Firmus Technologies withdrew its planned Australian float on Friday after investors balked at its proposed valuation and share price, as well as uncertainty about the company and the AI sector. The company had aimed to raise about $7 billion from public-market investors, but couldn’t persuade them to accept its terms. The planned listing put its market capitalisation close to $43.7 billion.
A public float meets the market’s limits
The offering could have been the second-biggest stock float in Australian history and the biggest initial public offering since Telstra listed in 1997. Firmus planned to use the proceeds, then borrow more, to build data centres in Australia, Singapore, Malaysia and Indonesia. That expansion depended on access to vast amounts of capital, and investors got to decide whether the proposed price made sense.
Oliver Curtis, Tim Rosenfield and Jonathan Levee founded Firmus in 2019. By Wednesday night, reports said bankers were considering cutting the share price and valuation towards $30 billion. The reported offer price fell from $11 to $8.25 by Thursday. By Friday, the float was off.
The company had previously said it secured $US2 billion ($2.86 billion) in investor commitments, including from Nvidia and Blackstone. Nvidia already held a 7.2 per cent stake and had reportedly provided funding and sold Firmus chips. That backing didn’t settle questions about the offer price, the limited detail in the prospectus or whether rapid growth in AI and data centres would last.
On Thursday morning, as Firmus tried to save the offering, it withdrew from an appearance at a federal parliamentary inquiry into AI. The company’s founders are seeking further private funding, and Firmus plans to attempt a Nasdaq listing next year. For now, the public market has withheld the approval the company needed for its chosen route to expansion.
Who controls the chips—and the price
Morningstar senior market strategist Lochlan Holloway said Firmus was being valued similarly to Woolworths. “The economics look good now and the growth is very fast now, but what that looks like again in five or 10 years is the big question mark,” he said. Investors’ doubts weren’t limited to the company’s pitch: commentators also cited fears of an AI bubble and concerns among politicians and the public about data-centre growth.
Philip Wohl of Reliance Investment Research said: “Firmus ran into a wall of AI anxiety right at the time that they had planned to list.” He described the company’s claim that market volatility prevented it from listing at its desired price as “a euphemistic way of describing exactly what happened”. He added that investors faced greater uncertainty around AI hardware and that Firmus “really didn't see the demand” at the pricing it targeted.
Ten Cap Investment co-founder and lead portfolio manager Jun Bei Liu called the IPO overpriced. “The valuation for the company was way too high to start with,” she said. “It is almost three times what the valuation was when they raised money just a few months ago.” Liu said Firmus may need listed shares because “it's a very capital-intensive business.”
Nvidia’s role also shows how much leverage sits with companies controlling scarce technology. Holloway said Firmus’s data centres could put Nvidia chips to use, and customers such as Meta and OpenAi may rely on companies such as Firmus to access them. “There is a very, very short supply of Nvidia's high-quality chips and a lot of demand for AI model training,” he said. “If you had access to those chips, which Firmus did with its partnership with Nvidia, you can basically name your price.”
Infrastructure, private money and another attempt
Investors questioned whether Australia's energy infrastructure could support data centres. Firmus’s website said it had two operational sites in Australia and Singapore and five sites “under development”. Floodline Research founder Anna Wu said the offer lacked detail and that Curtis’s criminal past added ambiguity. Curtis served a year in jail in 2016-17 for insider trading before co-founding Firmus as a bitcoin mining company in 2019.
Wu said the Nasdaq was a bigger, more accepting market for emerging technology, though tougher competition could make it harder for Firmus to attract the attention and valuation of US peers. The Nasdaq had recently reached record highs and was up 43 per cent in the past year. The ASX All Technology Index, which tracks 42 technology-linked stocks, had fallen almost 18 per cent that year, including 9 per cent in September alone. Australian technology listings lean more heavily towards software companies, which investors had sold off during the so-called “SaaSpocalypse” as they worried AI could undercut those companies’ business models.
Nvidia was contacted for comment after the failed float. Wu said Firmus was likely to seek a larger investment from the company, describing Nvidia’s global investment activity as “going around the world buying anything that it thinks is going to take over in the local market”. She called Nvidia’s long-term partnership with Firmus “an endorsement ... but also, it's a move towards expanding in the Australia-Pacific region and trying to integrate with the local infrastructure provider.”
Holloway said uncertainty had led more Australian companies to delay listings and that ASX IPO numbers had fallen considerably in recent years. With surplus private cash available, companies could remain in private hands instead of floating shares publicly; analysts expected delayed or withdrawn IPOs to become more common. Wohl summed up the pitch investors rejected: “There were just too many unknowns here and it required too much optimism for investors to take that leap of faith with Firmus.”