Firmus pulls $11-a-share ASX float as AI datacentre hype fails investors

Firmus Technologies has scrapped an $11-a-share listing that would have been the biggest debut on the Australian Securities Exchange since Telstra floated in 1997. The board pulled the offer after investor demand for its AI datacentre business failed to materialise.

The decision lands at an awkward moment for the AI infrastructure trade. Datacentre operators have spent the past two years pitching themselves to public-market investors as the picks-and-shovels play on the AI boom. Firmus is now a concrete data point that the pitch doesn't automatically sell.

What exactly happened?

Firmus had been preparing what Australian media described as the country's biggest company listing in decades. The offer price was set at $11 a share. That structure would have placed the debut alongside — and by some measures ahead of — Telstra's 1997 privatisation float, long the benchmark for ASX debuts.

The float is off. A Firmus spokesperson said the board decided that proceeding with the offer was no longer in the best interests of the company and its shareholders.

That wording matters. It's the standard formulation companies use when an IPO is withdrawn rather than merely postponed, and it puts the onus on market conditions rather than on any disclosed problem inside the business.

The simple takeaway: a company positioned as a flagship AI infrastructure listing couldn't find enough buyers at the price it wanted.

Why does the Telstra comparison matter?

Telstra's 1997 float remains the reference point for scale in Australian equity markets. Matching or beating it signals a listing of national market significance, the kind that pulls in retail investors and institutional money alike.

Firmus was lined up to be exactly that. The comparison tells you how large this deal was intended to be — and how much capital the AI datacentre narrative was expected to raise on the ASX.

When a float of that scale collapses at the pricing stage, it sends a signal to every other infrastructure operator queueing for a local listing. Investors, not issuers, set the ceiling.

What does weak demand tell us about AI datacentre bets?

The scrapped float is one deal, and one deal doesn't settle a debate. But the mechanism here is worth spelling out.

An IPO only works if institutions commit to buy the shares at the offer price before the company starts trading. If demand "fails to materialise" — the phrase used in reporting on the withdrawal — the underwriters can't guarantee the price. The issuer then chooses between cutting the price, shrinking the deal, or walking away.

Firmus walked away at $11 a share. That choice implies the gap between what the company believed it was worth and what the market would pay was too wide to bridge.

For AI infrastructure specifically, this is the tension to watch. The sector's pitch depends on sustained, large-scale spending from AI companies that themselves have not yet demonstrated durable profits. Public-market investors appear, in this case, to have looked at that chain of assumptions and declined to underwrite it at the asking price.

What comes next for Firmus?

The company hasn't announced a replacement plan. The spokesperson's statement — that proceeding was no longer in the best interests of the company and its shareholders — leaves the door open to a revived float, a private raise, or a sale, but commits to none of them.

What's on the record is narrow: the offer is withdrawn, the board made the call, and the $11-a-share price is dead for now. Everything beyond that is speculation.

The plain fact is that Australia's most ambitious AI infrastructure listing couldn't clear the market — and the next datacentre operator eyeing the ASX now knows the price of overreach.