On June 28, 2026, Australian AI infrastructure company Firmus Technologies said it had struck a strategic partnership with Nvidia to deploy 170,000 GPUs in a new Batam, Indonesia facility starting in 2027. Reuters reports Firmus expects up to $30 billion in revenue over six years from Nvidia‑powered cloud services aimed at emerging AI firms.
This article aggregates reporting from 2 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
This deal is a textbook example of how the race to AGI is now also a race to lock up long‑term compute supply. A 170,000‑GPU build‑out in Batam, coupled with a 360MW data centre, effectively turns Firmus into a regional AI utility for the next wave of model‑builders that can’t get onto the big hyperscalers’ waitlists. For Nvidia, it extends the DSX‑style model of seeding partners with hardware in exchange for revenue share rather than selling chips outright.
Strategically, it pushes serious AI infrastructure into Southeast Asia rather than the usual US–China–EU triangle. That matters for geopolitical resilience and for the emergence of a broader base of labs that can run frontier‑scale workloads without relocating to California or Beijing. It also makes the supply map more complex: by 2028, a non‑trivial slice of the world’s frontier‑grade compute may sit in Indonesian racks controlled by an Australian startup and Nvidia.
For the AGI race, more geographically distributed compute both accelerates progress and broadens the cast of contenders. The risk, as BIS highlights, is that if AI returns disappoint, leverage tied to these mega‑projects could become a vulnerability. In the meantime, Firmus just announced itself as a serious player in the global compute oligopoly.

