Firmus Technologies has struck an AI access agreement with Nvidia that will see the Australia-founded company buy Nvidia infrastructure and sell Nvidia-powered cloud services to AI Native customers and others. The deal hands Nvidia product revenue and a share of cloud revenue, while the hardware itself is set to be shipped into Batam, Indonesia, in the first year of the deal and the second year of the deal. The arrangement puts a private gatekeeper between advanced computing power and everyone else. Nice little setup.
Who Gets the Hardware
The contract contemplates delivery of 170,000 GPUs from the first quarter of 2027 to early 2028, with the GPUs to be located in Batam, Indonesia. That means the physical machinery behind advanced AI won’t be sitting in some public commons or open network. It’ll be controlled through a commercial deal, routed through Firmus Technologies, and tied to Nvidia’s revenue stream. The people and smaller firms at the bottom of the stack don’t get direct access. They get access on terms set by the companies that own the gear.
Firmus said the arrangement is meant to make advanced AI infrastructure more accessible to smaller and developing AI firms. That’s the sales pitch, anyway. The actual structure still runs through ownership, contracts, and revenue shares. Access comes packaged as a product, not a commons. The apparatus stays in private hands.
Who Takes the Cut
Nvidia will receive product revenue and a share of cloud revenue from the deal. That’s the core of it. The company that already sits near the top of the AI hardware pile gets paid twice: once for the infrastructure, and again for the cloud services built on top of it. The arrangement turns scarcity into a business model, then calls it access.
Firmus projected up to $30 billion in revenue over the first six years, based on customer commitments. That figure tells its own story. The promise of “access” is still anchored in commitments, revenue projections, and a market that decides who gets to use the machines and who doesn’t. The people doing the actual work of building AI systems remain dependent on a chain of corporate permission.
What They Call Access
The company says the deal is aimed at smaller and developing AI firms. That framing matters because it shows how the language of inclusion gets folded into the same old hierarchy. The infrastructure is not being handed over. It’s being sold, leased, and monetized. The cloud may sound weightless, but the power behind it is very concrete: 170,000 GPUs, a location in Batam, and a contract that sends money upward.
Firmus Technologies is Australia-founded, Nvidia is the hardware giant, and the customers are AI Native firms and others. Those are the players named in the deal. The rest of the world gets the familiar lesson. When access to advanced technology depends on corporate ownership, the gate stays locked even when the marketing says it’s open.
The contract runs from the first quarter of 2027 to early 2028 for delivery, with revenue expectations stretching over the first six years. That’s a long runway for private control dressed up as democratization. The machinery moves. The money moves. The hierarchy stays put.