Nvidia is about to start taking money from cloud service providers.
Nvidia is transforming its powerful balance sheet into market leverage, quietly evolving from a chip seller to the "central bank" of the AI compute ecosystem by providing financial backing to emerging cloud providers in exchange for revenue sharing.
On July 1, technology media The Information reported that Nvidia is offering financial guarantee commitments to young cloud providers that rent or sell its GPUs—if these companies fail to find enough AI developers to rent their compute capacity, Nvidia will buy back their unsold GPU capacity at an agreed price.
As a trade-off, Nvidia will take a certain percentage of revenue from these cloud providers, with the share gradually decreasing as the contract progresses. GPU cloud providers Firmus and Sharon AI have already joined the program, and three senior executives with business ties to Nvidia have confirmed the arrangement.
On July 1, Nvidia announced on its official website a new business model combining revenue sharing and credit support, allowing AI cloud providers to procure Nvidia infrastructure without fully bearing upfront capital expenditure, and offer compute services to downstream AI-native companies, model developers, and enterprise clients.
The report said that this project is referred to internally at Nvidia as the "AI Compute Partnership." A Nvidia spokesperson also confirmed the existence of the project. This move marks a major strategic shift for Nvidia:
On one hand, it lowers the financing threshold for emerging cloud providers to expand the customer base; on the other, Nvidia directly participates in profits downstream in the compute market via revenue sharing, further extending its control of the AI industry chain toward downstream sectors.
Model Shift: From Selling Chips to Sharing Cloud Revenue
According to Nvidia's official press release, Nvidia will receive additional revenue sharing from cloud services apart from standard product sales, thus creating a recurring revenue stream linked to usage. The core intention of this model is to break through the financing barrier that has long constrained startup AI companies from acquiring large-scale compute capacity.
Nvidia positions this framework as the "DSX AI Factory" model, targeting AI service scenarios requiring cross-region operations, high utilization rates, and multi-tenant accelerated computing.
Sharon AI and Firmus are the first cloud providers to participate in this model. Sharon AI plans to deploy up to 40,000 Nvidia Grace Blackwell GB300 GPUs; Firmus is building a DSX AI factory park in Batam, Indonesia, expected to scale up to 360 MW and equip up to 170,000 Nvidia GPUs. These two deployments directly demonstrate Nvidia’s latest advances in turning compute demand into deployable, financeable infrastructure.
Nvidia points out that historically, emerging AI companies have faced severe limitations in acquiring capital-intensive infrastructure—even signing long-term commitments has often not been enough to unlock financing for compute procurement. This means many AI-native companies, model developers, and inference providers encounter long waits when expanding compute capacity: site selection, power purchasing, construction, hardware installation—each stage can take months or even longer.
The new model promises: By realigning the economic structure, these groups can obtain full-stack accelerated compute capacity faster without waiting for traditional infrastructure buildout cycles.
Backstop Logic: Solving the Core Challenge of GPU Financing
Reportedly, GPUs are typically the most expensive component in AI data centers. For chip buyers with lower credit ratings, securing sufficient loans is itself a major hurdle.
A data center executive commented that Nvidia’s deals “kill two birds with one stone.” He explained: If Nvidia merely guarantees the lease of data center facilities, "you still face the challenge of 'how to finance GPUs'"; but if Nvidia promises to buy unsold compute within the facility, "the GPU financing problem is solved, as is the data center financing issue."
In other words, Nvidia’s backstop pledge essentially serves as a credit enhancement tool, allowing emerging cloud providers unable to secure bank loans to leverage greater capital and accelerate data center construction.
Strategic Intention: Breaking the Monopoly of Major Clients
Nvidia has clear strategic motives for these measures. Currently, a handful of large cloud providers—Amazon, Microsoft, SpaceX, Oracle, Meta, and Google—buy most of Nvidia’s chip capacity. However, several of these companies are developing competing AI chips in-house, posing a potential threat to Nvidia.
To reduce dependence on such giant customers, Nvidia has in recent years continuously supported a group of emerging GPU cloud providers led by CoreWeave. The “AI Compute Partnership” is a continuation and deepening of this strategy.
According to The Information, Nvidia is also in talks to offer financial guarantees for OpenAI’s large data center lease in Ohio, which, if fully built at current chip, labor, electricity, and material costs, would cost up to $500 billion.
Capital Investment: From Equity Stakes to Capacity Guarantees
Nvidia’s capital investment in this direction has already reached a substantial scale.
To date, Nvidia has invested billions of dollars in several emerging cloud providers for equity, and in some cases agreed to buy back their chips, involving companies like CoreWeave and Lambda, with total deals worth billions. The Information also reported that Nvidia’s own researchers use GPU servers bought back from Lambda.
In terms of capacity guarantees, Nvidia began pursuing such deals last fall. In September 2024, Nvidia pledged that if CoreWeave couldn’t find tenants, it would buy all its unsold capacity through 2032, with a contract value of $6.3 billion. This move eased investors’ concerns about CoreWeave’s highly leveraged business model, sending its stock up nearly 30% in the following week.
According to a regulator filing from Nvidia in May (covering the quarter ending in April), Nvidia has since pledged another $3.5 billion to guarantee customer data center leases in exchange for rights to buy their stock.
Overall, Nvidia is building a multi-layered interest binding mechanism: equity investment, capacity buyback, lease guarantees, and now revenue sharing. Each arrangement further strengthens Nvidia’s financial ties with downstream cloud providers, enabling Nvidia to directly share in the incremental commercialization gains of AI compute beyond chip sales.
Risk Warning and DisclaimerMarkets are risky and investments should be made cautiously. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individuals. Readers should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. All actions based on this article are at your own risk.