> ## Content Index
> Fetch the complete content index at: https://bytevyte.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Firmus IPO Collapse: Nvidia-Backed Data Center Firm Scraps $5 Billion ASX Listing [Update]
- URL: https://bytevyte.com/firmus-ipo-collapse-nvidia-backed-data-center-firm-scraps-5-billion-asx-listing-update/
- Published: 2026-10-11T12:10:28.000Z
- Updated: 2026-10-11T12:10:28.000Z
- Description: Firmus IPO scrapped: the Nvidia-backed data center operator walked away from its $5 billion ASX listing after weak demand and will raise privately.
- Author: Bytevyte Editorial
- Tags: ai-beats

**Firmus Grid Ltd. has abandoned its Australian share sale outright**, walking away from a listing that would have raised roughly $5 billion and stood among the largest ever attempted in Sydney. The Nvidia-backed data center operator had been marketing its Firmus IPO at A$11 a share, and the collapse leaves it raising private capital instead. The decision came within a day of the offer being repriced sharply lower.

The withdrawal is the sequel to the Firmus IPO repricing we reported on Oct. 8, when the deal was cut 25% after institutional demand failed to convert into firm orders. Firmus shelved the ASX process on Oct. 9, citing market volatility and market conditions, and said it would run a private fundraising round ahead of a potential New York share sale.

Had it completed, the float would have ranked among Australia's biggest-ever stock market listings and handed public investors a pure-play bet on AI data center capacity. It did not survive contact with the bookbuild.

## The Terms Firmus Was Asking For

Firmus had spent weeks building the case for the float. It planned to sell shares at A$11 each, about US$7.65, a price that implied an equity valuation near $30.6 billion. That was almost three times the $10.5 billion valuation the company reached in early August, when it closed a $2 billion round backed by Nvidia, Blackstone, Coatue Management and Jane Street.

The August round lifted total equity raised over the preceding year past $3 billion, giving Firmus one of the strongest private cap tables in Australian technology. The Firmus IPO was designed to convert that private support into public currency at a substantially higher mark.

The size of the raise set it apart from typical ASX technology listings. Deals of that magnitude are rare on the exchange, which is one reason the offer drew attention well beyond Australia.

| Metric                   | August 2026 private round                          | Planned ASX listing                  |
| ------------------------ | -------------------------------------------------- | ------------------------------------ |
| Implied equity valuation | $10.5 billion                                      | $30.6 billion                        |
| Capital raised           | $2 billion                                         | About $5 billion                     |
| Share price              | Not disclosed                                      | A$11 (about US$7.65)                 |
| Named backers            | Nvidia, Blackstone, Coatue Management, Jane Street | Public institutional investors       |
| Register structure       | Private, locked                                    | About 58% free to trade from day one |

The gap between those two valuations was the central obstacle. Firmus was asking public investors to accept an uplift of roughly 191% in about nine weeks, against a business whose data center capacity is still being built out.

Structure compounded the pricing problem. About 58% of the register would have been free to trade from day one, leaving little protection against selling pressure from existing holders in the weeks after listing.

## Why the Firmus IPO Ran Out of Buyers

Bookbuilding works on a simple test. A company sets a price, investors signal how much stock they will take at that level, and the issuer decides whether the demand is real. Firmus cleared the first stage when it told the market it had received indications of demand. It failed the second when those indications did not translate into committed orders at A$11.

Cutting the price by a quarter on Oct. 8 was the first attempt to bridge that gap. It failed to close it. Withdrawing the offer entirely was the second attempt, and the final one.

Timing is part of the story. Repricing and withdrawal happened inside roughly 24 hours, a compressed sequence for a deal of this size. Offers of this scale normally spend longer in the market before an issuer either accepts a discount or walks away. The speed suggests the shortfall was large rather than marginal, and that extending the book would not have changed the outcome.

The company's explanation points to market volatility and conditions, a framing that places the cause in the wider trading environment rather than in the offer itself. The distinction matters for a business that expects to return to equity markets. A deal pulled on price leaves a valuation question unresolved; a deal pulled on volatility leaves the company's own case intact.

The board's position is that the marketed terms did not reflect Firmus's business strength or its long-term growth outlook. Choosing private markets over a public listing preserves that argument while sidestepping a debut at a discount to the last private mark.

## What Private Financing Changes

The pivot shifts who carries the risk of the buildout. Private backers can hold a position through a multi-year construction cycle without quarterly earnings scrutiny, something public shareholders in a newly listed infrastructure company cannot do. In exchange, the terms of the next round will be settled privately, and no public reference price for Firmus will exist until a listing is attempted again.

A private round also keeps the register closed. Existing holders stay locked in, which removes the overhang that a 58% free float would have created. The trade-off is that Firmus gives up the daily price signal and the broader investor base a listing provides.

The August backers now hold stakes marked at $10.5 billion rather than the $30.6 billion the IPO would have established. A completed listing would have crystallised a paper gain and created a liquid market in the shares. A private round leaves that mark where it was, and leaves the eventual outcome to the next set of negotiations.

It also removes a benchmark. A completed IPO would have priced AI data center capacity in public markets and given other operators raising in Australia and offshore a reference point. Without it, those companies must negotiate against private valuations that are harder to verify from outside.

Nvidia's participation in the August round is the strongest signal in Firmus's favour. The chipmaker's equity stakes in data center operators sit alongside a broader effort to secure demand for its hardware downstream. That relationship did not persuade public investors to pay almost triple the private price.

## The Wider AI Funding Signal

The shelved float arrives as investors apply tighter tests to AI infrastructure businesses. Firmus was pitching an equity story built on future demand for compute, and the market answered by rejecting the price rather than the category. The offer carried Nvidia's backing, a $3 billion-plus equity history and a national buildout story. It could not clear at roughly three times its last private mark.

AI data centers are among the most capital-hungry assets in technology. Operators commit billions before a single rack is powered, and returns depend on contracts signed years in advance. That profile has drawn private equity and infrastructure capital, but public shareholders applying tighter tests to AI valuations now want more evidence before funding construction.

For the private funds now being asked to step in, the calculus differs from that of a public investor. A closed-end vehicle can underwrite a decade-long construction programme, price the risk privately and avoid the mark-to-market pressure that comes with a listed register. Firmus needs capital for capacity that will not generate revenue for years, which fits that mandate more naturally than a daily traded share.

Australia's exchange loses a listing that would have been a landmark. The ASX rarely hosts deals of this size, and the absence of a completed transaction removes a reference point for other domestic infrastructure candidates weighing floats.

Firmus has said it will pursue private funding ahead of a possible international listing. A New York debut would put the company in front of a deeper pool of specialist technology investors, and it would reopen the valuation question in a market that has spent 2026 reassessing AI infrastructure exposure.

The A$11 price is now a ceiling rather than a target. Any future attempt will be measured against it.

## Why this matters

For anyone underwriting AI infrastructure, the Firmus withdrawal is a repricing event. Capital for data centers remains available, but it is shifting toward operators with demonstrated cash flow and away from multiples built on forward capacity. Public market investors have drawn their line on that trade; private funds have yet to say where theirs sits.

*See our earlier coverage:* [*Firmus Grid IPO Repriced 25% Lower as AI Data Center Bets Face Public Market Test*](https://bytevyte.com/firmus-grid-ipo-repriced-25-lower-as-ai-data-center-bets-face-public-market-test/)

*AI-generated image.*

## Related Articles

- [Firmus IPO Seeks $5 Billion as Losses Test AI Data-Center Appetite](https://bytevyte.com/firmus-ipo-seeks-5-billion-as-losses-test-ai-data-center-appetite/)
- [Firmus Grid IPO Repriced 25% Lower as AI Data Center Bets Face Public Market Test](https://bytevyte.com/firmus-grid-ipo-repriced-25-lower-as-ai-data-center-bets-face-public-market-test/)
- [Firmus AI Infrastructure Funding Reaches $10.5B Valuation](https://bytevyte.com/firmus-ai-infrastructure-funding-reaches-10-5b-valuation/)

✔Human Verified

---

*Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.*