Firmus is seeking $5bn in its Australian float after doubling its valuation to $10.5bn

Firmus is reportedly seeking to raise up to $5bn through an Australian Securities Exchange listing, potentially making it one of the exchange's largest-ever floats. The company's targeted valuation rose from $5.5bn in April, alongside a $2bn fundraising plan, to more than $10.5bn by August. Investor meetings have begun, signaling advancing IPO preparations.

Analysis

A materially upsized AI-infrastructure float would be a liquidity event for Australian technology risk rather than proof of durable earnings power. The key question is whether Firmus is being valued on contracted, financed compute capacity or on headline demand; the latter deserves a steep discount given hyperscaler pricing power, rapid GPU depreciation, and the risk that customer commitments are cancellable. The valuation step-up over a few months raises late-cycle private-market marking risk and increases the probability of an aggressive deal structure, including cornerstone allocations and limited free float.

Near term, a successful bookbuild could pull domestic growth capital toward data-centre, power and AI-adjacent names, temporarily supporting the S&P/ASX All Technology Index (XTX) and Global X Semiconductor ETF (SEMI.AX). The second-order constraint is electricity: incremental compute deployment raises demand for grid access, firm generation and transmission, potentially benefiting Australian utilities and infrastructure suppliers more reliably than the compute operator itself. Conversely, a weak aftermarket performance would reset private AI-infrastructure valuations and pressure listed proxies whose multiples already assume sustained capacity scarcity.

The contrarian view is that public investors may demand proof that AI compute is a utility business, not a software business: high capital intensity, refinancing needs, hardware obsolescence and customer concentration can cap returns despite strong top-line growth. This is not yet a directional trade without disclosed revenue, EBITDA/FCF, debt commitments, customer terms, GPU purchase obligations, and the final valuation. Treat the prospectus as the catalyst; contracted-capacity coverage and leverage are the decisive variables over the next 1-3 months.

Key Decisions for Investors

• No pre-IPO position: place an alert for the prospectus and assess participation only if enterprise value is supported by multi-year contracted revenue, customer-grade disclosures and conservative debt; avoid if valuation relies on capacity projections rather than signed take-or-pay commitments.
• If the deal prices at a premium to global data-centre peers without positive free-cash-flow visibility, consider a post-lockup relative-value short versus long SEMI.AX or XTX, with a 3-6 month horizon. Falsifier: disclosed contracted EBITDA and financing terms demonstrate materially faster deleveraging than peers.
• Watch Australian power and grid-exposure names rather than treating the IPO as a pure AI beta signal; initiate only after identifiable power-procurement commitments are disclosed. The investable spillover requires incremental load to translate into contracted generation or transmission spend, not merely aspirational campus plans.
• For broader private-market exposure, reduce valuation sensitivity in unprofitable AI-infrastructure holdings if the offering trades below issue in its first 20 sessions; that outcome would be a cleaner read-through to funding conditions than the initial headline valuation.