Firmus’ flop is bigger than Oliver Curtis

You would be hard pressed to find any person in Australian business without an opinion on last week's cataclysmic Firmus float.

A lot of the rhetoric has so far centred around the eye-watering $43.7 billion valuation, the tricky escrow terms and the bemusing secrecy around the company’s prospectus that never was. To some, the failed IPO validates pre-existing scepticism of co-founder Oliver Curtis and whether someone who served a year in prison for insider trading (as a 21-year-old) possesses the judgment required to run a public company. His detractors have spent much of this week on an ecstatic victory lap. Australia is a country that leaps on hubris, yet this story is ultimately bigger and more complex than Curtis, the number of open buttons on his shirt or his proclivity for flying private. [[ It’s pretty common for loss-making AI infrastructure companies to fly private. The Roberts brothers, who run Australian neocloud IREN, own a Global Bombardier 7500 and use it regularly to jet between the US and Australia. US neocloud CoreWeave, which is also unprofitable, has another. ]]

There is a far more alarming problem that is beginning to affect AI infrastructure businesses everywhere: US investors are no longer happy to blindly buy into the never-ending data centre hype machine now that the cost of capital is rising and bond yields are at their highest levels in decades. That spells more problems looming for not only Firmus but practically every other Australian company highly leveraged to the AI boom.