Firmus adds fresh capital to an AI infrastructure push
Firmus, a Singapore-based AI data center provider backed by Nvidia, has announced a new $505 million funding round led by Coatue, reaching a $5.5 billion post-money valuation. The company says the latest raise brings its total to $1.35 billion over the past six months, a striking pace that places it among the more aggressively financed infrastructure players in the AI market.
The announcement is notable not just for the size of the round, but for what it represents. Firmus is not pitching a general cloud story or a software platform. It is building physical infrastructure for AI computing, centered on a network of energy-efficient data centers in Australia and Tasmania. The project is called Project Southgate, and it is positioned as an “AI factory” network rather than a conventional expansion of warehouse-style server capacity.
That distinction matters because the strongest money in AI continues to move toward the systems that can support large-scale model training and inference. In this case, investors are backing a company whose strategy is explicitly tied to the next wave of Nvidia hardware and to the facilities required to run it.
Project Southgate ties capital markets to next-gen chips
According to the supplied source material, Firmus is using Nvidia’s reference designs for its data center development. The company also plans to deploy Nvidia’s Vera Rubin platform, described as the chipmaker’s next-generation AI computing system that will succeed Blackwell and is expected to ship in the second half of 2026.
That timing links Firmus directly to the next cycle of AI compute deployment. Rather than building generic capacity and deciding on hardware later, the company is aligning site development with a specific computing roadmap. Investors appear to be rewarding that clarity. The new valuation marks a sharp increase from the company’s previous financing, when it raised AU$330 million, or about $215 million, at an AU$1.85 billion valuation, or roughly $1.2 billion, with Nvidia among the investors.
The contrast between those two valuations shows how fast the infrastructure narrative has changed. In a short period, Firmus has gone from a comparatively early-stage operator with a large project ambition to a multibillion-dollar company with enough capital to argue that it can become a meaningful regional AI infrastructure platform.
Source material also notes that Firmus began in cooling technologies for Bitcoin mining before shifting into AI. That background fits a broader pattern in which companies that learned to solve power density, thermal control, and hardware deployment for crypto have found a new market in AI computing. The underlying technical problems are not identical, but both sectors reward operational experience around energy, heat, and specialized equipment.
Why the latest raise stands out
The size and pace of the fundraising are significant on their own. A company adding $1.35 billion in six months is moving with the kind of urgency more often associated with large-scale industrial buildouts than with conventional venture-backed software. It suggests that investors see a limited window to secure strategic positions in AI infrastructure, especially in regions that can support new capacity and where operators can differentiate on energy efficiency.
Firmus is framing its network around that efficiency claim. The supplied source text does not provide detailed engineering metrics, power-use numbers, or capacity targets, so those points remain unquantified here. But even without those specifics, the company’s emphasis is clear: it is trying to present AI data centers not merely as bigger facilities, but as purpose-built systems optimized for the economics of next-generation compute.
That is where the Nvidia relationship carries extra weight. Using Nvidia reference designs can serve two purposes at once. It can accelerate deployment by narrowing design choices, and it can reassure investors and customers that the facilities are being built around an established hardware roadmap rather than around speculative assumptions.
The result is a cleaner commercial story. Firmus is not just promising demand for AI compute. It is linking geography, facility design, capital, and hardware strategy into one package.
Key facts from the announcement
- Firmus announced a $505 million funding round led by Coatue.
- The round values the company at $5.5 billion post-money.
- Firmus says it has raised $1.35 billion in six months.
- The company is based in Singapore.
- It is developing Project Southgate, an energy-efficient AI data center network in Australia and Tasmania.
- Firmus is using Nvidia reference designs.
- The company plans to use Nvidia’s Vera Rubin platform, expected in the second half of 2026.
What this says about the AI market
The Firmus raise reinforces a simple point: AI investment is still flowing heavily into foundational infrastructure, not just into applications. Money is chasing the ability to host and operate compute at scale, and the willingness to support a capital-intensive buildout suggests investors still believe that access to specialized AI infrastructure will remain constrained enough to justify premium valuations.
It also highlights the global shape of that race. Firmus is Singapore-based, building in Australia and Tasmania, and aligning with an American chipmaker’s future platform. That cross-border structure is part of the current AI economy. Capital, hardware roadmaps, and facility development are no longer confined to one national market, especially when projects are designed around regional energy and siting advantages.
At the same time, the announcement says less about current operations than about anticipated positioning. The company is being valued on the strength of its buildout narrative, strategic partnerships, and the expectation that demand for AI infrastructure will continue rising into the Vera Rubin cycle. Whether that valuation holds will depend on execution, but the market signal is already visible: investors are still willing to pay for scarce infrastructure stories with a direct path into AI compute demand.
The bigger takeaway
Firmus’s latest round is a reminder that the AI boom is not only about models and consumer products. It is also about concrete, cooling systems, electrical design, hardware integration, and site development. The companies attracting attention are often the ones that can translate those industrial requirements into a credible growth plan.
In that sense, Firmus is presenting itself as more than a data center operator. It is trying to become a regional platform for AI capacity timed to Nvidia’s next computing generation. Investors have now put real money behind that thesis.
For the broader market, the implication is straightforward. The scramble to secure AI infrastructure is still underway, and firms that can show a tight connection between capital, efficient facility design, and future chip deployment remain in a strong position to attract funding.
This article is based on reporting by TechCrunch. Read the original article.
Originally published on techcrunch.com




