Nvidia’s newly announced revenue-sharing model for AI computing capacity is both a global and domestic problem, that primarily affects new startups and smaller cloud providers that can promise long-term demand for graphics processing units but lack the capital to finance the infrastructure needed to deliver it.
Australia has become one of the largest destinations for AI data centre investment in the world, with data centre and server spending reaching an unprecedented share of the country’s total private investment earlier this year, according to previously reported figures.
Much of that capacity has gone to large, well-capitalised operators and a handful of major local players with balance sheets large enough to secure GPU allocations and finance construction upfront. Smaller Australian AI cloud providers have faced a harder path to the same Nvidia hardware, often competing against buyers able to pay for capacity well before demand is proven.
Nvidia’s new model is structured to reduce that barrier. Under the arrangement, participating AI cloud providers procure Nvidia infrastructure on behalf of startups and enterprise customers. With Nvidia earning a share of the resulting cloud revenue rather than requiring customers to finance hardware independently. In principle, that could allow a smaller Australian provider to access Nvidia’s newest hardware, including GB300 Blackwell-class GPUs, without the upfront capital previously required.
Nvidia’s initial named partners under the model (Sharon AI and Firmus) are both operating at regional and international scale. Firmus’s flagship project under the arrangement is a data centre campus in Batam, Indonesia, expected to scale to 360 megawatts. Nvidia has not named an Australian-based AI cloud provider among its initial partners, and the company’s announcement did not indicate the Australian market as a specific priority for the program.
Whether the model extends to Australian operators remains unconfirmed. Analysts and industry observers have not yet assessed whether the financing structure will be offered more broadly to regional AI cloud providers beyond Nvidia’s initial partners.
Australia’s own AI infrastructure efforts, including guidance from the Australian Signals Directorate on agentic AI security and Bendigo Bank’s push to build what it calls the country’s first agentic security operations centre, depend in part on the availability of competitively priced GPU capacity. A financing model that lowers the barrier to entry for smaller compute providers would be relevant to that broader buildout if it were extended to the Australian market, though no such extension has been announced.
