Nvidia is offering cloud providers sweeping financial guarantees to secure their unused GPU capacity, effectively acting as a central bank for the AI startup sector. The chip giant promises to lease back idle hardware itself if these providers cannot find enough customers, ensuring the expensive equipment gets financed while the data centre remains solvent. This strategy allows Nvidia to take a direct cut of the cloud revenue generated by its partners. A data centre executive noted that previous leasing models left providers struggling to finance the GPUs themselves, but this new approach solves both funding problems simultaneously. The arrangement is designed to reduce heavy reliance on Amazon, Microsoft, and Google, which currently purchase the majority of Nvidia chips while developing their own AI hardware. By diversifying its customer base, the company aims to loosen the grip of these tech giants on its business model.
The move shifts risk away from smaller players and creates a more stable demand stream for Nvidia’s hardware. It also forces competitors to contend with a market where compute costs are subsidised by the chip maker. This dynamic could alter how future cloud infrastructure is priced and built.
- Nvidia takes a percentage of cloud revenue in exchange for financing.
- Idle GPUs are leased back by Nvidia if customers do not materialise.
- Reliance on Amazon, Microsoft, and Google is intentionally being reduced.
