NextFin News - Australia's capital markets are being asked to do two opposite things at once: fund a US$5 billion artificial-intelligence IPO that would be the second-largest float in the country's history, while a property downturn strips tens of millions of dollars off trophy homes and a collapsed developer leaves more than 3,000 creditors chasing A$3.4 billion. The question is not whether Australia can list Firmus Technologies - it is whether the same pool of capital can absorb both the AI boom's promise and the property sector's bill.
The Week in Three Stories
On 21 September, a term sheet circulated to investors set out the timetable for Firmus Technologies' Australian Securities Exchange listing: an institutional bookbuild on 6 and 7 October, a prospectus lodged on 8 October, retail bidding from 12 to 19 October, and first trading on 22 October. The company is seeking to raise A$7 billion (about US$5 billion), or up to US$5.5 billion if the overallotment option is exercised, according to the term sheet. Firmus declined to comment. If completed, the float would be the fourth-largest IPO globally in 2026 and the biggest Australian listing since Telstra's US$10 billion offering in 1997.
Behind the IPO sits a capital stack that already tests the limits of local risk appetite. In February, Firmus closed a US$10 billion debt facility led by Blackstone's tactical opportunities and credit units; in April it raised US$505 million in equity led by Coatue at a roughly US$5.5 billion valuation. The company, founded in 2019 by Oliver Curtis, Tim Rosenfield and Jonathan Levee, says it has more than 900 megawatts of data-centre capacity under contract across seven sites it calls "AI factories," but only two are operational - one in Australia and one in Singapore.
The property side of the ledger tells a different story. On 18 September, the Supreme Court of New South Wales extended the convening period for the voluntary administration of Bathla Group to 13 September 2027 - a full year of runway that administrator Teneo Australia says is conditional on funding it has not yet secured. Bathla entered administration on 25 August owing about A$3.4 billion. On 7 September it secured A$4.7 million in short-term funding from five lenders, stood down 213 workers and suspended construction on projects not backed by those lenders. The extension buys time, not certainty.
And at the very top of the housing market, the cracks are visible in single transactions. Rothesay, a harbour-view mansion at 24 Pearson Street in Balmain East, sold for A$12.8 million in September - almost A$7 million below the A$19.76 million it fetched in 2022, during the pandemic property boom. The sale is a data point, not a trend. But it lands in a trophy-home market where discounts of A$20 million off original asking prices have already appeared, and where the Reserve Bank of Australia's cash rate of 4.35 percent is the highest in more than a decade.
What the Firmus Float Is Really Testing
The obvious read of the Firmus IPO is that Australia's market is finally mature enough to house a technology flagship. The deeper test is whether investors will pay for capacity that exists mostly on paper. The company's valuation ambition - reported at between US$30 billion and US$60 billion, or as much as A$90 billion - rests on management's stated target of US$5 billion in earnings before interest and tax by the 2028 financial year, a multiple of about 12 times that figure. That multiple is pitched as a discount to listed peer CoreWeave, which has traded at as much as 17 times forecast EBIT.
This is where the mechanism matters. Firmus is not being valued as a data-centre operator with contracted cash flows; it is being valued as a claim on the AI build-out that has not yet been built. Two of seven factories are running. Five are under development across Australia and Asia-Pacific. The contracts with customers including Meta and OpenAI are real, but the assets that generate revenue from those contracts are largely unbuilt. The valuation therefore depends on three things happening in sequence: capital must be raised, construction must be completed on time, and the AI demand that justified the contracts must still be there when the facilities open. Break any link and the 12-times multiple has no earnings to attach to.
"You have two groups of fund managers - those that want to assassinate it because they're not there, and those that want to cheer it because they are," said Gary Rollo, a portfolio manager at Montgomery Investment Management.
Rollo's point cuts to the structure of the bookbuild itself. With only institutional investors invited to the initial allocation, the first price is set by buyers who want exposure and by bankers who need the deal done. The second price - the one that matters for anyone buying after 22 October - will be set by a market that includes everyone else. An A$7 billion raise when the 10-year Australian government bond yield sits at 5.30 percent, near multi-year highs, is not the same transaction it would have been in 2021.
There is also a technical reason the ASX listing makes sense for Firmus, and it has little to do with Australian technology expertise. A company capitalised at A$50 billion or more would be large enough to attract mandatory ownership from index funds that track the S&P/ASX 200. That creates a built-in bid from passive managers who cannot ignore a stock of that weight. Bankers can sell that dynamic to large fund managers as a reason not to sit out the float. It is a market-structure advantage, not a judgment on the underlying business.
The Property Downturn Is Not One Story
The property side of the briefing contains two distinct problems that are often conflated. The first is a cyclical correction in trophy-home prices. The second is a solvency problem in the development sector, and it is structural.
The trophy-home correction is a function of interest rates and buyer psychology. With the cash rate at 4.35 percent, the cost of carrying a A$30 million property on debt is no longer trivial, and the pool of buyers willing to park capital in illiquid real estate has shrunk. The Rothesay sale - down A$7 million from 2022 - is consistent with a market where vendors who need liquidity meet buyers who know they have time. The A$20 million discounts reported in Sydney's top-end sales this year are the same mechanism at a larger scale.
Bathla is different. This is not a price correction; it is a balance-sheet failure that exposes how the development model worked when rates were low. A developer that sells its head office for A$29 million and rents it back eight months before collapse is not managing a temporary cash-flow gap. When the administrator's review finds more than A$730 million in overstated receivables, the question shifts from "when will funding return" to "what was actually owed." Receivables are claims on money that has not arrived. If A$730 million of them were overstated, then the A$3.4 billion creditor stack was built on assets that did not exist in the form reported.
The human and political dimensions are already visible. Nearly 1,000 subcontractors are owed about A$130 million and sit behind secured lenders in the repayment queue. The state's home warranty scheme could face a bill of up to A$365 million. Fair Trading fined Bathla's builder A$45,000 and suspended its director in the weeks before the collapse. And the company's own visa agency, Brothers Migration & Education, has lodged a claim of A$310,556.68 against the same corporate entity that hosts it - while founder Bhart Bhushan's wife, son and daughter have claimed almost A$1.8 million between them.
"The extension gives us the time needed to progress projects towards completion, but that outcome remains dependent on securing further lender support to maintain critical central functions at Bathla," administrator Stephen Longley said. "Locking in this funding is therefore an immediate priority and discussions are continuing."
That sentence contains the whole Bathla story: time has been granted, money has not. A one-year extension without funding is a runway to a decision, not a rescue.
The Second-Order Question: One Pool of Capital, Two Competing Claims
The first-order reading treats these as three separate stories. The second-order question is whether they compete for the same capital. Australia's equity market is small by global standards. An A$7 billion IPO would be one of the largest single equity events the market has absorbed in a decade. At the same time, the property sector is asking lenders to extend more credit into a development pipeline that just produced a A$3.4 billion failure, and the public sector may be asked to cover up to A$365 million in warranty claims.
The transmission channel runs through risk pricing. If the Firmus float succeeds at the high end of its range, it signals that growth capital is still available and that investors are willing to underwrite construction risk in exchange for AI exposure. If it prices at the low end or trades poorly after debut, the signal runs the other way: capital is scarce, and the premium for unbuilt capacity is too high. Either outcome feeds back into the property market, because developers and data-centre builders compete for the same contractors, the same grid connections, and increasingly the same political patience over power and water use.
The conventional wisdom is that the AI trade and the property trade are uncorrelated - one is technology, the other is real estate. That is wrong at the level that matters for Australia. Both are claims on the same finite inputs: electricity, construction capacity, and investor risk tolerance. A data-centre developer and an apartment developer in Western Sydney are bidding for the same pool of electricians and the same transmission lines. When one sector's capital cost rises, the other's does too.
The counterweight is Australia's A$4.4 trillion superannuation system, which allocates to equities and property through separate mandates with separate risk budgets. That scale is what makes the Firmus float plausible at all. But scale is not the same as appetite - and appetite is what this week will measure.
The Counter-Thesis
The strongest case against reading these events as connected is that Australia's market is deep enough to segment them. Superannuation funds allocate to equities and property through separate mandates with separate risk budgets. A disappointing tech IPO does not force a property fund to sell, and a developer collapse does not cap the valuation a growth fund will pay for AI infrastructure. On this view, Firmus will be judged on AI demand and power contracts, Bathla on construction law and creditor recoveries, and trophy homes on interest-rate expectations. The three stories share a country, not a mechanism.
That argument holds if capital is truly siloed. It weakens if the Firmus float underperforms badly, because underperformance would not stay inside the technology mandate - it would raise the equity risk premium that every growth asset in Australia is priced off. The falsifying signal is specific: if Firmus prices at the low end of its range and then trades more than 20 percent below its offer price within three months of the 22 October debut, the "segmented markets" thesis is wrong, and the property sector should expect tighter credit and a higher cost of capital. If, instead, the stock holds its issue price and the float is oversubscribed, the segmentation view survives and the property downturn remains a sector-specific story.
What Comes Next
Short term (the float itself): watch the bookbuild on 6-7 October and the retail window from 12-19 October. The size of the institutional book relative to the A$7 billion target, and whether the greenshoe is exercised, will set the first read on demand. The prospectus lodged on 8 October will carry the company's first published revenue figures - the first hard data on whether the US$5 billion FY2028 EBIT target has any foundation.
Medium term (funding and construction): for Bathla, the trigger is lender support. Without it, the 13 September 2027 extension is a formality that ends in asset sales and creditor haircuts. For Firmus, the trigger is construction progress at the five unbuilt factories - any delay pushes the earnings target further out and compresses the valuation multiple the market will accept.
Long term (structural read): the trophy-home market is cyclical and will recover when rates fall; the development-finance model exposed by Bathla is the one that needs rebuilding. Australia can list a A$90 billion AI company and still have a broken developer-funding system. The two are not contradictory - they are what happens when an economy transitions from property-led growth to infrastructure-led growth, and the capital markets are asked to finance the bridge.
The base case is that Firmus lists in October at a valuation below the top of its range, enough to clear the market but not enough to declare a new era. The upside case is an oversubscribed book that forces index funds to buy and pulls the whole Australian technology sector higher. The downside case is a weak debut that coincides with a Reserve Bank rate decision - the next meeting is 28-29 September - that keeps borrowing costs high and reminds investors that cheap money is not coming back.
Australia is not choosing between an AI future and a property past. It is being asked to pay for both at the same time, with the same money, and the market will decide which claim is senior.
Explore more exclusive insights at nextfin.ai.
