Report Says Nvidia Funded AI Startups That Then Bought Its GPUs

A new report from The Wall Street Journal, relayed by Wccftech and TechPowerUp, alleges that Nvidia helped inflate the current shortage of GPUs and memory. According to the report, the company invested in AI startups that went on to buy Nvidia hardware and build data centers around it. The account points to an August meeting between Jensen Huang and his senior finance team, and it lands just after Huang's departure as CEO was announced.

The allegations are reported claims, not findings. Nvidia has not been shown to have broken any rule, and the story rests on how investments and purchases were allegedly connected.

What is being alleged

The core accusation is circular demand. A chipmaker puts money into young AI companies, those companies use the capital to buy the chipmaker's accelerators, and the resulting orders make demand look larger and more urgent than it might otherwise be. Added to an already tight supply of GPUs and memory, that could push prices up and leave other buyers waiting longer.

The pattern is not new in the industry. Vendor financing, where a supplier helps fund its own customers, has a long history in telecom and enterprise hardware. It is legal and often defended as a way to grow a market. Critics see the risk differently: when the supplier's money becomes the customer's purchasing power, real end demand gets harder to measure, and shortages can look worse than they are.

Why the timing matters

The shortage is not abstract for developers and PC builders. Memory has become a bottleneck as AI data centers soak up DRAM and high-bandwidth memory. Dell's CEO has said the shortage will get worse in 2027. Reports of price hikes across GeForce RTX 50-series cards, production ending for several models, and speculation about very high prices for top-end cards have made the squeeze visible at retail.

If the allegation holds up, it would add a new angle to that story. Part of the scarcity would then be a product of strategy rather than just a mismatch between factory output and AI appetite. That is a significant difference for regulators, customers and investors, even if the physical constraints on memory and advanced packaging remain real either way.

The open-source angle

For open-source developers, a GPU market shaped by a single dominant supplier has long been a concern. Training and inference tooling is deeply tied to CUDA, which makes switching costly. Competing efforts such as AMD's ROCm, Intel's oneAPI and compiler-level projects like OpenAI's Triton and the MLIR ecosystem aim to loosen that grip, but they depend on hardware being available and affordable to the people writing the code.

Scarcity favors incumbents. When accelerators are hard to get, well-funded labs and startups backed by the supplier itself get first access, while independent researchers and small open-source teams are pushed toward rented cloud capacity or older hardware. A market where financing and supply allocation are entangled makes that imbalance harder to correct.

What to watch next

Several things will decide whether this story grows:

• Whether Nvidia responds in detail to the report and discloses how its startup investments relate to hardware purchases.
• Whether regulators or investors ask for more transparency about circular deals across the AI sector, not only at Nvidia.
• How the leadership transition affects the company's investment and allocation policies.
• Whether other chip and memory vendors face similar scrutiny, since several have taken equity stakes in customers.

For now, the report is an allegation built on internal discussions. What it does is sharpen a question developers have been asking for a while: how much of the current shortage comes from physical limits, and how much from how the money flows around them. The answer will shape GPU prices, cloud costs and the room left for open alternatives over the next few years.