Nvidia deliberately kept GPU prices high through circular financing deals
Nvidia’s alleged use of “circular financing” — investing in or guaranteeing AI startups so they could buy its GPUs — has triggered debate over whether the company artificially sustained high chip prices and demand. Some see the practice as aggressive but routine vendor financing akin to carmakers’ credit arms; others argue it distorts true market demand, entrenches Nvidia’s dominance in AI hardware, and may expose investors and the broader semiconductor supply chain to hidden risk. The thread also touches on related issues like export controls, antitrust scrutiny, and the impact of prolonged GPU scarcity on both cloud providers and consumers.
Allegations About Nvidia’s Financing Practices
- Thread centers on reports that Nvidia invested in or guaranteed financing for AI/data‑center startups, which then used that money to buy Nvidia GPUs, allegedly inflating demand and keeping prices high.
- Some commenters highlight concerns about “circular financing” and possible channel stuffing: Nvidia committed to huge wafer purchases from TSMC, then helped create buyers when hyperscalers began developing their own chips.
- There are mentions of internal tensions: the CFO reportedly opposed these financing deals, while a close relative of the CEO was involved in structuring them, raising perceived nepotism concerns.
Is It Illegal, Unethical, or Just Normal Vendor Financing?
- One camp says this is standard vendor financing used in many capital‑intensive industries (cars, aircraft, industrial equipment) and not inherently circular or fraudulent.
- Others argue that equity-based financing tied to mandatory Nvidia purchases, plus counting that demand as normal revenue, crosses into distortion or at least aggressive financial engineering.
- Some commenters go further, speculating about export‑control evasion and “smuggling” to China, while others explicitly label those claims as conspiratorial or unproven.
Artificial Scarcity vs Genuine Demand
- Some assert Nvidia “engineered scarcity” and forced customers to buy now or lose access to future generations, contributing to unused GPU stockpiles and inflated prices.
- Others counter that GPU rental prices remain very high and hyperscalers still buy all the GPUs they can get; this suggests genuine supply constraints and inelastic demand rather than a glut.
- It is noted that limited TSMC capacity and long fab lead times make over‑ or under‑supply structurally hard to correct quickly.
Anti‑trust, Market Power, and Remedies
- A number of commenters call for investigations or antitrust action, up to breaking up Nvidia or forcing open firmware/microcode.
- Others think the real issue is broader “financialization” of tech and capital markets, not Nvidia alone.
Competition and CUDA Lock‑in
- Discussion notes that Nvidia’s long‑term investment in CUDA and domain‑specific software (across many scientific fields) makes it hard for AMD/Intel or custom chips to displace them quickly, even if hardware competitors exist.