The company that powers nearly every AI model just said it's finished funding the companies that build them.
At the Morgan Stanley TMT Conference on Wednesday, Nvidia CEO Jensen Huang said the company's $30 billion investment in OpenAI "might be the last" direct equity stake in the company. He said the same about Nvidia's $10 billion commitment to Anthropic, calling it "probably" the final investment there, too.
His reasoning? Both companies are headed toward IPOs, and Nvidia doesn't see itself as a pre-IPO investor once they go public.
It's a tidy explanation. It also doesn't quite hold up.
Strategic investors regularly participate in funding rounds right up to a company's public debut. As TechCrunch noted, the IPO-timing rationale "raises more questions than it answers." And then there's the tone shift: In January, Huang was effusive about OpenAI, calling it "one of the most consequential companies of our time." Two months later, he's winding things down. On the earnings call, he framed the investments as a strategic play to expand Nvidia's ecosystem reach — a goal he suggested had already been achieved.
Huang dismissed the notion that there's tension between Nvidia and the labs. But the less comfortable explanations have nothing to do with personal relationships.
Nvidia controls roughly 90% to 95% of the market for advanced AI chips. When it pours billions into the same companies buying those chips, the arrangement starts to look circular. Michael Cusumano, a professor at MIT Sloan, put it bluntly: "Nvidia is investing $100 billion in OpenAI stock, and OpenAI is saying they are going to buy $100 billion or more of Nvidia chips... kind of a wash."
Rebecca Haw Allensworth, an antitrust professor at Vanderbilt Law School, flagged a sharper risk. "They're financially interested in each other's success," she said. "That creates an incentive for Nvidia to not sell chips to, or not sell chips on the same terms to, other competitors of OpenAI."
This isn't hypothetical. Nvidia structured its $20 billion acquisition of chip startup Groq last December as a licensing deal and talent acquisition, specifically to sidestep regulatory review. The company is clearly aware of where the lines are.
Nvidia isn't done spending, though. It's just spending differently. In January, the company announced a $1 billion co-innovation lab with Eli Lilly for AI-powered drug discovery. That's the kind of industry partnership that puts Nvidia's chips to work without the entanglement of owning a stake in a lab that buys your products. Nvidia's future isn't in picking which lab wins. It's in selling infrastructure to every company that wants to build.
Valley View
Nvidia spent the last two years playing both sides of the AI boom, selling the picks and shovels while buying stakes in the gold mines. That was always going to be temporary. Whether the pullback is driven by antitrust risk, IPO logistics, or simply recognizing that you shouldn't own your best customers, the signal is the same: the era of chipmakers bankrolling AI labs is over. What comes next will tell us whether this industry is maturing into something sustainable or still running on circular billions.
