Nvidia Pulls Back From OpenAI and Anthropic Bets

Nvidia Signals End of Direct AI Startup Investments

Nvidia CEO Jensen Huang declared at the Morgan Stanley Tech, Media and Telecom conference on March 4 that the chipmaker's recent investments in OpenAI and Anthropic are likely to be its last direct bets on AI startups. The announcement marks a potential turning point in how the world's most valuable semiconductor company relates to its biggest customers.

Huang's stated reasoning centers on timing: with both OpenAI and Anthropic expected to pursue initial public offerings later in 2026, he suggested the window for private-market investments is closing. But market analysts and industry observers have noted that his explanation leaves significant questions unanswered about Nvidia's true motivations.

The $30 Billion OpenAI Investment That Shrank

The investment Nvidia finalized as part of OpenAI's record-breaking $110 billion funding round came in at $30 billion — substantially less than the up to $100 billion Nvidia had once pledged to commit. Bloomberg reported that Huang explicitly ruled out reaching the original $100 billion figure, though Nvidia has not publicly explained why the commitment was scaled back so dramatically.

This follows Nvidia's $10 billion stake in Anthropic from November 2025, bringing the chipmaker's total direct investment in the two leading AI labs to approximately $40 billion. The sheer scale of these positions — in companies that are also among Nvidia's largest GPU customers — has raised governance and antitrust questions that may have contributed to the pullback.

"The opportunity to invest closes once they go public." — Jensen Huang, Morgan Stanley conference, March 4, 2026

Strained Relationship With Anthropic

Behind the IPO reasoning, tensions between Nvidia and Anthropic have been visible for months. Just two months after Nvidia's $10 billion investment, Anthropic CEO Dario Amodei took the stage at Davos and — without naming Nvidia directly — compared the practice of selling advanced AI chips to approved Chinese customers to "selling nuclear weapons to North Korea."

The comment landed poorly with Nvidia leadership, which derives significant revenue from Chinese markets within the bounds of U.S. export regulations. The diplomatic fallout has not been publicly resolved, and industry insiders suggest the Davos remarks accelerated Nvidia's decision to limit future exposure to Anthropic.

Nvidia now holds stakes in two companies pulling in sharply different directions — OpenAI pursuing Pentagon contracts while Anthropic refuses military applications — creating an uncomfortable position for a supplier that needs both as customers.

Bubble Concerns and IPO Reality

Growing concern about an AI investment bubble may also explain the retreat. Nvidia's combined $40 billion position in just two AI companies represents a concentration risk that few public companies would normally tolerate. With OpenAI's valuation reaching $730 billion and Anthropic's estimated above $100 billion, any correction in AI valuations could hit Nvidia's balance sheet hard.

Huang's claim that IPOs close the investment window has been questioned by investors who note that late-stage private investments regularly continue right up to and through public offerings. The real signal may be that Nvidia sees better risk-adjusted returns in its core business of designing and selling chips than in holding equity stakes in its own customers.

What This Means for the AI Industry

For engineers and professionals tracking the AI landscape, Nvidia's pullback signals a maturing market where the lines between supplier, investor, and competitor are becoming harder to manage. The move could accelerate both companies' IPO timelines as they seek alternative sources of capital, and it may open the door for other strategic investors to take larger positions.

The broader implication is that the era of mega-scale private AI investments may be peaking. As frontier AI companies prepare to go public, the industry is transitioning from a growth-at-all-costs phase to one where governance, conflicts of interest, and sustainable business models matter more.