Firmus Technologies just locked in OpenAI as an anchor customer for two new AI compute facilities in Malaysia, pushing the Australian infrastructure company’s total contracted capacity past 900 MW.
The multi-year agreement marks Firmus’ expansion into a fourth country, adding Malaysia to a portfolio that already spans Australia, Singapore, and Indonesia.
The scale of the buildout
Firmus currently operates two AI factories, one in Australia and one in Singapore. Five more are under development, with completion targets spread across the next 24 months.
The company’s latest funding round valued it above $10.5 billion. The new Malaysian facilities will deploy Nvidia’s next-generation Vera Rubin NVL72 systems and DSX platform. Firmus will pair those chips with its proprietary HyperCube liquid-cooling technology, designed to keep power consumption manageable as compute density ratchets higher.
Meanwhile, the company is also building a 360 MW campus in Batam, Indonesia, targeting a deployment of roughly 170,000 Nvidia AI accelerators at that single site.
Why Southeast Asia matters for AI compute
Malaysia in particular has become a magnet for hyperscale data center projects. The country’s combination of affordable electricity, political stability, and proximity to major undersea cable routes makes it attractive for companies that need to move enormous amounts of data quickly and cheaply.
The IPO question
Firmus is reportedly eyeing a potential IPO in late 2026, which could rank among Australia’s largest public offerings if it moves forward. The OpenAI partnership adds credibility to the IPO narrative, signaling revenue visibility and long-term demand.