Report: Nvidia funded startups to buy its own GPUs amid chip glut

Nvidia used its own investment arm to prop up demand for its chips, according to a Wall Street Journal report cited by Wccftech. The newspaper says the company put money into AI startups and small cloud providers that then turned around and bought Nvidia GPUs to build data centers, a pattern that kept the appearance of a tight market even as supply built up behind the scenes.

The Journal’s account centers on an August meeting between then-CEO Jensen Huang and his senior finance team, including chief financial officer and acting CEO Colette Kress. According to the report, Kress argued for pulling back on so-called circular financing arrangements, deals where Nvidia’s own capital flows back to it through customer purchases, and specifically pushed to shrink the company’s $250 billion agreement with OpenAI. The Journal reports that Huang responded by calling Kress a “deer in the headlights.”

The report ties the strategy to a supply problem at the chip-manufacturing level. Nvidia reportedly faced what the Journal describes as an “avalanche” of chips arriving from manufacturing partner TSMC without enough buyers lined up to take them. Investing in smaller cloud providers who would commit to purchasing Nvidia hardware offered one way to absorb that supply, according to the account.

The Journal also reports that Kress raised a separate concern: many of the small-scale cloud providers Nvidia had backed could not actually afford to buy its GPUs on their own. That left Nvidia’s investments at risk of becoming, in effect, money spent to create customers who might not generate a lasting return. The report connects Huang’s push for more investment in these providers to a broader shift among large cloud computing firms, which have been moving away from off-the-shelf Nvidia hardware toward custom-built chips of their own.

What this means

If the Journal’s reporting holds up, it suggests the GPU shortage that has driven prices and allocation decisions across the AI industry was shaped in part by Nvidia’s own financial maneuvering rather than by demand alone. That matters for any company that has been rationing budgets or redesigning products around the assumption that Nvidia chips are scarce because the market is simply that hot.

The report also raises a question the Journal itself does not resolve: how reliable the smaller cloud providers Nvidia funded are as long-term customers, given that Kress is reported to have flagged their limited ability to pay. What is not established in the available reporting is how widespread this financing pattern was across Nvidia’s investment portfolio, or how the company’s leadership has responded since Huang’s departure as CEO.

Nvidia has not been quoted in the available material responding to the specific allegations about engineered scarcity, and the extent of the circular financing arrangements beyond the OpenAI deal remains unclear from the reporting so far.