Nvidia-backed Firmus shelves its Australia IPO amid weak demand

By Splitfeed EditorialLast updated Oct 8, 2026, 22:30 UTC

On October 8, 2026, Firmus, an Australian AI data centre operator backed by Nvidia, closed its IPO bookbuilding after investor demand fell well short of what the company had sought. The listing had been described as a mega IPO on the Australian market, but support cratered during the bookbuilding process and the company ultimately shelved the offering entirely. No revised terms or postponed timeline were announced alongside the withdrawal.

The collapse matters beyond the immediate disappointment for Firmus because it tests one of the core assumptions driving AI investment cycles: that demand for physical infrastructure, the data centres, power supplies and networking buildout that underpin large-scale AI systems, represents a safer and more concrete bet than software-layer AI companies. Firmus carried Nvidia's backing, which had been a meaningful signal of credibility in recent capital markets, yet even that association was not enough to fill the book. The episode raises a genuine question about whether institutional investors believe current AI infrastructure valuations reflect realistic near-term returns, or whether appetite has begun to cool after several years of aggressive commitment.

None of the three camps had published reactions by the time this story was filed, so their likely positions can only be anticipated. The Pro-AI camp would typically argue that a single failed IPO reflects the idiosyncrasies of Australian capital markets and the specific sizing of this deal rather than any broader retreat from AI infrastructure, and that well-structured offerings with clearer revenue visibility will continue to attract capital. The Anti-AI camp would likely treat the cratering demand as confirmation that the AI investment cycle is entering a correction, with institutional money finally pricing in the gap between infrastructure spending and demonstrated returns. The Middle Ground camp would probably frame the failure as a useful market signal that separates genuine demand from hype-driven appetite, and argue that disciplined capital allocation is a healthier long-term outcome than an oversubscribed listing built on momentum.

The clearest indicator of which reading is correct would be whether Firmus attempts to refile at a lower valuation in the coming months, and how comparable AI infrastructure listings in larger markets, particularly the United States and Southeast Asia, perform before the end of 2026.