- Published on
Owed a billion dollars in Nvidia stock
- Authors
- Name
- DeepHorus
A former technical adviser to NVIDIA says the company owes him roughly $1 billion in stock after discovering that thousands of options granted during the chipmaker’s earliest years apparently vested but were never delivered. The adviser, who joined NVIDIA’s Technical Advisory Board in 1993, recently reconciled an old option agreement with NVIDIA’s corporate records and concluded that 9,375 shares remain unissued. He has since retained attorneys Allan Steyer of Steyer Lowenthal and Chris Burke of Korein Tillery to investigate the claim. The dispute centers on an old contract rather than NVIDIA’s modern compensation practices, and no judicial ruling has yet validated the alleged debt.
The adviser’s relationship with NVIDIA began after a meeting aboard his houseboat, the SS Vallejo, in Sausalito. NVIDIA co-founder Curtis Priem had noticed his work on nonlinear texture mapping while Priem was designing Sun Microsystems’ SPARCstation GX and Sense8 Corporation was porting virtual-reality software to the platform. The adviser implemented bilinear texture mapping with an Intel i750 DVI chip and later developed a faster biquadratic technique documented in US patent 5,796,426 and related patents. NVIDIA’s founders invited him to demonstrate the technology, believing it could help distinguish the company’s first graphics processor from competing products.
After joining the Technical Advisory Board, the adviser received an agreement for 25,000 NVIDIA options in September 1993. The contract said the options would be issued in quarterly installments, with all shares vesting one year after the grant date. He subsequently worked casually with Priem to port biquadratic texture mapping to NVIDIA hardware and helped prepare an Intel-sponsored virtual-reality demonstration shown in New York in 1993. NVIDIA eventually shipped the NV1 in 1995, but Microsoft’s DirectX software supported triangles rather than quads and lacked quadratic texture mapping, severely weakening the product’s prospects and leading NVIDIA to lay off a substantial portion of its staff.
In April 1996, however, NVIDIA’s chief financial officer informed the adviser that only 15,625 options had vested and instructed him to exercise them. He complied and then forgot the matter for nearly three decades. During a 2024 review of his documents, he discovered that the signed agreement appeared to provide for full vesting after ten quarters, or roughly two and a half years, rather than after the 30 quarters described by NVIDIA and its outside counsel. Because the vesting had already occurred under the agreement’s one-year language, the adviser believes every share in the original 25,000-option grant should have been issued.
NVIDIA has since conducted numerous stock splits, which the adviser describes as a cumulative adjustment of 480 times. Applying that factor to the 9,375 allegedly missing shares produces 4.5 million current NVIDIA shares, explaining how an obscure early option grant became a claim valued at approximately $1 billion. The amount is not a fixed cash obligation because its dollar value moves with NVIDIA’s market price, and it would also depend on taxes, legal fees, and any negotiated settlement terms. Nevertheless, the share count illustrates how an error involving a small 1993 grant could acquire extraordinary present-day value.
The adviser says his lawyers spent about a year exchanging correspondence with NVIDIA’s in-house and outside counsel, including Cooley, while citing case law and possible legal theories. NVIDIA reportedly did not challenge the authenticity of the option agreement, focusing instead on the claim’s age and expiration of the applicable statute of limitations. At a settlement meeting, Cooley’s position was, in the adviser’s account, essentially that NVIDIA would defend any lawsuit. After considering the risk of litigation, the adviser and his attorneys concluded that the roughly three-decade delay likely barred them from bringing an ordinary contract claim.
The episode highlights the importance of preserving option agreements, tracking vesting dates, and reconciling historical equity records before corporate actions or splits make discrepancies harder to resolve. It also raises questions about how long semiconductor companies may be exposed to disputes involving handwritten or informal compensation practices from their founding eras. For NVIDIA, the practical exposure may ultimately be limited by the statute of limitations, while the claimant may need to pursue a tolling agreement, contractual workaround, or other legal avenue. Regardless of the final outcome, the story offers a rare glimpse at how foundational graphics technology, long-forgotten grants, and modern AI-era valuations can intersect decades after a company’s earliest days.
