On October 9, 2026, Australian AI data centre operator Firmus withdrew its planned A$7 billion initial public offering on the ASX, citing market volatility and valuation concerns. The float, backed by Nvidia and major investors, had been pitched as Australia’s biggest listing since Telstra in 1997.
This article aggregates reporting from 3 news sources. The TL;DR is AI-generated from original reporting. Race to AGI's analysis provides editorial context on implications for AGI development.
Firmus’s pulled IPO is an early sign that public markets are not willing to accept any price for AI infrastructure stories, even ones wrapped in Nvidia branding and “AI factory” narratives. The company was trying to raise A$7 billion on a proposed A$44 billion valuation while still operating only a couple of sites, with most capacity under development. When investors pushed back on the price, and partners like CDC stepped away, the deal collapsed. That does not kill AI data centres, but it punctures the assumption that everything labelled “AI infrastructure” will float effortlessly.
For the AGI race, this is less about one company and more about the cost of capital. Training and serving frontier models depends on large, long-lived infrastructure bets. If public investors start demanding more realistic valuations, more transparent build-out plans and better risk sharing, some aggressive expansion timelines will need to be revised or pushed into private markets. That could slow the most speculative capacity build, but also steer capital toward better-governed, more credible operators.
The episode also underscores the reputational risk attached to AI infrastructure. Firmus has been at the center of local controversies over planning, energy use and water in Australia. As those concerns become more salient to mainstream investors, data centre operators will need to demonstrate not just demand for compute but thoughtful stewardship of the physical systems that support it.



