Firmus axes $5.5B IPO after 'greedy' A$44B valuation revolt

By the TechPulse USA News Desk

They didn't cut the price. They didn't delay the listing. Firmus — the Nvidia-backed Australian data centre company that wanted to go public at a A$43.9 billion valuation, the biggest Australian float since Telstra in 1997 — killed the whole thing on Friday morning. One week the books were "indications of interest well above the offer size." The next, the company walked away and blamed "recent market volatility." Read the second sentence again: a A$44 billion company does not need market volatility as an excuse. What it needed was buyers. It didn't have them.

We pulled the Reuters, Nikkei and Straits Times versions of Friday's announcements side by side before writing this — the facts line up, which matters because this story has been spinning for a week and the numbers keep getting rounded in different directions. What follows is what actually happened to Australia's most hyped IPO, who got burned, and what it says about the AI funding boom's new mood.

Photo: Melbourne, where Australian AI data centre operator Firmus runs existing facilities. Its withdrawn IPO proceeds were meant to fund construction of a broader data centre network across Australia and Southeast Asia. (Bob Tan, via Wikimedia Commons, CC BY 4.0)

What actually happened

Firmus announced on Friday, October 9, that it had withdrawn its application to list on the Australian Securities Exchange. The deal had been priced at A$11 a share, aimed at raising up to A$7.9 billion — roughly US$5.5 billion — at a valuation of A$43.9 billion (about US$30.7 billion). Trading was pencilled in for October 23. That date is now fiction.

The official reason, from the company statement: "recent market volatility and prevailing market conditions." The board, it said, had concluded "that the terms on which the Offer could proceed would not appropriately reflect the strength of the Company's business and long-term growth outlook." That is corporate language for a much plainer fact, which investors said out loud all week: not enough big funds were willing to pay A$11 a share.

Because consider the timeline. Days before the withdrawal, the term sheet going around said indicative orders were already sufficient to cover the offering — enough to put the company on track for a roughly US$30 billion valuation. Then Bloomberg reported on October 8 that Firmus had shut its bookbuilding process early as investor support faltered. Twenty-four hours later, the IPO was dead. Something between the indications of interest and the actual commitments did not survive contact with reality.

Firmus says it will now chase private capital instead, pursuing "alternative public and private-market options," in its words. Translation: back to the drawing board, where the pricing is negotiated one fund at a time instead of in front of the whole market.

The "greedy" verdict

The investors who passed on the deal were not shy about why. Ron Shamgar, head of Australian equities at Tamim Asset Management, told Nikkei Asia the company had simply overreached: "Obviously they got a bit too greedy." On the numbers, Shamgar said the valuation looked "quite excessive" next to comparable "neo-cloud" companies — offered at roughly 13 times earnings before interest and taxes on a two-to-three-year outlook, which, he said, "isn't necessarily cheap for essentially what is a startup that has a lot of execution risk."

That's the sentence that kills a A$44 billion valuation. Firmus operates two data centres. The rest of the valuation — the pipeline of facilities across Australia and Southeast Asia, serving customers including OpenAI as its anchor tenant and Meta — is buildings that don't exist yet. The IPO money itself was needed to help fund their construction. Investors were being asked to pay 2029 prices for 2026 infrastructure, and enough of them declined that the book wouldn't close.

The other complaints, relayed by investors to Reuters and the Straits Times, read like a checklist of everything that makes public-market buyers nervous about AI infrastructure right now: a rapidly rising valuation (one analyst noted the company was valued at roughly $1 billion a year ago and came asking the market to believe $44 billion), a hefty debt pile, the execution risk of an ambitious buildout, and the awkward fact that about 58 per cent of the shares would have been free to trade from day one — an open invitation for early holders to flood the market.

The cleanest summary came from Maxence Visseau, Dubai-based chief investment officer at Arkevium Capital, speaking to the Straits Times: "Investors still believe in AI. What they won't do is pay any price for companies that spend huge amounts on data centres, depend on a few big customers and promise profits years from now." Rising borrowing costs and tightening data-centre financing conditions did the rest.

Photo: Nvidia CEO Jensen Huang at Computex Taipei. Nvidia backs Firmus, and its imprimatur did heavy lifting in the IPO's valuation story. (NVIDIA Taiwan, via Wikimedia Commons, CC BY 2.0)

The collateral damage

The blast radius is not limited to Firmus. Maas Group, an Australian construction services provider that pledged an additional A$300 million to Firmus back in August — taking its stake to 3.2 per cent — requested a trading halt on Friday pending what it called a material announcement about its contractual arrangements with the data centre operator.

Maas was already bleeding before the halt. Its shares plunged more than 20 per cent on Thursday, after reports surfaced that Firmus was reconsidering the deal's terms — erasing about A$517 million of market value in a day and leaving the stock down more than 30 per cent from the record high it hit earlier this month. The company said it expected the halt to lift by Monday, once it had updated the market. That update is going to be an interesting read.

Australia's shrinking market takes the hit

Here's the part that should worry people beyond Firmus's shareholder register. Australia's benchmark index is heavily skewed toward the Big Four banks and giant miners; fund managers had genuinely hoped a big tech listing would diversify the market and give investors something new to buy. Instead, the collapse is another entry in a miserable ledger.

Oscar Oberg, lead portfolio manager at Wilson Asset Management — which has been a Firmus investor since last year — did not pretend otherwise. "It's really bad," he told Reuters, on the capital-market implications. And then, with disarming honesty: "We owned shares early, I'm not going to hide behind that. But I just think it would have been good for the market to have something new."

Oberg's broader point lands harder than the Firmus story alone: "The lack of IPOs has been very frustrating for a long, long time for us … that to me is the most disappointing part." The numbers back him up. There were just A$1.37 billion of new share sales in Australia in the first nine months of 2026, according to LSEG data — the most since 2021, but a rounding error next to global rivals. Hong Kong saw 118 companies raise around $50 billion via IPOs in the same period. At A$43.9 billion, the Firmus float would have been Australia's second-largest ever, behind only Telstra's $10 billion privatisation in 1997 — and, per Dealogic, the fourth-largest public offering in the world this year, behind SpaceX, CXMT and Cerebras.

The ASX, already dealing with a shrinking roster of listed companies after years of privatisations and takeovers, just lost its best near-term hope of a flagship tech listing. Investors we saw quoted across Friday's coverage kept coming back to the same word: disappointing. Not surprising — disappointing.

2026's growing IPO casualty list

Firmus is not an isolated misfire. 2026 has been littered with high-profile listings that were abandoned, delayed or reworked as investors demanded valuation discipline — a pattern Reuters rounded up on Friday alongside the Firmus news.

Wall Street brokerage Clear Street withdrew its planned US IPO in February after first delaying the deal and slashing its fundraising target. Smart-ring maker Oura postponed its planned US listing in September, shelving a deal that could have valued it at $15 billion. Nuclear equipment maker Holtec withdrew its IPO in September, and homeowners insurer Bamboo Insurance postponed its own late last month. Add Firmus to the list and the theme is unmistakable: the window is open for plausible valuations and shut for fantasy ones.

Which is the right way to read this for the AI trade. Nobody is saying investors have stopped believing in AI infrastructure — Visseau said it plainly, and the money still flowing into private AI rounds says it too. What's changed is the price of admission to the public markets. As we noted when Firmus's valuation tripled in under two months on the way to this IPO, the "fairytale" math always depended on public investors paying private-market prices. They declined.

It's the same selectivity showing up elsewhere in AI-land this week: OpenAI's revenue came up $20 billion short of what investors had been led to believe, and the Nasdaq noticed. The market is still happy to fund the AI buildout — Musk is still buying hundreds of thousands of Nvidia chips — but the era of "any price for AI exposure" is ending, one failed bookbuild at a time.

Our verdict

Reuters' Breakingviews desk summed the week up as a failed IPO putting "fear before FOMO," and that's exactly right — but only half the story. The other half is that the FOMO was the problem. Firmus's bankers tried to price a company with two operating data centres at 13 times forward EBIT on the strength of buildings it hadn't built, customers it hadn't billed, and a Nvidia halo that turns out to be worth less than the pitch decks assumed. The market's answer was a shrug.

The honest reading: this is the AI funding boom growing up in public. Private investors can keep writing $40 billion checks to AI chip startups on vibes and term sheets. Public investors get quarterly reports, and they're starting to ask what, exactly, they're buying. Firmus's answer — "market volatility" — will fool nobody. The volatility was in the price, and the price was the company's own invention.

Watch what happens next: whether Firmus can raise private capital at anything near the A$11-a-share valuation it just failed to sell publicly. If the private round comes in lower, we'll know the real verdict. If it comes in higher, we'll know the private markets are still living in the fairytale. Either way, the bookbuild told the truth first.

Sources: Reuters (October 9, 2026 — withdrawal, Reuters investor interviews, Oberg quotes, Maas trading halt, 2026 IPO roundup); Nikkei Asia (October 9 — Shamgar "greedy" quote, valuation and raise figures); The Straits Times (October 9 — Visseau quote, OpenAI/Meta customer details, 58% free-float detail); Bloomberg via Startup Fortune (October 8–9 — bookbuild shutdown, US$30B track); Dealogic via Reuters (4th-largest global IPO); LSEG via Reuters (Australia A$1.37B IPO tally).

About the author: TechPulse USA's news desk covers the AI industry across the US, Europe, Canada and Australia — model launches, lab politics, and the numbers behind the hype.