They have Nvidia, Europe’s rearmament, nuclear power or connected health on their side. On paper, everything seems to be in place to win over Wall Street and other stock markets. Yet these five highly anticipated companies have opted to postpone their IPOs. They blame market conditions. This is a convenient explanation, behind which lie far more concrete questions about valuation and profitability.
There are invitations investors would rather decline. Firmus, KNDS, Oura, Holtec Nuclear and PhonePe have all had that experience. They operate in attractive sectors, had ambitious stock market plans and ultimately postponed, suspended or withdrew their listing projects.
The refrain is almost always the same: market conditions are supposedly not right. The argument is not necessarily false. But an IPO is also a negotiation between shareholders who want to sell at the best possible price and investors seeking sufficient returns for the risk they are taking. And sometimes, the two sides simply cannot agree on the bill. Here are five different stories that each illustrate, in their own way, the return of a certain discipline to valuations.
Firmus: even Nvidia is no longer enough to convince investors
Firmus traded cryptocurrencies for artificial intelligence. Founded in Australia in 2019, the operator builds and runs data centers whose computing power it markets to AI companies. Backed in part by Nvidia, the group was considering an IPO that could have valued it at AUD 30.6bn before abandoning the transaction in October. Officially, market volatility put an end to its immediate ambitions.
But the market did not appear ready to sign a blank check. The proposed valuation represented nearly three times the AUD 10.5bn valuation set during a private financing round in early August. Investors also questioned the group’s debt, its still limited experience building AI data centers and the uncertainties surrounding a major partnership. For Joseph Koh, a portfolio manager at Blackwattle Investment Partners, the asking price was based on future performance that assumed nearly flawless execution. An ambitious assumption for a company still in the midst of its rollout.
The moral of the story: in AI, billions invested do not guarantee billions in profits. Firmus still has to turn its data center projects into operating capacity and then into cash flow. Nvidia’s backing lends credibility to its industrial strategy, but eliminates neither execution risk nor funding risk. Investors are not necessarily challenging the future of AI: they are mainly refusing to pay today for a future that still has to be built.
KNDS: wars fill order books, but not necessarily trading books
European rearmament could have rolled out the red carpet for KNDS. The Franco-German group manufactures Leopard 2 tanks, Caesar howitzers, armored vehicles and ammunition, among other products, selling them to governments before providing maintenance and modernization services. With a listing among Europe’s major stock market projects for 2026, the manufacturer put the process on hold as early as July. Once again, the group is waiting for more favorable market conditions.
Yet military demand is not lacking. With an order backlog of €33.1bn at the end of 2025, KNDS has considerable commercial visibility. The problem lies elsewhere: producing more requires expanding industrial capacity, securing supplies and controlling costs. An order does not automatically become a profit. Added to that are questions about valuation and ownership in a case where industrial and political interests play an important role. A valuation of around $15bn had been discussed.
The moral of the story: a war can fill factories without guaranteeing margins. European rearmament gives KNDS a powerful growth engine, but investors still have to assess the cost of ramping up production and the profitability of its contracts. The order book describes what the company could sell tomorrow; it does not say exactly how much it will earn. Even in defense, commercial visibility does not eliminate the need to negotiate on price.
Oura: a $15bn ring can feel a little tight
Oura succeeded in turning a ring into a health-tracking device and then into a subscription. The Finnish manufacturer markets its smart rings and charges for access to personalized analyses of sleep, activity and recovery. Buoyed by growth and the launch of new products, the group was considering a US listing that could have valued it at up to $15bn. At the end of September, it ultimately postponed the transaction, citing market uncertainty.
Product enthusiasm is not enough to justify any multiple, however. While subscriptions carry a gross margin far higher than that of the rings, hardware sales still represent the bulk of revenue. Oura therefore remains exposed to manufacturing costs, competition and replacement cycles. Investors are also questioning the long-term loyalty of subscribers. And the proposed offering included a significant proportion of shares sold by existing shareholders, rather than newly issued stock to fund growth.
The moral of the story: a good product does not automatically make a good stock. Oura has the ingredients of an appealing model, combining connected hardware and recurring revenue. But investors have to determine what share of its value comes from durable subscriptions and what share depends on the next generation of rings. The question is not whether customers like Oura, but how much shareholders should pay for that loyalty.
Holtec Nuclear: nuclear power is in favor, but its IPO may not be
Nuclear power is making a comeback, but Holtec Nuclear failed to capitalize on the stock market momentum. The US group manufactures equipment for the nuclear industry, among other activities, handles spent fuel management and develops reactor technologies. It generates revenue from equipment sales, industrial services and service contracts. Holtec was considering raising up to roughly $900m before withdrawing its IPO filing in September after an initial postponement. The group cited unfavorable conditions in the markets and in the nuclear sector.
The paradox lies in artificial intelligence. The growth of data centers is fueling expectations for electricity consumption and, consequently, interest in nuclear power. But when investors question the profitability of spending on AI, that caution can spread to companies presented as indirect beneficiaries of the revolution. Lukas Muehlbauer of IPOX Research notably highlighted the link between Holtec’s investment thesis and forecasts for data center electricity demand. The group is therefore exposed to questions surrounding a sector whose investment cycles are particularly long.
The moral of the story: electricity needs do not instantly translate into profits. Several years can pass between an anticipated increase in demand, industrial orders and their impact on financial results. Nuclear power’s comeback remains a long-term prospect, but investors are no longer necessarily willing to pay today for all of its future benefits. Even a favorable structural trend cannot shield a valuation.
PhonePe: millions of payments, but not yet a recipe for profits
In India, PhonePe has turned the smartphone into a wallet. Backed by Walmart, the fintech allows consumers to pay for purchases through instant UPI payments, often using a simple QR code. Because these standard transactions generally generate no fees for the platform, the company is seeking to monetize its network through financial services and offerings for merchants. It was considering a listing at a valuation of between $9bn and $10.5bn before suspending its project in March, citing geopolitical tensions and global volatility.
Behind the millions of users lies a less flattering question: how do you turn considerable volumes into profits? The size of a network does not guarantee its profitability, particularly when its flagship service is free. Bernstein analysts notably pointed out that PhonePe remained loss-making while its rival Paytm was approaching breakeven. Another distinctive feature: the IPO was supposed to be based on the sale of existing shares, including by early investors. The transaction would therefore have allowed them to sell part of their stakes without directly providing new capital to PhonePe.
The moral of the story: being indispensable does not mean being highly profitable. PhonePe benefits from a powerful network effect, but still has to prove its ability to capture that value sustainably. Investors are not paying solely for the number of transactions completed: they want to know what share of that activity will ultimately flow into profits. A gigantic market is not necessarily a gold mine for the company operating in it.
Five stories, one bill
From artificial intelligence to defense, and from nuclear power to digital payments, these five companies have little in common. Yet their missed appointments with the stock market tell a similar story: growth prospects are no longer always enough to justify the valuations being sought.
The phenomenon is not limited to these five names. Clear Street, Bamboo Insurance, Amaero and CopperTech Metals have also postponed or abandoned their listing plans. The circumstances vary, and market volatility is a genuine obstacle. But behind this recurring explanation, investors are also examining the strength of future profits, funding needs and the risks specific to each company.
An IPO is not a beauty contest (Keynes). It forces a company to compare its ambitions with the price investors are actually willing to pay based on their expectations. While some companies were able to benefit from a period of euphoria when merely mentioning artificial intelligence was almost enough to attract markets, the situation appears to have changed. Investors are now more demanding when it comes to valuations and profitability prospects. Some of these companies will undoubtedly return to the market when conditions improve, but perhaps at a less ambitious price.
