Firmus Secures $505M at $5.5B Valuation
Firmus, an Australian AI-focused data-center operator backed by Nvidia, raised US$505 million at a A$5.5 billion valuation ahead of a planned ASX IPO later this year. The round, led by Coatue Management with continued strategic backing from Nvidia, will finance expansion of GPU-heavy facilities, including a US$1.37 billion flagship campus in Launceston, Tasmania designed to host 36,000 Nvidia GB300 chips and leverage hydroelectric power. The capital targets GPU deployment, power and cooling systems, and additional facility build-out to support high-density training and inference workloads in standardized Nvidia-compatible environments.
What happened
Firmus, an Australian company that builds and operates data centers optimized for large-scale AI workloads, secured US$505 million in new funding at a A$5.5 billion valuation. The round was led by Coatue Management and retains Nvidia as both investor and strategic partner. The company is preparing for an initial public offering on the Australian Stock Exchange later this year.
Technical context
Firmus’ product is infrastructural — high-density GPU clusters, high-speed interconnects, storage, and purpose-built power and thermal systems tuned for sustained AI training and inference. Their operating model is managed infrastructure that provides standardized, Nvidia-aligned environments so customers can deploy and scale multi-node workloads without reengineering facility-level systems.
Key details
The new capital will expand data-center capacity, GPU deployment, and the supporting power/cooling infrastructure required for continuous high-compute operation. Firmus is building a A$1.37 billion flagship campus in Launceston, Tasmania, designed with water recycling, battery storage, and hydroelectric power sourcing; the site is slated to host 36,000 Nvidia GB300 chips. SiliconANGLE notes this funding follows roughly A$715 million raised in 2025 (including a A$327 million round) and that Australian Financial Review characterized this as the third and final raise ahead of the IPO.
Why practitioners should care
Firms like Firmus materially change how organizations procure large-scale GPU capacity. By integrating facilities, validated Nvidia hardware stacks, and managed operations, Firmus reduces friction for teams that need sustained, repeatable access to thousands of high-performance GPUs for model training and inference. For ML engineers and infrastructure teams, providers of standardized, turnkey GPU capacity can shift capital planning, procurement timelines, and choices between on-prem, colocation, and cloud-managed options.
What to watch
Track actual GPU availability timelines (especially GB300 deployments), the commercial terms for access (dedicated vs. managed/shared capacity), network/topology details for multi-node training, and the IPO prospectus for unit economics and customer concentration. Also monitor how Firmus’ Tasmania campus impacts regional power sourcing and cost-per-GPU-hour compared with hyperscale cloud providers.
Scoring Rationale
This funding round and Nvidia partnership materially affect GPU capacity availability and commercial options for large-scale model training. It’s highly relevant to practitioners managing compute strategy, though not an algorithmic breakthrough. Freshness is high (one day old), so minor freshness adjustment applied.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
