Jensen Huang Says Nvidia Is Pulling Back From OpenAI and Anthropic — But His Explanation Doesn’t Add Up

Nvidia GPU chips with AI company connections on dark tech background

Jensen Huang has announced that Nvidia is “pulling back” from its investments in OpenAI and Anthropic — the two companies that happen to be Nvidia’s biggest AI chip customers. The explanation, according to TechCrunch, “raises more questions than it answers.”

When the company that manufactures virtually every GPU used to train frontier AI models is also an investor in the companies buying those GPUs, the conflict of interest isn’t subtle. It seems Nvidia has finally acknowledged what everyone else already knew: you can’t be both the dealer and a player at the same table.

The Uncomfortable Dual Role

Nvidia’s position in the AI industry is unprecedented. The company supplies roughly 80-90% of the GPUs used for AI training worldwide. OpenAI, Anthropic, Google, Meta, and every other major AI lab are essentially dependent on Nvidia hardware to build their models.

At the same time, Nvidia had invested in several of its biggest customers, including both OpenAI and Anthropic. This created a web of conflicts:

  • Chip allocation decisions — When GPUs are scarce (as they have been), how do you fairly allocate chips between companies you’ve invested in and companies you haven’t?
  • Pricing leverage — An investor has access to financial information and strategic plans. A supplier sets prices. Combining those roles creates obvious advantages.
  • Competitive intelligence — As an investor in both OpenAI and Anthropic, Nvidia potentially had insight into both companies’ strategies, roadmaps, and capabilities.
  • Antitrust exposure — Regulators worldwide are increasingly scrutinizing AI industry concentration, and Nvidia’s dual role as supplier-investor was an obvious target.

What Jensen Actually Said

Huang’s public explanation for the pullback has been notably vague. Rather than citing specific regulatory concerns or ethical principles, he framed it as a strategic decision about where Nvidia’s focus should be — language that typically translates to “our lawyers told us this was becoming a problem.”

The timing is also telling. This announcement comes amid:

  • The Pentagon labeling Anthropic a “supply chain risk”
  • The DOJ and FTC stepping up antitrust enforcement in AI
  • Growing scrutiny of the cozy relationships between AI chip makers and AI labs
  • Multiple countries launching investigations into AI market concentration

Nvidia pulling back now feels less like a proactive decision and more like getting ahead of regulatory action that was coming anyway.

The Antitrust Shadow

Nvidia is already under antitrust scrutiny. The company’s dominance in AI chips has drawn attention from regulators who worry that control over the hardware layer gives Nvidia outsized influence over the entire AI industry.

Investing in the companies you supply amplifies that concern dramatically. If Nvidia can see OpenAI’s financial data and control how many H100s they get, the power dynamic is wildly asymmetric. Even if Nvidia never actually abused that position, the potential for abuse is enough to trigger regulatory action.

By divesting from AI companies now, Nvidia can tell regulators: “Look, we’ve cleaned up our own house. We’re just a chipmaker, not an AI conglomerate.” Whether regulators buy that argument is another question entirely — Nvidia’s 80%+ market share in AI GPUs is a monopoly problem regardless of its investment portfolio.

Impact on OpenAI and Anthropic

For OpenAI, losing Nvidia as an investor is largely symbolic at this point. The company has raised over $30 billion and has SoftBank trying to borrow $40 billion more to invest. Nvidia’s stake, while significant, isn’t make-or-break.

For Anthropic, the pullback comes at a more sensitive time. The company is fighting the Pentagon’s “supply chain risk” label, watching its commercial growth surge from the ChatGPT backlash, and trying to raise additional capital. Losing an investor who also happens to be your primary hardware supplier adds complexity — though it may actually help Anthropic’s narrative that it operates independently from Big Tech influence.

The Bigger Question: Who Controls AI?

Nvidia’s pullback is a symptom of a larger question the AI industry hasn’t answered: who should be allowed to control multiple layers of the AI stack?

Today, a handful of companies dominate:

  • Nvidia — chips
  • Microsoft/Google/Amazon — cloud infrastructure
  • OpenAI/Anthropic/Google — frontier models

When these companies invest in each other, partner with each other, and depend on each other, the resulting web of relationships makes it nearly impossible for new entrants to compete. Nvidia stepping back from one thread of that web doesn’t untangle it — but it’s an acknowledgment that the current structure is unsustainable.

The Bottom Line

Jensen Huang pulling Nvidia’s investments from OpenAI and Anthropic is the right move, but for the wrong reasons. This isn’t about principle — it’s about regulatory self-preservation. The conflicts of interest were obvious from day one, and Nvidia is only addressing them now because the legal and political pressure became impossible to ignore.

The AI industry’s dirtiest secret isn’t what the models can do. It’s how intertwined the companies building them are. Nvidia divesting is a start, but the real cleanup — separating chip makers, cloud providers, and AI labs into genuinely independent entities — is a conversation nobody in Silicon Valley wants to have.