When UniSuper's chief investment officer used a Friday note to explain why the fund would not buy Firmus Technologies' abandoned float, he did more than pass on a hot deal. He drew a line that every default super fund should copy: if you cannot audit the cash flows behind an AI infrastructure story, you do not belong in members' retirement money.
Members pay the tuition when FOMO wins
Firmus sought a valuation above $43 billion with an $11 entry price, then reportedly cut toward $8.25 before withdrawing entirely. Nvidia's logo on the cover was meant to signal strategic legitimacy, but it could not answer questions about debt-funded data-centre builds, customer concentration, or the circular financing the Reserve Bank now warns could puncture global stability. Retail investors who queued through brokers were spared only because institutions refused to clear the book.
Super funds are not venture capital partnerships. They owe millions of Australians a liquidity and diversification mandate that conflicts with locking capital into pre-profit hyperscale bets. When CIOs stay silent during hype cycles, they implicitly invite politicians to treat super balances as national industrial policy piggy banks.
The RBA already wrote the risk chapter
This week's Financial Stability Review explicitly linked AI sentiment shifts to potential disorder in bond and equity markets. Super trustees who ignore that paragraph are breaching their own risk frameworks. Sitting out Firmus is not anti-innovation; it is recognition that innovation can be financed elsewhere without forcing nurses and teachers to underwrite Oliver Curtis's Nasdaq backup plan.
What trustees should do next
Publish refusal criteria before the next mega-deal launches. Disclose how much private-market AI exposure you already carry through unlisted infrastructure. And when issuers withdraw floats, ask ASIC to publish anonymised feedback so retail investors learn why the book failed.
Firmus's flop is embarrassing for bankers, not for members whose funds declined to play. That outcome should be celebrated, then codified—so the next Telstra-scale pitch meets a super sector that says no early and in public.
Transparency beats theatre
Trustees who stayed quiet during the Firmus marketing blitz should publish post-mortems explaining how they scored the deal against their own unlisted tech holdings. Members deserve to know whether any default fund held indirect exposure through private-market sleeves that will now mark down. Silence invites the conclusion that super is either reckless or opaque—neither is acceptable when compulsory balances exceed $3 trillion.
Politicians tempted to cheer domestic AI champions should instead fund CSIRO and university commercialisation paths that do not require listing at Telstra-scale valuations on day one. The Firmus withdrawal is a market outcome, not a national failure—but it is also a warning that retirement money is not venture capital unless members opt in with eyes open.
APRA should ask whether any fund's investment governance policies still allow IPO allocations without independent valuation committees. If the answer is yes, members should demand reform at annual member meetings rather than waiting for the next AI poster child to fail.
Industry funds that did participate in earlier private Firmus rounds must now mark those holdings against public comparables that no longer exist, a valuation exercise trustees should explain in annual statements.