Firmus scraps $44bn ASX float as investor demand evaporates for AI datacentre venture

Bankers had vastly overestimated demand for a startup with only two operational sites
Firmus faced mounting skepticism over its $44 billion valuation despite its limited operational footprint and early-stage status.
Mark

So Firmus was going to be the biggest IPO in Australia since Telstra in 1997, and they just walked away from it. What actually happened?

Mimi

The bankers overestimated demand. They thought they could raise $7 billion at $11 a share, but when they started taking orders from institutional investors midweek, the interest wasn't there. So instead of going ahead with a smaller float, the board decided to pull the whole thing.

Luke

But why did demand evaporate? Was it just market conditions, or was there something specific about Firmus that spooked investors?

Mimi

A lot of it was the valuation relative to where the company actually is. They're valued at $44 billion but they only have two small operational datacentres. They're a startup with big forecasts.

Mark

And they had backing from Nvidia and major Wall Street firms. Doesn't that count for something?

Mimi

It does, but it doesn't guarantee retail or institutional investors will bite. Those backers believed in the vision, but the broader market wasn't convinced the numbers added up.

Luke

There's also the CDC deal falling apart. That's a separate story, but it signals something about Firmus's strategy, doesn't it?

Mimi

Yes. CDC walked away because Firmus's business model shifted. CDC is about critical infrastructure and resilience. Firmus became more commercial and flexible. They couldn't align.

Mark

So now Firmus has to raise money from private investors instead. Is that actually harder, or just different?

Luke

It's definitely harder in one sense—they can't tap the public markets, so their options are narrower. But private investors might be more patient with a startup's trajectory than public market investors would be.

Mimi

The founders also lose out significantly. Oliver Curtis, Tim Rosenfield, and Jonathan Levee were expecting much larger paper wealth from a public listing. That's gone now.

Mark

And what about the investors who were already in? The ones who were hoping to use the IPO as an exit?

Luke

That's the real question. Early investors can't sell into the float now. They're stuck waiting for another path—another private round, an acquisition, or eventually a different public listing. Maas Group's stock dropped 20 percent just on the news.

  • A $7 billion ASX listing scheduled for October 23 was pulled after bankers discovered they had dramatically overestimated institutional demand, leaving the $11-per-share price indefensible.
  • Skepticism had been building for weeks over whether a startup with only two operational datacentre sites could credibly carry a $44 billion valuation into public markets.
  • The collapse sent immediate shockwaves outward — early investor Maas Group watched its share price fall more than 20 percent in a single day as the market recalibrated.
  • A separate $73 billion joint venture with veteran datacentre operator CDC, called Project Southgate, also unraveled as the two companies' philosophies on commercial flexibility versus critical infrastructure resilience proved irreconcilable.
  • Firmus must now pursue private capital to fund its expansion, operating without the public scrutiny or market validation a listing would have imposed — while its founders and early investors absorb substantially reduced financial prospects.

In the arc of technological ambition, few moments are as clarifying as when the market refuses to believe a story it has been told. Firmus Technologies, a young Australian AI datacentre company backed by Nvidia and major Wall Street institutions, abandoned what would have been the country's largest stock market debut in nearly thirty years after institutional investors declined to validate its $44 billion valuation — a figure resting on two small operational sites and a vision of sprawling liquid-cooled infrastructure across Australia and Asia. The collapse, announced in October 2026, reveals the widening tension between the pace at which AI ambition is being priced and the slower, more demanding rhythm at which trust is actually earned.

Firmus Technologies abandoned its planned debut on the Australian Securities Exchange, walking away from what would have been the country's largest IPO since 1997. The company, backed by Nvidia and prominent Wall Street firms including Blackstone, Jane Street, and Coatue, had been preparing to raise $7 billion at a $44 billion valuation, with listing scheduled for October 23. By midweek, it was clear that institutional investors were not buying the story. Bankers had badly misjudged demand, urgent discussions about cutting the share price followed, and the board ultimately decided the float no longer served the company or its shareholders. A private funding search was announced in its place.

The reversal was striking for a company that had positioned itself as a transformational force in Australia's AI infrastructure landscape. Firmus had persuaded some of the world's most sophisticated investors that it could build a vast network of liquid-cooled AI datacentres across Australia and Asia — despite operating only two small sites and carrying no meaningful track record. As the listing approached, that confidence gave way to hard questions about the gap between its forecasts and its reality. The company's founders, including Oliver Curtis, Tim Rosenfield, and Jonathan Levee, faced a significant reduction in their anticipated wealth.

The damage spread quickly. Maas Group, an early Firmus investor, saw its shares fall more than 20 percent as markets absorbed the news. Concerns had circulated that existing investors were relying on the public float as their exit, potentially leaving retail investors exposed — though in the end, public enthusiasm never materialised at all.

A further blow came from the collapse of Project Southgate, a $73 billion joint venture with CDC, a veteran datacentre operator, designed to build sovereign, renewable-powered AI infrastructure with Nvidia as anchor customer. The partnership fell apart as the two companies' visions diverged. CDC's chief strategy officer told a parliamentary committee that when Firmus had been focused on sovereign Australian AI capability, the alignment made sense — but as Firmus shifted toward a more commercially flexible model, the fit deteriorated. CDC's model demanded extreme resilience and rigorous process; Firmus was pursuing something leaner and faster.

Firmus now faces the quieter, more constrained path of private capital markets. Its major backers have not publicly withdrawn, but the company that was meant to anchor Australia's role in the global AI infrastructure race will build its network away from public scrutiny — without the validation, or the discipline, that a market listing would have demanded.

Firmus Technologies pulled the plug on what would have been Australia's largest stock market debut in nearly three decades, abandoning a $44 billion valuation after it became clear that investor appetite for the AI datacentre venture had simply dried up. The company, backed by chip maker Nvidia and major Wall Street firms including Blackstone, Jane Street, and Coatue, had been preparing to raise $7 billion through an initial public offering scheduled for October 23 on the Australian Securities Exchange. By midweek, the picture had shifted entirely. Bankers working on the deal had dramatically overestimated how much demand existed among institutional investors, prompting urgent conversations about slashing the proposed $11-per-share price. Rather than proceed with a diminished offering, Firmus's board decided the float no longer served the company's interests or those of its shareholders. A company spokesperson announced on Friday morning that Firmus would now turn to private markets and explore alternative funding paths, promising shareholders more details as those options developed.

The collapse marks a stunning reversal for a company that had been positioned as a transformational player in Australia's AI infrastructure landscape. Firmus operated only two small datacentre sites but had convinced some of the world's most sophisticated investors that it could build a sprawling network of liquid-cooled AI facilities across Australia and Asia. That confidence evaporated as the float approached. Skepticism had been mounting for weeks over whether a startup in its early phase could justify a $44 billion valuation, particularly given its limited operational footprint and the gap between its ambitious forecasts and its actual track record. The company's founders—Oliver Curtis, his cousin Tim Rosenfield, and Curtis's former brother-in-law Jonathan Levee—stood to see their anticipated wealth substantially reduced.

The timing of the withdrawal sent shockwaves through related parts of the market. Maas Group, an early investor in Firmus, saw its share price plummet more than 20 percent on Thursday as traders absorbed the implications. There had been growing concern, reported earlier by Guardian Australia, that existing Firmus investors were counting on the public float as their exit strategy, potentially leaving retail investors holding the bag if enthusiasm for the company cooled. That scenario never materialized because the enthusiasm never reached the public markets in the first place.

The unraveling accelerated after a separate partnership collapsed. CDC, a veteran datacentre operator, had announced what was billed as a $73 billion joint venture with Firmus called Project Southgate, aimed at building sovereign, renewable-powered AI infrastructure with Nvidia as an anchor customer. That deal, first announced in October of the previous year, fell apart this week as the two companies' visions diverged. Speaking before Parliament's joint standing committee on AI on Friday, CDC's chief strategy officer Dr. Jack Dan explained that Firmus's business model had shifted significantly since the partnership began. When Firmus was more focused on building sovereign Australian AI capability, the alignment made sense. But as Firmus evolved into a more commercially oriented venture, the fit deteriorated.

Dan described CDC's approach as centered on critical infrastructure with extreme resilience and near-total availability—requirements that demanded rigor, assurance, and process. Firmus, by contrast, was pursuing a leaner, more commercially flexible model for developing datacentres. Those competing philosophies proved incompatible. Dan noted that CDC had existed for nearly two decades in relative obscurity, but datacentres had suddenly become central to Australia's national conversation about AI capability and sovereignty. That heightened scrutiny may have also contributed to the partnership's unraveling, as both companies faced pressure to justify their strategies and commitments.

With the ASX float abandoned, Firmus must now chase capital from private investors to fund its expansion plans. The company has not disclosed how much money it needs immediately or what timeline it faces. Its backers remain substantial—Nvidia and the Wall Street firms that supported the float have not publicly walked away—but the path forward is now far more constrained and uncertain than it appeared just weeks ago. The company that was supposed to anchor Australia's position in the global AI infrastructure race will instead operate in the shadows of private markets, building its network without the scrutiny or validation that a public listing would have provided.

The board decided that proceeding with the offer was no longer in the best interests of the company and its shareholders
— Firmus Technologies spokesperson
When things start diverging significantly from that, we continue to follow our core mission
— Dr. Jack Dan, CDC chief strategy officer, on why the partnership with Firmus ended
Contattaci Domande frequenti