Nvidia once again beats all expectations with quarterly revenue of $96.2 billion, is reportedly acquiring Hugging Face for $12.9 billion, and has a foot in the door at almost every AI company that matters. Jensen Huang is the most generous man in the tech sector. However, anyone who accepts money from a sugar daddy knows there is a catch. In this case: eternal loyalty to his chips.
The figures Nvidia presented this week once again defy all imagination: $96.2 billion in revenue, more than a doubling year-on-year, of which $89 billion came from data centers. For the current quarter, the company is targeting $108 billion. A money machine like that has to do something with all that cash. CEO Jensen Huang chose his destination: his own customers.
Bank of America calculated that Nvidia has already committed about $70 billion in direct stakes, with $30 billion for OpenAI, up to $10 billion for Anthropic, and $5 billion for Safe Superintelligence as highlights. In the first four months of 2026 alone, Nvidia pumped over $40 billion into interests.
“We don’t pick winners, we have to support everyone,” Huang summarizes. The Dutch research institute SOMO did the math: nine out of the ten best-funded AI startups in the world have Nvidia as a shareholder.
It doesn’t stop at startups. Last year, Nvidia invested $5 billion in arch-rival Intel, a stake that grew to about $30 billion in seven months, and holds $21 billion in SpaceX. The latter company promptly declared Nvidia its exclusive AI partner. That’s how it works with a sugar daddy.
You simply buy away the competition
Those who don’t need money are convinced otherwise. Chip startup Groq, founded by the creators of Google’s TPU and worth barely $6.9 billion last year, received $20 billion in cash around Christmas for a ‘non-exclusive license’ to its inference technology.
The CEO and almost the entire top management moved with it. The largest deal in Nvidia’s history was structured so that no merger regulator was involved. U.S. Senators Warren and Blumenthal are now demanding clarification.
The largest deal in Nvidia’s history was structured so that no merger regulator was involved.
And now there is Hugging Face, the go-to platform for open-source AI. According to The Information, Nvidia is paying $12.9 billion for it, although official confirmation is currently lacking.
The logic is there: OpenAI, Google, Amazon, and Anthropic are building their own chips, so a thriving open-source ecosystem keeps the rest of the market neatly dependent on Nvidia. Anyone who downloads a model has to run it somewhere.
The money merry-go-round
This is where the story gets delicate. Nvidia invests in companies that use that money to buy Nvidia chips. Bernstein analyst Stacy Rasgon already warned of this ‘circular’ construction during the OpenAI deal, and examples have been piling up ever since.
Nvidia guarantees up to $105 billion in rent and energy obligations for the mega data center that OpenAI leases in Ohio: eight gigawatts, twenty years, exclusively filled with Nvidia chips. If OpenAI doesn’t pay the rent, Jensen covers the difference.
Furthermore, with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, Nvidia is setting up financing platforms intended to mobilize $500 billion in external capital, with computing power as collateral.
With the Korean SK Group, it signed for another $500 billion in AI factories and memory chips. The Wall Street Journal calculated that nine tech giants together keep $3,000 billion in mostly AI-related obligations off-balance sheet.
The skeptics are stirring. Michael Burry, known from The Big Short, compares Nvidia to Cisco in 2000. Huang dismisses it all: AI companies are the first generation of startups that need tens of billions of dollars. “The risk is limited because the infrastructure is reusable. My only regret is that we didn’t invest earlier and more.”
And Europe? Europe just pays
Fair is fair: the demand is real. AWS tripled its order this week with two million additional GPUs, and according to Bloomberg, Google is leasing about $33 billion worth of Nvidia capacity. That is cash from the richest companies in the world.
However, look at who is in Jensen’s will. OpenAI, Anthropic, Intel, SpaceX, Groq, CoreWeave: all American. Europe only appears as a mistress at best. While Nvidia did participate three times in a round for Mistral, that was a stake in a round of barely 1.7 billion euros, compared to $30 billion plus $105 billion in guarantees for OpenAI.
European sovereignty, built on American chips.
The much-discussed industrial AI cloud that Nvidia is building with Deutsche Telekom? One billion euros for ten thousand Nvidia GPUs: European sovereignty, built on American chips. All sovereign AI initiatives on our continent are not sponsorships, but a sales channel.
The sugar daddy pampers his American favorites; Europe gets the bill. For the European CIO, the calculation is sobering: almost every AI euro, whether it flows through AWS, Google Cloud, OpenAI, Mistral, or a sovereign cloud, ends up with Nvidia today.
The bill always comes
It is important to emphasize: this is no pipe dream. The revenue is real, the chips are the best in the world, and Huang has a point when he says that AI capital operates on a scale that the traditional financial system cannot handle.
On the other hand, when one company is simultaneously the supplier, shareholder, rent guarantor, and banker for its own customers, the word ‘demand’ loses its meaning as a market signal. Every chip sold then proves nothing more than that the money merry-go-round is still spinning.
Sugar daddy relationships rarely end well: either the money runs out, or the partner grows up and leaves with their own chip under their arm. In both scenarios, someone is left with the bill. Want to bet it’s not Jensen?