Nvidia's $40 Billion Spending Spree Revives Dot-Com Bubble Fears

Nvidia's AI financing strategy is no longer a side story. The chipmaker is putting real money behind the same companies that buy its chips, and you don't need to be a bubble obsessive to see why Wall Street is watching the loop.

Nvidia has committed more than $40 billion to AI equity deals in 2026, according to a CNBC tally published in May. The largest piece was its $30 billion investment in OpenAI. That investment was part of a funding round that OpenAI said included $50 billion from Amazon and $30 billion from SoftBank. Reuters put the round's valuation at $840 billion including the new capital, while OpenAI later said the raise had grown to $122 billion. That is a lot of money moving into one company. It is also money moving into one of Nvidia's biggest chip buyers.

That is the issue. Nvidia has also put money into CoreWeave, Nebius, Corning and IREN, all tied in different ways to the AI infrastructure stack that depends heavily on Nvidia GPUs. Its latest disclosures added more scale to the picture: Nvidia revealed a SpaceX stake worth about $21 billion as of June, the Financial Times reported, and Tom's Hardware noted that its Intel position had grown to roughly $30 billion. The numbers are too large to treat as venture side bets. This is balance-sheet strategy.

Wedbush Securities analyst Matthew Bryson put the concern plainly in a note cited by CNBC, saying Nvidia's investments fit "squarely into the circular investment theme" worrying investors about AI demand. He also said the same deals could help Nvidia build a "competitive moat" if the company executes well. Both things can be true. A supplier can build a stronger ecosystem and still make it harder for outsiders to tell how much demand would exist without its own capital pushing the system along.

The Dot-Com Comparison Is Not Random

The old example is telecom vendor financing. Between the late 1990s and 2001, equipment makers such as Lucent, Nortel and Cisco lent or promised billions to carriers buying their gear. Lucent's vendor financing portfolio reached $8.1 billion by September 2000, according to Treasury and Risk. TheStreet reported at the time that Nortel had $3.1 billion in commitments, with $1.4 billion drawn, while Cisco had promised $2.4 billion in customer loans.

The carriers used that financing to buy network equipment. The suppliers booked the sales. Growth looked clean until the customers couldn't pay and capital markets shut. Then the same loop that flattered revenue started pulling it down.

Nvidia's deals are not a carbon copy. Most are equity stakes or strategic investments, not classic customer loans, and Nvidia is printing cash in a way Lucent was not when the telecom cycle cracked. Nvidia's data center business is real. It's large, and it's profitable. You can see the chips being used. You can see the demand in cloud spending, and in the compute burned on training and inference.

Still, the mechanism is familiar enough to make investors uncomfortable. When Nvidia funds a customer, and that customer buys Nvidia systems, the two sides of the transaction stop looking fully independent. That's not an accounting scandal by itself. It is a question of quality. You want to know whether revenue is being pulled by end demand or helped along by the seller's own balance sheet.

Nvidia Is Trying To Widen The Circle

The OpenAI deal shows how quickly the structure has changed. In September 2025, OpenAI and Nvidia announced a letter of intent to deploy at least 10 gigawatts of Nvidia systems, with Nvidia intending to invest up to $100 billion as each gigawatt came online. The first phase was targeted for the second half of 2026 on Vera Rubin systems. By February, OpenAI had announced a broader funding package with Amazon, Nvidia and SoftBank, and its own update described 3 gigawatts of dedicated Nvidia inference capacity and 2 gigawatts of training capacity on Vera Rubin systems.

That shift matters. Nvidia isn't just buying orders here - it's trying to make itself the center of the whole AI buildout, financially and technically bound to its own supply chain. A stake in CoreWeave gives it exposure to a GPU cloud provider; a stake in Nebius puts it close to another AI infrastructure buyer. Corning matters because optical fiber and glass sit inside the data center buildout. Intel matters because Nvidia wants custom data center products and x86 systems tied more tightly to its platform.

Frankly, the newest move is the most revealing one. The Financial Times reported in August that Nvidia had partnered with Apollo, Blackstone, BlackRock, Brookfield, Goldman Sachs and KKR on an effort to mobilize more than $500 billion for AI infrastructure financing. That takes the circular-financing worry and tries to answer it with outside capital. If banks and private investors are underwriting more of the buildout, Nvidia can argue the demand is broader than its own checkbook.

That does not settle the matter. It just raises the price of being wrong.

If AI usage keeps growing fast enough, Nvidia's stakes will look like early positions in the companies and assets that needed its chips most. If growth slows, the same network becomes a risk map: customers, suppliers, financing partners and equity holdings all leaning on the same assumption that compute demand keeps compounding. Nvidia has earned the benefit of being taken seriously. It has not earned a free pass from arithmetic.

Also read: Anthropic's Dario Amodei Blames AI Backlash on a Crisis of Trust • Grok 4.6 matches GPT-5.6 Sol on benchmarks while undercutting its own $300 plan • OpenAI's New Ultrafast Mode Runs GPT-5.6 Sol on Cerebras Chips, Not Nvidia

Source