•
The Supreme Court took the Nvidia shareholder case, heard argument, then dismissed it in one sentence. The Ninth Circuit's ruling stood, and in March 2026 a district judge certified the class.
This post used to say the Supreme Court would decide whether the Nvidia shareholder suit could proceed. It did not decide. On December 11, 2024 the Court dismissed the writ of certiorari as improvidently granted, a one-sentence order issued after full briefing and oral argument. The Ninth Circuit's decision stood, the case went back to the district court, and in March 2026 a judge certified the class.
The lesson for anyone tracking litigation is worth stating plainly: a cert grant is not a preview of a ruling. Roughly one case a term ends this way.
| Item | Detail |
|---|---|
| District case | In re NVIDIA Corporation Securities Litigation |
| Court | U.S. District Court, Northern District of California |
| Docket | 4:18-cv-07669-HSG |
| Filed | December 21, 2018 |
| Supreme Court case | NVIDIA Corp. v. E. Ohman J:or Fonder AB, No. 23-970 |
| Argued | November 13, 2024 |
| Disposition | Certiorari dismissed as improvidently granted, December 11, 2024 |
| Class certified | March 25, 2026 |
| Claims | Exchange Act section 10(b), Rule 10b-5, and section 20(a) |
Between 2017 and 2018, cryptocurrency mining drove enormous demand for graphics processing units. Miners bought consumer gaming cards by the pallet.
Shareholders allege that Nvidia understated how much of its gaming GPU revenue actually came from crypto mining, and that executives knew the real figure. The class period runs from May 10, 2017 to November 14, 2018. When the crypto market fell in late 2018, gaming GPU sales fell with it, the stock dropped sharply, and investors say they were surprised because the company had described gaming demand as independent of mining.
The claims are the standard securities fraud set: section 10(b) and Rule 10b-5 for the misstatements, section 20(a) for control person liability against the named executives.
The Private Securities Litigation Reform Act requires a securities fraud complaint to plead facts giving rise to a strong inference of scienter, meaning intent or conscious recklessness, before any discovery happens. That standard is why most of these cases die at the pleading stage.
The plaintiffs got past it partly by relying on an outside expert analysis estimating crypto-driven revenue, and on allegations that internal reports gave executives the real numbers.
Nvidia asked the Supreme Court to adopt two rules. First, that a complaint relying on internal documents must plead their specific contents. Second, that plaintiffs cannot satisfy the PSLRA by relying on an outside expert's reconstruction of what the company must have known.
Both rules, if adopted, would have raised the pleading bar across every securities class action in the country. That is why the case drew the attention it did, and why the dismissal was an anticlimax.
A dismissal as improvidently granted leaves no precedent. The Ninth Circuit's opinion remains binding in that circuit and persuasive nowhere else. Commentary from A&O Shearman and the Federalist Society reads it as a deferral rather than a resolution, and the pleading question will come back on another vehicle.
Class certification changes the shape of the case. It converts individual claims into aggregate exposure, which is usually what moves a defendant from litigating to negotiating.
Expect a Rule 23(f) petition for interlocutory review of the certification order, summary judgment briefing on loss causation, and a fight over the damages model. Nvidia's own periodic filings with the SEC are a reliable, free public source for the company's characterization of the case at each stage.
Pull the current filings here: In re NVIDIA Corporation Securities Litigation. Post-certification dockets in securities cases move in long quiet stretches punctuated by heavy filing weeks, which is exactly the pattern case alerts handle better than manual checking.
Judge Gilliam's record on class certification and summary judgment in securities cases is the most useful predictor of what happens next. The federal judge lookup pulls it.
The Court does not explain these orders, so reading them requires inference.
The usual explanations are that the case turned out to be a poor vehicle, because the record was muddier than the petition suggested or the question was not squarely presented, or that the argument revealed the parties were closer than the briefing implied, leaving no clean rule to announce.
What it does not mean is that the Court endorsed the decision below. A dismissal leaves the lower court judgment in place without approving it, and creates no precedent anywhere.
The same term produced a dismissal in a parallel securities case, which is why practitioners read the two together as a signal that the Court was not ready to touch PSLRA pleading standards rather than as a comment on either case.
Certification orders in securities cases turn on the presumption of reliance from Basic v. Levinson and on whether the defendant rebutted it by showing the alleged misstatements had no price impact. Goldman Sachs Group v. Arkansas Teacher Retirement System sharpened that inquiry in 2021.
The order is the most informative document in the case right now, because it tells you how the court evaluated price impact evidence, which is the same evidence that will drive any damages model.
For a Supreme Court case that did produce a merits ruling on intermediary liability, read Cox Communications v. Sony Music. For a securities-adjacent fraud prosecution where the appellate process ran to completion, see the Theranos verdict. And for how the docket is the only reliable source when coverage of a case gets ahead of the record, see how journalists investigate federal court cases.
Read the available docket record. These pages reflect the records collected so far; check the source for subsequent filings.
District Court, N.D. California · Filed December 21, 2018