Firmus Raises $2 Billion for APAC AI Infrastructure
Firmus raised $2 billion in an equity round on August 7, valuing the AI infrastructure company at more than $10.5 billion post-money, Reuters reports. Coatue and Nvidia made follow-on investments, while Blackstone-managed funds and Jane Street participated. Reuters reports that the capital is earmarked for Project Southgate AI training and inference infrastructure in Australia and expansion into other Asia-Pacific markets.
Firmus raised $2 billion in equity financing, taking its post-money valuation above $10.5 billion, according to Reuters. The August 7 round included follow-on investments from Coatue and Nvidia, plus participation from Blackstone-managed funds and Jane Street.
Reuters reports that the valuation is nearly double the $5.5 billion valuation attached to Firmus' April financing. Bloomberg likewise reported the $2 billion round and identified the capital as funding the next phase of the company's Project Southgate AI Factory rollout in Australia and expansion across Asia-Pacific markets.
Capital for AI factory build-out
Firmus develops AI infrastructure using Nvidia's DSX AI Factory Reference Architecture, Reuters reports. The company said the new financing would accelerate Project Southgate's Australian training and inference infrastructure and support expansion into other Asian markets.
"This investment allows us to move on multiple fronts at once," Firmus co-CEO Oliver Curtis told Reuters. "We're scaling across Australia while fast-tracking our capacity to expand into the wider Asia-Pacific region, including the early steps behind our recently announced Indonesia development that will serve AI-native customers."
Capital Brief reported that Blackstone Tactical Opportunities funds and other Blackstone vehicles participated alongside global trading and technology firm Jane Street. Startup Daily reported the round as A$2.85 billion and placed the company's valuation at A$15 billion, the Australian-dollar equivalent of the US$10.5 billion figure reported by Reuters.
Investor appetite and infrastructure constraints
The financing adds another large private-market transaction to the AI infrastructure build-out, where data-center capacity, accelerator supply, power, networking, and deployment expertise can constrain model-training and inference workloads. Reuters quoted Datt Group managing director Emanuel Ajay Datt as saying Firmus' valuation increase demonstrates private capital's view of AI infrastructure as a capital-scarce opportunity.
For ML practitioners, the operational relevance is indirect but material. Companies undertaking comparable regional capacity expansions can increase the availability of GPU-backed cloud services closer to users and data sources, while also raising the importance of evaluating latency, data-residency requirements, interconnects, and service commitments rather than treating accelerator access as a commodity.
Reuters also reported that Firmus and Nvidia reached an agreement in late June under which Firmus would buy Nvidia infrastructure and sell Nvidia-powered cloud services. The reported combination of hardware supply, data-center construction, and cloud distribution illustrates one way these elements can be combined for large AI compute deployments.
Key Points
- 1Firmus secured $2 billion at a $10.5 billion post-money valuation, demonstrating continued investor demand for AI compute infrastructure.
- 2The round funds Project Southgate's Australian training and inference capacity, with Reuters reporting intended expansion across Asia-Pacific markets.
- 3Comparable regional AI factory build-outs make latency, data residency, power availability, and GPU service commitments more consequential for practitioners.
Scoring Rationale
This is a major $2 billion financing for AI infrastructure, with Nvidia and Blackstone among the participating investors and a rapid valuation increase. It matters to ML practitioners because regional training and inference capacity can affect access to GPU-backed compute services, though the immediate product impact remains indirect.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
