October 9, 2026, (Inside AI) — Firmus Technologies pulled its planned A$44 billion ($31 billion) initial public offering on October 9, the morning it was due to publish its prospectus, after institutional investors rejected the A$11 per share valuation. The Australian data centre developer, which had positioned itself as a neocloud operator serving artificial intelligence workloads, cited "recent market volatility and prevailing market conditions" in a statement announcing the withdrawal.
The collapse marks the first major AI infrastructure listing to fail at the pricing stage in 2026, a year that has otherwise seen sustained appetite for data centre and compute assets. It also signals that public market investors are beginning to apply stricter scrutiny to AI-adjacent companies with limited operating histories and shifting business models.
Firmus had sought a valuation of almost A$44 billion, a figure that would have placed it among Australia's largest listed technology companies. The company operates data centres designed for high-density AI training and inference workloads, a segment that has attracted billions in private capital over the past two years. But its transition from a traditional colocation provider to a neocloud specialist raised questions about execution risk, particularly after the loss of a key partner.
According to sources familiar with the offering, prospective shareholders balked at the price given the company's limited track record as a neocloud operator and the departure of a significant commercial partner. The identity of that partner has not been disclosed. Inside AI could not independently verify the circumstances of the partnership termination.
In its statement, Firmus said the terms "would not appropriately reflect the strength of the Company's business and long-term growth outlook." The company added it would pursue "alternative public and private-market options" for raising capital. That language suggests Firmus may return to private investors or explore a trade sale, though no timeline was provided.
Neocloud Economics Face Public Market Test
The withdrawal exposes a widening gap between private and public market valuations for AI infrastructure assets. Private rounds for neocloud providers have routinely priced companies at 20 to 30 times forward revenue, fueled by demand from AI model developers. Public market investors have historically applied lower multiples to data centre operators, typically 10 to 15 times forward revenue, reflecting concerns about capital intensity and customer concentration.
Firmus is not the first AI infrastructure company to encounter this disconnect. Several peers have delayed listings or downsized offerings in 2026 as public investors demanded clearer paths to profitability. The sector's heavy reliance on debt and equity financing to build capacity has made it sensitive to interest rate expectations and credit market conditions.
Market volatility in early October, driven by renewed uncertainty over global growth and central bank policy, provided a convenient backdrop for the withdrawal. But the decision appears to have been driven more by valuation resistance than by broad market conditions. Other listings in the same week proceeded at revised price ranges, suggesting investors remained selective rather than uniformly risk-averse.
The failed IPO also raises questions about the neocloud business model itself. These companies lease compute capacity to AI developers, often on short-term contracts. That structure provides flexibility but limits revenue visibility. Public investors tend to favor longer-term contracts and diversified customer bases, neither of which is typical for early-stage neocloud operators.
Firmus had attempted to address these concerns by emphasizing its long-term growth outlook and the strength of its business. But without a published prospectus, investors had limited information to evaluate those claims. The company's decision to withdraw before releasing financial details may have been intended to preserve optionality for a future private raise.
For the broader AI infrastructure sector, the outcome is a cautionary signal. Companies that have raised capital at elevated private valuations may face difficulty achieving similar marks in public markets. That could slow the pipeline of AI-related IPOs and push more firms toward private credit or strategic partnerships.
Firmus said it would continue to pursue alternative options, but provided no further details. The company's next move will be closely watched by investors and peers alike, as it tests whether the neocloud thesis can attract capital outside the private markets. For now, fear of overpaying has trumped fear of missing out.
