Nvidia on July 2 announced an optional financing vehicle that lets the chipmaker collect hardware revenue and then take a percentage of the cloud sales that hardware generates, trading token credits to cash-strapped AI companies in exchange for a slice of their future earnings. The first named partners are Australia's Sharon AI and Singapore-based Firmus Technologies, both building large GPU clusters under the deal.
How the revenue-sharing and credit-support model works
Participating AI clouds buy Nvidia infrastructure and sell Nvidia-powered cloud services to end customers. Nvidia pockets its usual product margin on the hardware, plus a recurring, usage-linked cut of whatever cloud income that gear earns. Developers get token credits to fund compute access now, but they give up an undisclosed portion of future sales in return.
The structure targets a financing gap Nvidia identified: even signed, long-term commitments often fail to convince lenders to fund large-scale deployments, leaving smaller clouds unable to borrow against demand they have already generated. Neither Nvidia nor the partners have disclosed the revenue-split percentages.
Sharon AI and Firmus commit capacity
Sharon AI disclosed in an 8-K filing that the agreement runs six years and covers 72 MW of new Australian data center capacity designed to Nvidia's DSX AI factory spec, scaling to as many as 40,000 Grace Blackwell GB300 GPUs. The Nasdaq-listed neocloud also holds a separate revenue-share facility of up to $200 million with investor Digital Alpha, meaning portions of its income are now pledged to two different parties.
Firmus is building a DSX-aligned campus in Batam, Indonesia, expected to reach 360 MW and house up to 170,000 Nvidia GPUs. The scale signals that both partners are betting heavily on the model to expand supply without purely debt-driven financing.
From circular investments to royalties
Over the past year, Nvidia poured cash directly into its own customers-including a $30 billion stake in OpenAI's $110 billion funding round and backing for xAI's Colossus 2 financing-moves that drew circular-financing criticism. The new model flips the script: instead of investing capital that boomerangs back as GPU orders, Nvidia extends credit support and collects a royalty on partners' sales for years.
That royalty ties a slice of Nvidia's income to utilization rather than hardware shipments alone. If partner clouds can't keep racks rented, the usage-linked stream shrinks. The risk is real given the depreciation pressure already bearing down on operators who pay off hardware that Nvidia refreshes roughly every year.
Why this matters for Sales
For sales teams inside AI cloud providers, the structure changes unit economics permanently. Every deal must now account for a Nvidia revenue share layered on top of infrastructure costs, which can compress margin or force higher prices for end customers. Sellers at startups eyeing token credits gain a path to compute without upfront capital, but they hand over future revenue that can complicate long-term forecasting and investor pitches.
Sales leaders negotiating large cloud contracts should ask potential partners point-blank whether Nvidia's royalty applies, because it can shift capacity availability and cost projections. And if other hardware suppliers follow Nvidia's lead with similar revenue-share models, vendor comparisons will need new line items-making cost-of-sale conversations more complex than a simple per-GPU rate.
Your membership also unlocks:
