the next-generation GPUs or for prices to drop“); Ex. NN at 6, 12-13. In this way, there is some mismatch in subject
On the other hand, it‘s notable that analyst reports directly linked the disappointing November 2018 revenue results and guidance back to prior statements about cryptocurrency‘s impact on revenue. Generally, the Court does not find a dissection of subjective analyst reporting to be dispositive - or sometimes even compelling - evidence regarding price impact. “[M]arket commentary can provide insight into the kind of information investors would rely upon in making investment decisions - and therefore can serve as indirect evidence of price impact - [but] commentary touching upon only the same subject matter . . . cannot be enough.” ATRS, 77 F.4th at 104. Conversely, Defendants cannot meet their burden merely by pointing to some analyst reports that support their views, and analyst reports that directly tie the disclosure to the misstatements are important evidence. See, e.g., Sjunde AP-Fonden v. Goldman Sachs Grp., Inc., 798 F. Supp. 3d 416, 477 (S.D.N.Y. 2025) (finding defendants’ evidence insufficient, in part given the contrary commentators who “did explicitly tie the stock price drop . . . to the corrective disclosure“).
Given analyst reports that specifically related the disclosure to the alleged misstatements, Defendants have not convinced the Court that it is likely that there was no price impact. Compare Ferris, 2023 WL 2337364, at *10 (finding an inference of price impact where “market analysts linked the prior allegations to the new revelations“), with In re Kirkland, 2024 WL 1342800, at *9 (“The record shows that no analysts referenced or discussed either of the M&A Statements at all - let alone drew the inference that Kirkland was not considering acquisitions.“). Analysts from Macquarie Research, Morgan Stanley, and VentureBeat apparently drew the exact inference in November 2018 that Plaintiffs posit - slower-than-expected clearance of excess inventory revealed that Gaming revenue had been more impacted by mining demand than they thought - even if it wasn‘t explicit in the disclosures. Cf. Jaeger, 746 F. Supp. 3d at 1036 (“An event can [also] be corrective when it merely discloses a consequence of concealed information (for example, an earnings miss), even if the connection between the cause and the consequence is not
Defendants argue that Morgan Stanley later published a January 2019 report stating that “we don‘t think that we underestimated the impact [of cryptocurrency].” Sur-Reply at 6 (quoting Dkt. No. 279-12, Ex. JJJ at 3). However, that report also states that “[w]e can‘t just blame cryptocurrency” and “we don‘t think [cryptocurrency is] the majority of the incremental weakness that we are seeing here.” Ex. JJJ at 3 (emphasis added). While Morgan Stanley‘s analyst apparently recognized that some portion of NVIDIA‘s inventory excess was driven by other factors (including support for the Turing architecture taking longer than expected), id. at 3–4, this report did not fully walk back the prior comments. More importantly, even assuming Morgan Stanely‘s November report incorrectly linked the corrective disclosure to the alleged misstatements and “the majority of analysts drew no connection between NVIDIA‘s disclosures in November 2018 and historical crypto contribution to Gaming,” Sur-Reply at 6, the Court still must grapple with the handful of reports that did make that connection.43 It is Defendants’ burden to prove the absence of any price impact. In light of these reports, Defendants have not met that burden, even if the evidence could support an inference that much or even most of the price impact came from other causes.
The Court also notes that internal company emails support its conclusion here. Just before the November 2018 disclosure, NVIDIA‘s then-VP of Investor Relations and Strategic Finance opined in response to a question from Huang that one reason “the market isn‘t pricing in a bigger miss” following news that AMD had one or two quarters of post-crypto channel inventory was in part “because of comments we‘ve made on . . . ring-fencing the crypto impact in OEM.” Dkt. No.
Correction of Channel Management Comments.
Defendants also argue that the disclosure did not correct Huang‘s statements that Defendants were “masters at managing [their] channel,” “understand the channel very well,” and “have plenty of opportunities . . . to manage the inventory” because “[t]he channel developments and pricing dynamics disclosed in November 2018 had not happened (and thus could not have been disclosed) at the time of the August 2018 statement.” Opp. at 24. They argue that the disclosure was the “materialization of a known risk about inventory dynamics.” Id.44
It is undisputed that Defendants’ excess inventories were disclosed in their Form 8-K in August 2018. See FAC ¶ 216 (noting that “NVIDIA‘s Form 8-K filed the same day [as the earnings call] disclosed that the Company had seen its inventory balloon by 37% the previous quarter, and several analysts asked questions about the glut during the call“); Ex. DD at 18 (Form 8-K); see also Zurek Rpt. ¶ 165. Analysts were apparently aware that there was some risk that this inventory could spill over into future quarters. Zurek Rpt. ¶ 180. Defendants also “regularly cautioned investors that [they] might ‘build inventory anticipating demand for a product that does not materialize.‘” Opp. at 25 (quoting Ex. T at 15); Zurek Rpt. ¶ 179. They argue that the fact that NVIDIA “was ultimately unable to predict how market participants would react” does not
Plaintiffs suggest that the channel management comments persuaded the public not to be concerned about the inventory glut, and that the revelations about the failure to offload the inventory revealed that Defendants’ confidence was misrepresented. See Mot. at 13; Reply at 14–15; FAC ¶ 217. Plaintiffs point to (1) an analyst report that “[t]he large shortfall in guidance due to a bloated channel due to crypto-currency is in sharp contrast to the comments around channel inventory from the company at the last earnings call,” Mason Reply ¶ 154(a) (emphasis omitted); (2) internal June 2018 comments that [REDACTED]; and (3) internal attribution of the later stock price inflation to the channel management comments. Reply at 15. Plaintiffs also suggest that August 2018 analyst reports stating that “[NVIDIA] had de-risked its portfolio from crypto” are evidence that the August 2018 comments about channel management had misled investors. Mason Reply ¶¶ 153–54 (emphasis omitted). It‘s not clear if most of the de-risking comments referred to NVIDIA‘s assurances about channel management, or the fact that cryptocurrency would no longer be a (volatile) part of GPU demand moving forward, but at least one analyst appears to have specifically credited the channel management assurances when stating that “inventories [were] not out of line.” See id. ¶ 153(f).
First, it‘s unclear if the June 2018 internal presentation citing [REDACTED] contradicts the claim that NVIDIA was a master at clearing its channel. Dkt. No. 270-10, Ex. 31 at 4–5, 19–20; see also Dkt. No. 270-11, Ex. 32 at 12 (August 2018 presentation [REDACTED]). For one thing, the Court cannot tell if this was a reference to [REDACTED] or [REDACTED], the latter of which would be consistent with Defendants’ forecast of no mining impact in Q3 FY19. And, as Defendants note, “the internal slide deck [Plaintiffs] cite for this proposition actually projected [about a third of the backlogged weeks of] inventory to sell through by the end of the quarter,” suggesting Defendants thought their channel management statements were accurate until shortly before the November disclosure. Sur-Reply at 6, 7 n.2 (citing Ex. 32 at 19 and additional estimates throughout Q3 FY19).
Given this, the evidence here supports the theory that the November 2018 disclosure corrected the channel management comments. Defendants have not severed the relationship between any of the misstatements and the undisputed back-end price impact following the November 2018 disclosure. Because the Court concludes that adequate evidence supports Plaintiffs’ contentions that the back-end price impact from the November 2018 disclosure is attributable to the alleged misstatements about both NVIDIA‘s reliance on crypto and its channel
iii. Other Individualized Questions of Reliance
Defendants also attempt to rebut the Basic presumption by showing that individual class members would have traded even knowing the allegedly concealed truth, and they argue that “[r]esolving these individual reliance issues for a substantial portion of the class would overwhelm common issues.” Opp. at 26. The Supreme Court has suggested that a defendant could rebut the presumption by showing that “an individual plaintiff traded or would have traded despite his knowing the statement was false.” Basic, 485 U.S. at 248; see also Halliburton II, 573 U.S. at 269 (“[I]f a defendant could show . . . that a plaintiff would have bought or sold the stock even if he had been aware that the stock‘s price was tainted by fraud, then the presumption of reliance would not apply.“). In some cases, courts have found that individual investors were “indifferent to the fraud” alleged and granted judgment for those investors because the Basic presumption had been rebutted. See In re Vivendi Universal, S.A. Sec. Litig., 183 F. Supp. 3d 458, 466 (S.D.N.Y. 2016) (”Vivendi“) (summary judgment); GAMCO Invs., Inc. v. Vivendi Universal, S.A., 838 F.3d 214, 221 (2d Cir. 2016) (”GAMCO“) (judgment after bench trial).
Defendants introduce evidence that “scores of putative class members ‘purchased NVIDIA common stock as part of a long-term AI-focused strategy, and thus for reasons unrelated to . . . the allegedly concealed information.‘” Opp. at 27 (quoting Wood Rpt. ¶¶ 24, 65–76). First, the general fact that some investors purchased a stock as part of a long-term strategy does not mean that the investors would have traded had they known about the claimed fraud. Cases like Vivendi came at summary judgment or after trial, when there was substantial evidence regarding particular investors’ strategies. See Vivendi, 183 F. Supp. 3d at 466–68 (discussing “indifferent value investor” who had regular meetings with defendant‘s senior management and for whom “the market price . . . was not important“); GAMCO, 838 F.3d at 221 (discussing the “extensive evidence” and noting that “on this record, . . . in this case, and with regard to this particular fraud, GAMCO would still have viewed Vivendi‘s securities as a profitable investment” (emphasis in original)). Second, these decisions notably did not decertify the class, instead discussing the “occasional class members that cannot survive an individualized rebuttal outside of the class
Ultimately, Defendants ask the Court to infer that individual inquiries must be conducted for each investor by (1) picking off a handful of investors—like Wood, Wood Rpt. ¶ 65—that might still have traded if they‘d known about the fraud; and (2) identifying alternate reasons why some unidentified number of investors would buy or hold the stock regardless of short-term price fluctuations. That is insufficient. Compare Van v. LLR, Inc., 61 F.4th 1053, 1069 (9th Cir. 2023) (finding predominance potentially defeated where record evidence indicated that nearly 20% of the transactions at issue would require individualized inquiries), with Halliburton II, 573 U.S. at 276 (“That the defendant might attempt to pick off the occasional class member here or there through individualized rebuttal does not cause individual questions to predominate.“).
Defendants’ novel theory would represent a seismic change in the law: it would allow a defendant to rebut reliance in almost every large securities fraud class action just by explicating some long-term investment strategies relating to the relevant stock. Defendants like NVIDIA would be almost completely insulated from such class actions under this theory. The Court rejects this reading of Basic. While this case may involve some individualized inquiries down the line, the Court cannot conclude from the record that they predominate over the common issues.
b. Damages
Finally, Defendants argue that Plaintiffs have not shown that damages can be calculated by class-wide evidence consistent with their theory of liability. Opp. at 29. Defendants largely repeat their argument that Dr. Mason‘s “boilerplate” out-of-pocket methodology cannot withstand the rigorous analysis required under Comcast Corp. v. Behrend, 569 U.S. 27 (2013), because Dr. Mason fails to explain how he would disaggregate variable daily inflation, inflation attributable to non-actionable misrepresentations, and damages for materialization of a known risk. Opp. at 29–30.
Predominance requires that “damages are capable of measurement on a class[-]wide basis,
Defendants rely on a handful of out-of-circuit cases for the proposition that Dr. Mason must resolve issues with his unexecuted model at this stage, but the large majority of “[c]ourts in this District have regularly found that the methodology proposed by Plaintiffs satisfies Comcast.” SEB Inv. Mgmt. AB v. Wells Fargo & Co., No. 22-CV-03811-TLT, 2025 WL 1243818, at *8 (N.D. Cal. Apr. 25, 2025) (allowing Dr. Mason‘s out-of-pocket methodology at class certification stage for stock inflation case); City of Sunrise Firefighters’ Pension Fund v. Oracle Corp., No. 18-CV-04844-BLF, 2022 WL 1459567, at *7–*10 (N.D. Cal. May 9, 2022) (collecting cases and addressing a similar expert report generally opining on how to construct an out-of-pocket model and disaggregate confounding information); see Dkt. No. 263-2 (citing dozens of cases rejecting similar Comcast arguments in securities cases). Plaintiffs also offer several substantially similar expert reports from other cases where Comcast challenges were rejected. See, e.g., Dkt. Nos. 263-3, 263-4 & 263-5.
The Court is not persuaded that the specific factual challenges to calculating class-wide damages in this case warrant deviation from this well-trodden path. Other courts have rejected
An out-of-pocket model addressing the difference in inflation at purchase and inflation at sale does track Plaintiffs’ theory of liability, even if other complicating factors in calculating inflation and addressing loss causation may arise in the model. “At the merits stage, . . . plaintiffs will be required to establish all the necessary elements, including loss causation. They will also be required to establish a damages model capable of disaggregating causes of inflation outside of the actionable harms alleged herein. If they cannot, the class may be decertified.” Junge v. Geron Corp., No. C 20-00547-WHA, 2022 WL 1002446, at *9 (N.D. Cal. Apr. 2, 2022). But for now, the Court is satisfied that damages are capable of being measured on a class-wide basis using an event study and an out-of-pocket methodology, and the Court finds that individualized damages issues will not predominate over common issues. Cf. In re SanDisk LLC Sec. Litig., No. 15-CV-01455-VC, 2018 WL 4293336, at *2 (N.D. Cal. Sept. 4, 2018) (“The admittedly short portion of the [expert] report addressing this damages methodology, coupled with its general acceptance, suffices to show for class certification purposes that class-wide damages can be determined without excessive difficulty and attributed to [plaintiffs‘] theory of liability.” (quotation
IV. CONCLUSION
The Court DENIES Defendants’ motion to exclude the damages-related opinions of Dr. Mason. Dkt. No. 257. The Court GRANTS Plaintiffs’ motion for class certification, Dkt. No. 237, and certifies a class of investors defined as:
All persons or entities who purchased or otherwise acquired the common stock of NVIDIA between August 10, 2017, and November 15, 2018, inclusive. Excluded from the Class are Defendants, the officers and directors of NVIDIA at all relevant times, members of their immediate families and their legal representatives, heirs, agents, affiliates, successors or assigns, Defendants’ liability insurance carriers, and any affiliates or subsidiaries thereof, and any entity in which Defendants or their immediate families have or had a controlling interest.
Mot. at 7; FAC ¶ 247.
The Court further ORDERS that Lead Plaintiffs Lannebo Kapitalförvaltning AB and Stichting Pensioenfonds PGB are appointed as Class Representatives and ORDERS that Lead Counsel Kessler Topaz Meltzer & Check, LLP and Bernstein Litowitz Berger & Grossmann LLP shall serve as Class Counsel.
The Court further SETS a case management conference in this case on April 21, 2026, at 2:00 p.m. The hearing will be held by Public Zoom Webinar. All counsel, members of the public, and media may access the webinar information at https://www.cand.uscourts.gov/hsg. All attorneys and pro se litigants appearing for the case management conference are required to join at least 15 minutes before the hearing to check in with the courtroom deputy and test internet, video,
The Court also ORDERS the Clerk to update the case caption to match the caption used in this order.
IT IS SO ORDERED.
Dated: 3/25/2026
HAYWOOD S. GILLIAM, JR.
United States District Judge
Notes
Exhibits G, J, N, R, W, Y, EE, and NN are transcripts of earnings calls and public statements at conferences and a shareholder meeting, and the Court may take judicial notice of “what was in the public realm at the time.” Von Saher v. Norton Simon Museum of Art at Pasadena, 592 F.3d 954, 960 (9th Cir. 2010) (quotation omitted). For the same reason, the Court may take judicial notice of the analyst reports in Exhibits K, FF, and OO, and the publicly available news articles and websites in Exhibits H, L, P, S, U, V, X, BB, and II. Similarly, Exhibits D, I, M, T, DD, JJ, MM, PP, QQ, RR, SS, and TT are publicly available SEC filings, and their accuracy cannot reasonably be questioned. Metzler Inv. GMBH v. Corinthian Colls., Inc., 540 F.3d 1049, 1064 n.7 (9th Cir. 2008) (noting that SEC filings are properly subject to judicial notice). For each of the above exhibits, the Court takes judicial notice “solely as an indication of what information was in the public realm at the time.” Von Saher, 592 F.3d at 960. Finally, the Court may take judicial notice of historical cryptocurrency prices in Exhibit YY where, as here, their accuracy cannot reasonably be questioned. Metzler, 540 F.3d at 1064 n.7 (addressing historical stock prices).