Nvidia shifts to revenue-sharing model with Australian cloud provider SharonAI
In brief
- Nvidia supplies hardware and takes revenue share from SharonAI's Australian data center cloud services
- SharonAI deploys 40,000 NVIDIA Grace Blackwell GPUs initially, scaling to 55,000 by mid-2027
- AI Compute Partnership targets regional neocloud providers outside Amazon, Microsoft, Google
- Revenue-sharing model creates predictable cash flows contingent on cloud partners acquiring customers
A New Model for Infrastructure Revenue
Nvidia's revenue from this arrangement comes from two streams: traditional hardware sales, and a recurring cut of cloud service revenue. The partnership, which Nvidia is calling the AI Compute Partnership program, targets what the chipmaker sees as "neocloud" providers—regional or specialized cloud companies that operate outside the shadow of Amazon, Microsoft, and Google.
SharonAI is moving fast. The company plans to deploy up to 72 megawatts of data center capacity in Australia, starting with an initial installation of 40,000 NVIDIA Grace Blackwell GB300 GPUs. The target is to exceed 55,000 total NVIDIA GPUs by mid-2027. To fund the build-out, SharonAI closed a $1.6 billion strategic financing round specifically aimed at supporting the Nvidia collaboration. Stock in SharonAI surged as much as 10% in premarket trading following the announcement.
Why This Matters for Nvidia's Business
The revenue-sharing model addresses a structural challenge in hardware sales. Traditional chip sales are lumpy—big orders, then silence. Recurring revenue from cloud operations smooths cash flow and justifies higher valuation multiples. Revenue-sharing arrangements create the kind of predictable, subscription-like cash flows that justify higher valuation multiples, a dynamic familiar to software companies but less common in semiconductor.
There's a catch. The risk is that Nvidia's income from these deals depends on the cloud provider actually winning customers and generating meaningful utilization of the deployed GPUs. If SharonAI builds the infrastructure but can't fill it with paying workloads, Nvidia's recurring revenue evaporates. The chipmaker is betting that neocloud operators—smaller, more agile than hyperscalers—can carve out profitable niches in regional AI markets.
