Nvidia offers AI startups compute now, payment later

Instead of just selling chips, Nvidia is offering AI clouds a revenue-sharing and credit-support model built to get GPUs into the hands of companies that could not otherwise afford them.

Nvidia is changing how it gets paid. The company announced on Wednesday a new arrangement in which AI cloud providers can access large volumes of its chips in exchange for a share of the revenue those chips eventually generate, rather than paying the full cost upfront.

The logic, as Nvidia frames it, is a capital problem. Emerging AI companies have historically had limited access to the capital-intensive infrastructure needed to train and run large models, and even long-term customer commitments have often not been enough to unlock financing for compute.

Nvidia’s answer is to let AI clouds buy its hardware and resell Nvidia-powered cloud capacity, with Nvidia collecting standard product revenue on the chips and then a further cut of whatever the cloud earns from renting them out.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!

It is the same compute crunch that has sent valuations soaring for GPU resellers like Runpod, which hit a $1 billion valuation this June renting out chips it does not own.

Two companies are already running on the model. Sharon AI, an Australian AI cloud operator, is deploying up to 40,000 Nvidia Grace Blackwell GB300 GPUs across a six-year, 72-megawatt agreement, a deal its cofounder and chief executive James Manning called “a pivotal moment” for the company’s push into sovereign, large-scale AI compute.

Firmus, the other early partner, is building a much larger campus. The Australian firm is developing a 360-megawatt Nvidia DSX AI factory in Batam, Indonesia, that will eventually house up to 170,000 GPUs across Nvidia’s Grace-Blackwell, Vera-Rubin and Vera platforms.

Bloomberg has reported that Firmus expects between $25 billion and $30 billion in committed offtake agreements over the deal’s first six years, a scale that only makes sense if compute demand from AI-native customers keeps climbing. Nvidia named Baseten, Fireworks AI and Together AI as examples of the customers this is meant to serve.

These are companies that need immediate, elastic access to AI cloud capacity for training, fine-tuning and high-volume inference without committing to years of hardware procurement themselves, a different customer to the hyperscalers Nvidia has courted for a decade.

It is a bet on the long tail of model builders, agent platforms and enterprises that want frontier compute but not the balance-sheet risk of building a data centre.

The arrangement also gives Nvidia something it has not had at this scale before, a recurring, usage-linked income stream layered on top of hardware sales. The model pairs revenue sharing with credit support, effectively helping smaller AI clouds finance the purchase in the first place.

It is not a loan, but it functions like vendor financing with an equity-like upside attached. None of this changes what Nvidia sells, and the chips still cost what they cost.

What changes is who can afford to buy them and on what terms, which matters more than it sounds. Site selection, power procurement, construction and hardware bring-up can take years before a startup ever runs a workload, and Nvidia’s pitch is that AI cloud partners can compress that timeline by selling capacity that already exists.

The company has already committed more than $40 billion to direct AI equity investments this year, spanning OpenAI, Nebius and dozens of smaller rounds. A revenue-sharing compute model does something similar without touching the cap table, keeping the balance sheet exposure with its cloud partners instead of on its own books.

Nvidia has not disclosed how many AI clouds it expects to sign on this basis, or whether the Sharon AI and Firmus terms will be standardised across future partners.

It also deepens a dependency that has already drawn scrutiny, as an increasing share of the AI industry’s growth becomes contractually tied to Nvidia’s own success.

If the model works, more compute reaches more startups faster than the traditional buy-it-outright approach allowed. If AI-native demand cools, Nvidia is now exposed to that slowdown twice, once through chip sales and again through the cloud revenue it has agreed to share.