When Nvidia wired $30 billion to OpenAI earlier this year, it did not just write the largest check in its 33-year history. It tied its fortunes to a single customer more tightly than any chipmaker has done in modern memory. The investment anchored a funding round that Bloomberg reported totaled $122 billion, pushing OpenAI’s valuation to $852 billion. Amazon led with $50 billion, SoftBank added $30 billion, and a handful of smaller backers filled the rest, according to The Associated Press. The numbers are staggering on their own. What has drawn sharper scrutiny is the structure behind them: Nvidia is bankrolling the very company that funnels billions back to Nvidia for chips.
The scale of Nvidia’s bet
The $30 billion OpenAI stake accounts for roughly three-quarters of what Bloomberg estimates is $40 billion Nvidia has poured into AI startups in 2026. Nvidia has not confirmed that total in any regulatory filing. What it has confirmed publicly is a letter of intent with OpenAI to deploy 10 gigawatts of Nvidia-powered computing systems. To put that in perspective: 10 gigawatts is enough electricity to run roughly 7.5 million American homes, or about every household in the state of Ohio. As a data center buildout, it would dwarf anything a single company has ever attempted and lock OpenAI into Nvidia’s hardware ecosystem for years.
Nvidia can afford the gamble, at least on paper. The company ended fiscal year 2026 with over $43 billion in cash and short-term investments, and its market capitalization has hovered above $3 trillion. But concentration risk is real. In its most recent 10-K, Nvidia disclosed that a “different indirect customer” (widely understood to be Microsoft, OpenAI’s cloud host) accounted for 13% of total revenue. A $30 billion equity position in OpenAI deepens that exposure. If OpenAI thrives, Nvidia profits twice: once on the stock, once on the silicon. If OpenAI stumbles, Nvidia absorbs losses on both fronts simultaneously.
Why critics call it circular
Bloomberg’s investigation into what it labeled “circular AI financing” laid out the mechanics in blunt terms. Nvidia invests billions in an AI startup. That startup uses the capital to buy Nvidia GPUs. Nvidia books the resulting sales as revenue, which lifts its stock price and frees up capital to invest in the next startup. Repeat. The same pattern applies to Amazon and Microsoft, which pour money into AI companies that then pay those same cloud providers for infrastructure.
“This is the venture capital equivalent of a perpetual motion machine,” one semiconductor analyst told Bloomberg, “and perpetual motion machines don’t exist.” The concern is not that any single transaction is improper. The concern is that the entire chain depends on one assumption: that AI products will eventually generate enough independent revenue to justify the spending. If that revenue never materializes at scale, the loop breaks. Nvidia would face a simultaneous hit to its investment portfolio and its core chip business. Amazon and Microsoft could be left operating data centers built for workloads that no longer exist. Bloomberg’s analysis warned that the interconnected nature of these bets means a downturn would not stay contained to one balance sheet.
None of the companies involved have directly rebutted the circular-deal characterization. Nvidia has not published a detailed strategic rationale for the $30 billion stake beyond the letter of intent. OpenAI has not disclosed what share of its new funding is earmarked for hardware purchases versus research, hiring, or other costs. Without those details, outside analysts are left estimating how much of the $30 billion will cycle straight back to Nvidia’s top line.
The Microsoft and Amazon entanglement
OpenAI’s funding round did not happen in a vacuum. Microsoft, which has committed roughly $13.75 billion to OpenAI since 2019 and hosts the company’s workloads on Azure, remains deeply embedded in OpenAI’s operations. Amazon’s $50 billion entry, the largest single commitment in the round per the AP, positions Amazon Web Services as a potential second cloud provider for OpenAI and challenges Microsoft’s near-exclusive hosting arrangement.
This three-way dynamic adds another layer to the circularity problem. Microsoft earns cloud revenue from OpenAI. Amazon now holds a massive equity stake and stands to win cloud contracts. Nvidia supplies the GPUs that both clouds run. Each company’s financial health is partially contingent on the others continuing to spend. For investors trying to assess risk, the question is pointed: do these relationships represent genuine strategic alignment, or is every player propping up every other player in a structure that only works as long as no one stops spending?
OpenAI’s for-profit pivot complicates the picture
There is another variable that has received less attention. OpenAI has been in the process of converting from a capped-profit entity controlled by a nonprofit board to a full for-profit corporation. That transition, which OpenAI announced in late 2025, reshapes what Nvidia’s $30 billion actually buys. Under the old structure, investor returns were capped at 100 times the original investment. Under a for-profit model, Nvidia’s stake could appreciate without limit, but it also means OpenAI faces conventional pressure to deliver returns to shareholders, not just advance AI research.
The conversion has drawn its own scrutiny. California’s attorney general and a group of former OpenAI employees have raised questions about whether the nonprofit’s assets are being fairly valued in the transition. For Nvidia, the structural shift matters because it determines the governance and financial incentives of the company in which it just made its biggest bet.
Regulatory silence, for now
No public antitrust or securities investigation has been announced in connection with Nvidia’s OpenAI investment as of June 2026. But the pattern fits a category regulators have already flagged. In January 2024, the Federal Trade Commission issued 6(b) orders to Alphabet, Amazon, Anthropic, Microsoft, and OpenAI, compelling them to provide information about major AI investments and partnerships. The FTC said at the time that it wanted to understand whether such deals function as de facto acquisitions that sidestep merger review. The agency has not named Nvidia’s OpenAI stake specifically, but the structure, a dominant supplier taking a multibillion-dollar position in a key customer, is precisely the kind of arrangement the FTC said it was scrutinizing.
In the European Union, the European Commission has similarly signaled interest in how large AI investments interact with competition law. Whether either body moves from inquiry to enforcement depends on evidence that is not yet public. But the sheer size of the numbers involved makes prolonged regulatory indifference unlikely.
What the money actually buys
At its core, the transaction is straightforward: Nvidia is paying for guaranteed demand. By investing $30 billion in OpenAI and securing a letter of intent for 10 gigawatts of deployment, Nvidia has effectively pre-sold years of chip production to one of the world’s most compute-hungry organizations. For OpenAI, the deal provides capital and a committed hardware partner at a moment when GPU supply remains tight and competition from Google’s TPUs and AMD’s Instinct MI300X accelerators is intensifying.
The risk is that “guaranteed demand” is only as durable as OpenAI’s business. ChatGPT and its API generate substantial revenue. The Financial Times reported in late 2025 that OpenAI had surpassed $5 billion in annualized sales. But OpenAI has also disclosed significant operating losses, and the gap between revenue and the cost of training and running frontier models remains wide. If that gap does not close, the 10-gigawatt buildout becomes an albatross rather than an asset, and Nvidia’s $30 billion stake becomes a monument to overconfidence in a market that had not yet proven it could pay for itself.
Nvidia’s next earnings will start to answer the question
Nvidia has made the largest single corporate investment in an AI company ever recorded. That investment binds Nvidia’s financial future to OpenAI’s success more tightly than any supplier-customer relationship in recent tech history. Amazon and SoftBank have made similarly enormous commitments, creating a funding structure where the biggest AI companies are simultaneously investors in, customers of, and suppliers to one another.
Whether this is visionary capital allocation or a fragile web of mutual dependency will not be settled by today’s valuations. It will be settled by whether OpenAI and its peers can convert unprecedented compute spending into products that businesses and consumers actually pay for at scale. Nvidia’s next earnings report, expected in late August 2026, will offer the first concrete look at how the company accounts for its OpenAI stake and whether the 10-gigawatt deployment is moving from letter of intent to actual purchase orders. Until then, the $30 billion sits on the books as both Nvidia’s boldest bet and its most exposed position.
