Nvidia Reportedly Scales Back OpenAI Investment to $30 Billion After $100 Billion Plan Stalled

What Happened

Nvidia is "close to finalizing" a $30 billion direct equity investment in OpenAI, Reuters and the Financial Times reported on February 19-20, 2026. CNBC independently corroborated the same figure on February 19. Neither Nvidia nor OpenAI has officially confirmed the deal; all sources consistently use language such as "close to finalizing" and "expected."

The $30 billion figure replaces a much larger — and structurally different — commitment. In September 2025, Nvidia and OpenAI announced a strategic partnership under which Nvidia intended to invest up to $100 billion, but the funds were tied to infrastructure deployment milestones: capital would unlock as each gigawatt of Nvidia systems was deployed across a planned 10-gigawatt buildout. That agreement never progressed to a binding contract. By late January 2026 it was widely reported as stalled; Nvidia CEO Jensen Huang publicly denied a collapse on January 31 while stopping short of confirming the full $100 billion figure. The new $30 billion structure carries no milestone conditions, separating Nvidia's ownership stake cleanly from its vendor relationship.

Nvidia's investment is part of a broader round targeting more than $100 billion in total, expected to value OpenAI at approximately $730 billion pre-money (roughly $830 billion post-money, after new capital closes). SoftBank is reportedly committing $30 billion, Amazon up to $50 billion, and Microsoft a smaller sum. OpenAI's 2025 revenue exceeded $20 billion, and the company plans to direct a significant share of the new capital toward purchasing Nvidia chips for AI training and inference infrastructure.

Why It Matters

The original $100 billion plan represented one of the most unusual financial structures in tech history: a chip supplier tying equity investment directly to hardware purchases. Its collapse into a conventional $30 billion direct stake tells a more honest story about both companies' confidence levels than the headline number does.

Multiple reports indicate that Nvidia executives harbored private reservations about OpenAI's valuation premium and customer concentration risk — Nvidia already generates a substantial share of its data center revenue from a handful of large AI companies, and a nine-figure stake in its single largest customer would deepen that exposure. The reduction to $30 billion reflects measured commitment, not the unconditional endorsement the September 2025 announcement implied. As Cleo Capital managing director Sarah Kunst put it: "Back and forth between an investor and a startup playing out in the media isn't normal" (CNBC, Feb. 2026).

The deal now creates a new structural reality: Nvidia owns equity in its largest customer. Jensen Huang stated "there is no question we plan to participate" in OpenAI's funding round (TechStartups, Feb. 2026), and Sam Altman confirmed OpenAI "will remain a gigantic customer" of Nvidia (Benzinga, Feb. 2026). But antitrust law professor Rebecca Haw Allensworth identifies a risk in that arrangement: "This creates an incentive for Nvidia to not sell chips to, or not sell chips on the same terms to, other competitors of OpenAI" (Tom's Hardware, Sept. 2025). Neither company has responded substantively to that concern. The FTC and DOJ have signaled heightened scrutiny of AI industry consolidation, though no formal regulatory review of this deal has been announced as of February 2026.

A counterweight to Nvidia's leverage is already in motion. OpenAI is reportedly negotiating with AMD, Groq, and Cerebras for inference workloads, driven by performance concerns with Nvidia's chips on inference-heavy tasks including coding and AI-to-software applications. The equity stake deepens the financial relationship; OpenAI's supplier hedging limits how deep the hardware dependency can actually run.