KOHEN AVOCATS · PARIS, FRANCE
NVDA stock drop: can investors sue for securities fraud and what must the claim prove

Hassan Kohen is a Paris-based lawyer admitted to the Paris Bar (Avocat au Barreau de Paris). Initial strategic analysis: €80 incl. VAT, personal reply within 24 hours.

100% confidential · Legal professional privilege · No obligation

Article generated by artificial intelligence, through a process designed and supervised by the firm

Sources: U.S. court opinions and official U.S. government publications, read for this article

Maître Hassan KOHEN, avocat au Barreau de Paris
Hassan KOHEN
Attorney at the Paris Bar

NVDA stock drop: can investors sue for securities fraud and what must the claim prove

Speak to a lawyer 06 89 11 34 45

NVDA stock is back in heavy US search demand amid another sharp selloff. For American readers sitting on losses, the practical question is immediate: does a falling chip stock support a federal securities fraud suit, and what must the claim actually prove? A price drop alone never answers that question. Under federal law, an investor needs a materially false or misleading statement, facts giving rise to a strong inference that the speaker acted knowingly or recklessly, and a lawsuit filed within strict time limits. Press chatter about valuation, analyst targets, or online claims of hidden billions proves none of those elements.

The controlling courtroom story is the NVIDIA cryptomining litigation. Investors alleged that the company understated how much of its gaming graphics revenue came from cryptocurrency miners, and that its officers spoke falsely when they minimized crypto’s role. The United States Court of Appeals for the Ninth Circuit revived key claims in August 2023, holding that the complaint adequately alleged both falsity and scienter against the company and its co-founder. In December 2024 the Supreme Court dismissed NVIDIA’s petition as improvidently granted, leaving that revival standing. Separately, the Securities and Exchange Commission settled charges in May 2022 over inadequate cryptomining disclosures in the company’s fiscal 2018 quarterly reports.

The discussion that follows is general in scope: it describes checks useful in comparable situations and makes no findings about NVIDIA’s situation. It explains what the three decisions below require of any federal securities fraud complaint, how the governing statutes frame falsity, scienter, and deadlines, and where the SEC enforcement track differs from a private damages suit. As of this writing, no court has held that any particular recent NVDA price move was caused by fraud; only a filed complaint’s well-pleaded allegations, tested under these standards, can begin that inquiry.

I. What the federal courts decided in the NVIDIA cryptomining case

A. The Ninth Circuit revived the fraud claims against the company and its co-founder

NVIDIA is one of the world’s largest producers of graphics processing units, chips developed for gaming that also perform the repetitive computations used in crypto mining. According to the opinion, the investors’ amended complaint alleged that gaming-segment revenue was largely driven by crypto-related sales while the company’s public statements minimized crypto’s impact. In Ohman v. NVIDIA Corp., 81 F.4th 918 (United States Court of Appeals for the Ninth Circuit, August 25, 2023), No. 21-15604 (official slip opinion), the court reversed the dismissal in the part that mattered most to investors. The Ninth Circuit held that the investors stated a claim for relief under Section 10(b) and Rule 10b-5 against Huang and NVIDIA, but not against Kress and Fisher. The amended complaint sufficiently alleged that, during the class period, co-founder Jensen Huang made false or misleading statements about crypto-driven demand and did so knowingly or recklessly, while the allegations against fellow defendants Kress and Fisher fell short. The court likewise affirmed dismissal of the control-person claims against Kress and Fisher, vacated dismissal of those claims as to Huang, and remanded for further proceedings.

The opinion’s reasoning gives every later plaintiff a template and every later defendant a warning. On falsity, the panel credited detailed allegations, including former-employee accounts and a Prysm Group expert estimate that crypto-related revenue was understated by $1.125 billion, showing that gaming revenue depended on miners far more than the company’s statements admitted. On scienter, the panel held that the complaint tied the contents of internal data sources about crypto demand to the specific statements closely enough to support a strong inference that Huang spoke knowingly or recklessly. The court was explicit about the governing principle it applied: while the Private Securities Litigation Reform Act significantly altered pleading requirements in private securities fraud litigation, it did not impose an insurmountable standard. For readers, the application is direct: a complaint that only points at a stock chart and asserts fraud will fail, while a complaint that matches each challenged statement to contemporaneous internal facts about demand can survive, exactly as the Ohman plaintiffs’ core claims did.

Two boundaries in the decision matter as much as the revival. First, the court decided pleading sufficiency, not ultimate truth: holding that allegations, taken as true, state a claim is not a finding that the statements were in fact false or that any investor is owed damages. Second, the revival was selective, keeping claims against Huang and the company while ending them against Kress and Fisher, which shows that courts test scienter defendant by defendant rather than treating an entire executive team as one speaker. Those limits explain why later headlines about settlement talks or class certification, whatever press reports may claim, cannot be read as judicial findings of liability.

The control-person rulings in the same opinion deserve separate attention because investors often assume every officer shares liability. The court affirmed dismissal of the Section 20(a) claims against Kress and Fisher while vacating dismissal as to Huang and remanding, tying secondary liability to a viable primary violation committed by the controlled person. As the opinion put it for one of the dismissed officers, the only alleged primary violation was that committed through the company itself, which could not support that officer’s control-person claim. The check for any new complaint is therefore twofold: first plead a primary Section 10(b) violation by someone, then plead facts showing the officer actually controlled that violator, rather than relying on title or seniority alone.

B. The SEC settlement punished disclosure failures without deciding private damages

Parallel to the private lawsuit, the regulator extracted its own price. In the Commission’s May 2022 NVIDIA announcement, the Securities and Exchange Commission announced settled charges against the technology company for inadequate disclosures concerning the impact of cryptomining on its gaming business. The order finds that, during consecutive quarters of fiscal year 2018, the company failed to disclose that cryptomining was a significant element of its material revenue growth from gaming graphics units, even though it reported material gaming-revenue growth in two quarterly reports while holding information that cryptomining drove much of it. The full administrative order (File No. 3-20844) is available in the SEC’s May 6, 2022 NVIDIA order. Without admitting or denying the findings, the company agreed to a cease-and-desist order and to pay $5.5 million.

That settlement helps and limits private plaintiffs at the same time. It helps because an official federal order confirms that disclosure about cryptomining demand was deficient in the fiscal 2018 reports, which is a sourced public fact rather than an internet theory. It limits because an SEC settlement is not a judgment that any investor class is entitled to damages, and the “without admitting or denying” form means the order cannot be quoted as a confession of fraud. The two tracks also run under different rules: the Commission enforces disclosure duties in the public interest, while a private plaintiff must plead falsity, scienter, reliance, economic loss, and loss causation for its own purchases, and must survive the Reform Act’s screening before reaching discovery. An investor who treats the $5.5 million order as proof of a personal claim skips every element a court will actually test.

The practical check before citing the settlement is therefore narrow. Confirm the order’s date, number, and operative findings from the Commission’s own documents, note exactly which reports and quarters it covers, and resist extending it to other periods or other statements. The fiscal 2018 quarterly reports are inside the order; later disclosures, later GPUs, and later price moves are not. A complaint about a 2025 or 2026 NVDA drop must plead its own false statements and its own scienter facts, and cannot borrow the 2022 order as a substitute.

The order’s coverage is as important as its penalty. It addresses consecutive quarters of fiscal year 2018, gaming graphics units designed and marketed for gaming, and the failure to disclose that cryptomining formed a significant element of material revenue growth in two quarterly reports. A cease-and-desist order directs future compliance; it does not compute anyone’s damages, certify any class, or extend its findings to other quarters, other products, or later statements about artificial intelligence demand. An investor complaint that cites the settlement must therefore stay inside its four corners, and every allegation outside them needs its own pleaded facts and its own scienter showing.

The disclosure duty behind the settlement is worth stating precisely because investors often misread it. Quarterly reports must discuss known trends and uncertainties that make past results a poor guide to the future, which is why the order faults NVIDIA for reporting gaming growth without the cryptomining context it allegedly required. That duty belongs to the reporting company at filing time; it does not create a general obligation to update every prior statement whenever markets move later. A complaint about a later drop must therefore identify a statement that was false when made or that omitted a then-known fact the reports required, not merely a forecast that proved wrong.

II. What an investor must prove and check before suing over a drop

A. Falsity and scienter must be pleaded with particularity, witness by witness

The statute that creates the claim is short and the case law enforcing it is demanding. Under Section 10(b) of the Securities Exchange Act, 15 U.S.C. section 78j(b), it is unlawful “To use or employ, in connection with the purchase or sale of any security” any manipulative or deceptive device in contravention of Commission rules. The rule that implements it, Rule 10b-5, 17 C.F.R. section 240.10b-5, makes it unlawful, among other things, “To make any untrue statement of a material fact or to omit to state a material fact necessary in order to make the statements made, in the light of the circumstances under which they were made, not misleading,” in connection with a securities transaction. And the Reform Act, 15 U.S.C. section 78u-4, forces the complaint to “state with particularity facts giving rise to a strong inference that the defendant acted with the required state of mind”, an inference that holds only if a reasonable person would deem it cogent and at least as compelling as any opposing inference one could draw from the facts alleged. Those three provisions are the whole examination: a false or misleading statement, a culpable state of mind pleaded defendant by defendant, and everything else a court reaches only after those two are satisfied.

A February 2024 federal decision shows how courts test the final link in the chain, loss causation, even when falsity is adequately pleaded. In Reckstin Family Trust v. C3.ai, Inc., No. 22-cv-01413-HSG (United States District Court, Northern District of California, February 22, 2024), investors accused an artificial-intelligence software company and several officers of misrepresenting partnership and product claims. The court granted in part and denied in part the motions to dismiss, allowing some claims past the pleading stage while dismissing others with leave to amend. The Court finds that Plaintiffs have done enough at this stage to plead loss causation under this corrective disclosure theory. For a reader sitting on NVDA losses, the application is practical rather than comforting: even a complaint that quotes striking statements still needs trade-by-trade allegations tying an economic loss to the revelation of the truth, because a court tests each corrective disclosure separately instead of treating a falling price as proof.

The mirror image, where witnesses fail that test, is equally instructive. In Hollywood Firefighters v. Atlassian Corp., No. 3:23-cv-00519-WHO (United States District Court, Northern District of California, August 13, 2024), Judge Orrick dismissed a second amended securities complaint because the bulk of the new material came from one new confidential witness whose account was confusing, inconsistent, and ultimately unhelpful to scienter. The court dismissed the second amended complaint because the new confidential-witness allegations were confusing, inconsistent, and ultimately unhelpful to scienter, granting one final leave to amend. The lesson for an investor reading any complaint, including one about NVDA, is that amendment is a limited resource: the Atlassian plaintiffs had already survived one dismissal without prejudice and still failed, and courts grant final chances, not endless ones. A complaint’s length proves nothing; the reliability of each witness, tested line by line, proves everything.

Omission liability carries its own limit, and the rule’s text states it. A company need not disclose every internal fact; liability attaches when an omission makes the statements actually made misleading in light of their circumstances. That is why the Ohman complaint focused on affirmative statements minimizing crypto’s role rather than on silence alone, and why the SEC order faults the quarterly reports for presenting gaming growth without the cryptomining context the reports allegedly required. A new complaint that attacks pure silence, with no misleading statement to anchor it, starts from a weaker position than one that quotes a specific reassurance and then pleads the internal facts contradicting it.

B. Timeliness, reliance, and loss complete the claim, and each has its own trap

Falsity and scienter open the courthouse door, but three further requirements decide whether anyone walks through it. First, the claim must be timely. Under 28 U.S.C. section 1658(b), a private securities fraud action “may be brought not later than the earlier of” two years after discovery of the facts constituting the violation or five years after the violation itself. The two clocks run together and the earlier one wins: an investor who learns the facts in year four still loses the claim at year five, while an investor on notice in year one can be barred in year three. Because discovery turns on when a reasonable investor should have connected public disclosures to the alleged fraud, corrective disclosures, restatements, and pointed press coverage all start arguments about notice that defendants press hard. Anyone holding NVDA losses should therefore date three things before calling the claim viable: the challenged statements, the moment the truth allegedly emerged, and the purchase dates, since only purchases in the affected window can belong to any class.

Second, the complaint must connect the lie to the loss. The statute covers deception in connection with the purchase or sale of any security, which means the plaintiff must have transacted, and must tie an economic loss to the revelation of the truth rather than to the market’s general mood. A chip stock can fall because crypto prices collapse, because data-center spending pauses, or because rates move; none of those explanations, standing alone, states a claim. The Ohman litigation illustrates the discipline: the investors tied specific gaming-revenue statements to internal crypto-demand facts and then to the alleged corrective events, which is why the falsity and scienter rulings mattered. An investor whose only evidence is a red portfolio and a bearish article has correlation without a claim, and courts dismiss exactly those complaints.

Third, the Reform Act channels the litigation itself. Competing investors seek appointment as lead plaintiff, the court controls whether and how the case proceeds as a class, and settlements require judicial approval after notice to the class. Those procedures protect absent investors, but they also mean that filing first confers no ownership of the case and that individual losses below the cost of proof may be better pursued through a class than alone. The practical checks before any filing are therefore documentary and chronological: preserve trade confirmations, keep the statements and disclosures that allegedly misled, calendar the two- and five-year clocks from the statute, and test every element against the decisions above rather than against headlines. Where those checks fail, the honest answer is that the drop, however painful, is not a lawsuit.

Dating discovery is where many claims quietly die, and the discipline is practical rather than theoretical. The two-year clock starts with facts constituting the violation, which courts measure by when the investor knew or should have known enough to plead the claim, not by when the investor felt certain of winning it. Earnings-call corrections, amended filings, enforcement announcements, and detailed investigative reporting can each start that argument, and defendants routinely build timelines from them. An investor should therefore reconstruct the paper trail month by month: the first public hint that gaming revenue included undisclosed mining demand, each subsequent confirmation, and the investor’s own trades against those dates. Gaps in that reconstruction become the defendant’s statute-of-limitations motion, and the five-year outside date in section 1658(b) ends the debate regardless of when discovery occurred.

Conclusion

A falling stock price asks a question that only documents can answer. The affirmed appellate law requires a materially false or misleading statement pleaded with particularity, a strong inference of scienter against each speaker tested through reliable witnesses, a transaction connected to the deception, and a filing inside both limitations clocks. The NVIDIA revival shows how detailed internal-demand allegations can satisfy those tests; the C3.ai ruling shows how loss causation must still be pleaded trade by trade; the Atlassian dismissal shows how unreliable witnesses sink even an amended complaint; and the Commission’s settlement shows that regulatory punishment and private recovery are separate proceedings with separate proof. Each of those outcomes turned on papers a reader can inspect: complaints, orders, quarterly reports, and computation of dates.

Applied to the current search interest, that discipline yields a short checklist. Identify the exact statement claimed to be false and the document that contains it. Match it to contemporaneous facts the speaker allegedly knew. Name the witnesses or documents that prove that knowledge and test their reliability. Date discovery and purchase against the two-year and five-year limits. And treat every online claim of concealed billions as unproven until a complaint pleads it and a court sustains it. Where the checklist holds, counsel can evaluate a claim on its merits; where it fails at any step, the market loss remains what most market losses are, a risk realized rather than a wrong committed.

None of this requires believing or disbelieving any particular headline about the latest selloff. It requires matching statements to knowledge, knowledge to dates, and dates to the statute, using the same kinds of documents the decisions above tested: complaints with named witnesses and internal data, quarterly reports read against what the company knew, and dockets that show what courts actually decided. An investor who can assemble that record has the beginning of a case worth evaluating; an investor who cannot has something equally valuable, an early answer that saves the cost of finding out late.

📄 Official circular