When a company pulls an IPO and says the market got its value wrong, treat the claim with some suspicion. Firmus's board says the proposed terms did not reflect its "operational performance or long-term growth prospects." Perhaps so. But the buyers it had spent months courting looked at the same business and declined to pay the asking price. That is information, and it is more useful than anything in the press release.
What Firmus sells is the promise of AI capacity: sheds full of Nvidia chips, hungry for electricity and cooling. Among its prominent backers is also its chip supplier. That circularity, where the company selling the picks and shovels helps fund the prospectors, is one reason sceptics worry about the whole AI buildout. A public listing would have forced Firmus to lay out how much of its future depends on a few customers, a few power contracts and one supplier's goodwill. Fund managers read those pages closely, and they appear to have blinked.
The usual reading is that this is a story about investors. A Left columnist should say it is also a story about disclosure. An IPO is one of the few moments when a company that will consume a large slice of a public grid has to open its books to the public. Going private means raising money from a small circle of institutions under confidentiality. The load on the grid, and on water and land, does not become any more private. The Australian Energy Market Operator has projected that data-centre demand will grow several-fold this decade. Someone pays for the transmission, the firming capacity and the transition delays that come with it. Historically that has been a mix of ratepayers and taxpayers, with the biggest users negotiating hard for the cheapest deal.
The best counterargument is that the market worked. A shaky offering was pulled before retail investors, and the ordinary Australians who might have bought in were spared a possible loss. That is true, and I would rather see a pulled float than a bad one sold to mums and dads. But it is not the whole picture. The "private market" Firmus now turns to is, to a large degree, our superannuation system. Compulsory retirement savings, pooled by industry funds, are among the largest sources of patient capital in Australia, and they are increasingly parked in unlisted infrastructure and technology. Workers will take the risk of this buildout whether or not they ever see a prospectus. What changes is how much they are told, and how much say they have.
That is where the Left's interest should sit. The question is not whether Australia should host AI infrastructure; it may be a decent source of jobs and regional investment, and the renewable-rich states have a real advantage. The question is on what terms. Operators who want cheap power should be signing binding commitments to build or buy new renewable generation rather than bidding it away from households and manufacturers. They should publish their water and energy use. They should pay a fair share of network upgrades. Cheap electricity is a public good, and no private company should treat it as a free input just because it wears the Nvidia badge.
The market's hesitation has handed governments an opening. While the capital is nervous, regulators and state planning authorities have unusual leverage to attach conditions. When the sector's confidence returns, they will not. A pulled IPO should be read as a cue to set the rules for the next wave of data centres, before the money comes back.
How it may affect me
Most people won't see a direct effect, but two channels matter. First, if you hold an industry or retail super fund, part of your balance may already sit in unlisted data-centre and AI-related assets; your fund's annual report or investment-options page should show infrastructure and private-market holdings. Second, rapid data-centre growth can push up network and capacity costs that eventually appear in electricity bills. The size and timing of any such effect is uncertain and depends on regulators' decisions.
