In a surprise announcement at a major Wall Street conference, Nvidia CEO Jensen Huang said the company’s recent investments in OpenAI and Anthropic are “likely” to be its last, but his brief explanation left analysts and rivals with more questions than answers.
Nvidia, the chipmaker that has become the undisputed engine of the global AI boom, may be stepping back from its role as an investor in the very startups it helped power. In a remarks at the Morgan Stanley Technology, Media and Telecom (TMT) conference in downtown San Francisco on Wednesday, CEO Jensen Huang told the audience that the company’s recent stakes in OpenAI and Anthropic will likely be its final investments in both firms.
“Once they go public, as they are anticipated to, the opportunity to invest in a significant company like this closes,” Huang said, explaining that the IPO window effectively shuts the door on further Nvidia capital into these AI labs. “That’s just the way the market works,” he added, according to a transcript of the event.
On the surface, the logic sounds straightforward: OpenAI and Anthropic are preparing for public listings, and with Nvidia already sitting on tens of billions of dollars in equity, there may be less appetite to keep pouring money into companies that are already among the most valuable AI startups in the world. Yet Huang’s explanation delivered in a short, throwaway line has reignited speculation about deeper strategic, political, and ethical tensions behind the scenes.
Nvidia’s “Last” Investments in OpenAI and Anthropic
Over the past year, Nvidia has positioned itself as both a hardware supplier and a strategic investor in the AI ecosystem. In September 2024, the company announced a landmark deal to invest up to $100 billion in OpenAI, timed with massive infrastructure commitments that would see OpenAI buying tens of billions of dollars of Nvidia chips. Industry watchers, including MIT Sloan professor Michael Cusumano, described the original structure as “kind of a wash,” noting that “Nvidia is investing $100 billion in OpenAI stock, and OpenAI is saying they are going to buy $100 billion or more of Nvidia chips.”
By February 2026, however, the numbers had shifted. A final agreement valued OpenAI’s latest funding round at around $110 billion, with Nvidia committing $30 billion far short of the original $100 billion figure. Huang has since acknowledged that deploying the full $100 billion “probably isn’t in the cards,” signaling a scaling‑back of ambition even before Wednesday’s TMT appearance.
In parallel, Nvidia announced a $10 billion investment in Anthropic in November 2025, alongside a “deep technology partnership” that tied Nvidia’s top‑tier GPUs to Anthropic’s Claude models. Those equity stakes, combined with Nvidia’s near‑monopoly on AI training hardware, painted the company as both a supplier and a co‑owner of the AI future.
Now, Huang is suggesting that era may be ending. “Our investments are focused very squarely, strategically on expanding and deepening our ecosystem reach,” he said in a prior earnings‑call transcript referenced by Nvidia staff after his remarks. “In that sense, the investments in OpenAI and Anthropic have already achieved that objective.”
A “Clean” Supplier Relationship or Something More?
Huang’s wording is intentionally vague, but several dynamics make his explanation feel incomplete to many observers. First, Nvidia is already “minting money” from selling chips to OpenAI, Anthropic, and other AI companies, which means additional equity stakes are less critical for financial upside. As one analyst quoted in TechCrunch noted, Nvidia does not need to “goose its returns” by piling more capital into firms that are already among its biggest customers.
At the same time, the circular nature of earlier deals where Nvidia’s equity investments were tied to massive hardware purchases has raised concerns about an artificial valuation bubble. Some investors now worry that the “wash” structure masked deeper risk, prompting Nvidia to pare back commitments and shift toward a cleaner, more traditional supplier‑customer relationship.
Huang also dismissed the idea that there is “bad blood” between Nvidia and either OpenAI or Anthropic as “nonsense.” Yet tensions have been evident, particularly in the case of Anthropic. In January 2026, Anthropic CEO Dario Amodei took the stage at Davos and, without naming Nvidia directly, compared the act of U.S. chipmakers selling high‑performance AI processors to approved Chinese customers to “selling nuclear weapons to North Korea.” The remark, which clearly pointed to Nvidia and AMD, drew sharp criticism from industry and government figures and highlighted the increasingly fraught intersection of AI, geopolitics, and national security.
Geopolitics, Blacklists, and Competing Loyalties
That friction has only deepened in recent weeks. Just days before Huang’s Morgan Stanley comments, the Trump administration moved to blacklist Anthropic, labeling the company a “supply‑chain risk” and instructing federal agencies and military contractors to stop using its technology. The decision came after Anthropic refused Pentagon requests to loosen safeguards that would have allowed its models to be used in autonomous weapons or mass domestic surveillance.
“Anthropic’s stance creates a real complication for us,” one unnamed government official familiar with the situation told Axios, speaking on condition of anonymity. “Here you have a major AI lab that’s deeply embedded in federal systems, yet unwilling to align with certain national‑security requirements.”
In contrast, OpenAI has moved closer to the U.S. defense establishment. In late February 2026, the company announced a deal with the Pentagon that would expand the use of its models in defense and intelligence applications, a move that Anthropic has publicly labeled “mendacious” and that many civil‑liberties groups have criticized. The dueling alignments, OpenAI moving toward military‑aligned work, and Anthropic maintaining strict ethical boundaries have left Nvidia straddling two very different poles.
Huang’s simple explanation that investment windows close when firms go public feels “hard to square with how late‑stage private investing actually works,” as one TechCrunch analysis put it. Several venture capital insiders pointed out that large funds routinely invest in companies months, even weeks, before IPOs, suggesting that the timing alone does not fully justify Nvidia’s pullback.
Conflict‑of‑Interest and Nvidia’s Own AI Play
Another layer of the puzzle is Nvidia’s own expanding role in AI services. The company has aggressively moved beyond selling GPUs into offering cloud AI platforms, enterprise software stacks, and consulting aimed at building and deploying AI models. This shift places Nvidia in potential competition with the very startups it once backed, creating a conflict‑of‑interest headache if it remains a major shareholder.
“Holding equity in OpenAI and Anthropic while building rival AI services is a recipe for tension,” said one AI‑focused venture partner who spoke to TechBuzz.ai. “Customers will start to ask, ‘Who does Nvidia really work for?’” The risk, analysts warn, is that Nvidia’s hybrid role like hardware supplier, investor, and service provider could undermine trust at a time when openness and transparency are under intense scrutiny.
Huang has long argued that “if you want things to be done safely and responsibly, you do it in the open,” criticizing companies that operate in what he has described as “a dark room” and then declare their work “safe.” Those remarks, which he directed at Anthropic’s leadership in 2025, laid bare philosophical differences between Nvidia’s open‑ecosystem approach and Anthropic’s more guarded, safety‑first stance.
What This Means for the AI Landscape
For the broader AI industry, Nvidia’s move signals a potential turning point in how capital and power are structured. For years, the AI boom has been fueled by a symbiotic relationship: startups build models, Nvidia supplies the chips, and both use equity stakes to cement alignment. If Nvidia now steps back from the investor role, other players like large tech firms, sovereign‑wealth funds, and traditional venture capital may need to fill the gap.
“This could reshape funding dynamics for AI startups that have relied on Nvidia’s backing alongside its hardware,” noted TechBuzz.ai. OpenAI and Anthropic, already among the best‑capitalized AI labs, may feel less pressure, but smaller rivals could find it harder to secure “hardware‑plus‑equity” deals of the kind Nvidia once offered.
Consumer‑level data also hint at how these political and philosophical divides are playing out. Within hours of the Pentagon‑related announcements, Anthropic’s Claude app shot to the top of the free‑app rankings on Apple’s U.S. App Store, briefly overtaking ChatGPT, which had long dominated the category. According to Sensor Tower metrics cited in TechCrunch, Claude was outside the top 100 apps in the U.S. at the end of January; by late February, it had surged into the top five.
What Jensen Huang Didn’t Say
Throughout the Morgan Stanley event, Huang declined to elaborate on the geopolitical or philosophical undercurrents that may have contributed to Nvidia’s shift. He also did not address whether the company is considering similar exits from other AI‑focused investments, or what, if any, regulatory or internal compliance pressures are shaping the decision.
When asked for clarification, a Nvidia spokesperson referred reporters to the transcript of the company’s fourth‑quarter earnings call, where Huang reiterated that investments are “focused very squarely, strategically on expanding and deepening our ecosystem reach.” “In that sense,” the spokesperson said, “the investments in OpenAI and Anthropic have already served that purpose.”
Yet to many observers, that line feels more like a corporate‑speak placeholder than a full explanation. The depth and speed of the shifts, OpenAI’s pivot toward defense work, Anthropic’s placement on a national‑security blacklist, and the growing unease over circular financing deals suggest Nvidia is recalibrating not just its balance sheet, but its role in the AI ecosystem.
Looking Ahead
With OpenAI and Anthropic both expected to go public in 2026, Nvidia’s stake in both companies will become a major topic for public‑market investors. Analysts will scrutinize how much of its exposure the company intends to keep, and whether it will gradually sell down its positions as the IPOs proceed.
In the meantime, Huang’s comments at the Morgan Stanley conference have done more than just signal a retreat from direct investment. They have thrown a spotlight on the tangled web of commercial, political, and ethical relationships that underpin the AI race and raised a fundamental question that Nvidia has yet to answer fully: how the company wants to position itself in an era where chips, code, and national security are inextricably intertwined.
Comments