Capital Shifts Infrastructure
Firmus wrapped up a $2 billion equity round this week, sending its post-money valuation past $10.5 billion, nearly double the $5.5 billion the Australia-based company was valued at during its earlier round back in April. That is a sharp jump considering the last raise was only 4 months ago, and it says a lot about how fast money is moving toward AI infrastructure right now. Nvidia and Coatue Management both came back for another round, which tells you these investors are not just testing the waters; they are doubling down. The deal fits into a wider pattern where capital is increasingly chasing the physical backbone of AI instead of the models sitting on top of it.
Compute Capacity Accelerates
The company said the proceeds will let it accelerate the next phase of Project Southgate, its initiative to build out AI training and inference infrastructure across Australia, while it works on expanding into other Asian markets. As more companies build AI directly into how they operate day to day, this kind of physical groundwork, training and inference capacity, data centres, and reliable power is starting to matter just as much as the software running on top of it. With compute demand still running well ahead of what is actually available worldwide, rounds like this show just how urgent the race to build capacity has become.
Infrastructure Becomes Strategic
Nvidia's decision to invest again in this round makes a lot more sense once you look at the relationship between the two companies. Firmus builds its data centres on Nvidia's DSX AI Factory Reference Architecture, and the two firms struck a deal back in late June under which Firmus buys Nvidia infrastructure directly to power the AI factories it is racing to build.
So this is not just Nvidia backing a customer; it is backing the very infrastructure running on its own chips. That shift reflects a bigger change in how AI competitiveness gets defined; it is no longer only about who has the smartest model; it comes down to who actually controls the compute, energy, and data centre space needed to keep everything running. That is likely to squeeze enterprises and cloud providers, who may find themselves fighting harder for AI-ready infrastructure, especially in places where power and space are already in short supply. Governments will need to pay attention too, since energy planning and infrastructure policy may need to move faster than they are used to.
Investment Thesis Evolves
Firmus nearly doubling its valuation in under 4 months, from $5.5 billion in April to over $10.5 billion now, says a lot about where investor confidence is headed. Backing Project Southgate and the broader DSX buildout is not being treated as a background bet anymore; it is becoming a strategic asset in its own right. As the AI industry keeps maturing, the companies quietly building data centres and locking down compute capacity could end up shaping the space just as much as the ones building the actual AI models. Firmus' quick climb offers an early look at how that shift is already taking shape.
The race for AI leadership is increasingly becoming a race to build and finance the infrastructure behind it.