AI Agent UpdateNeocloud slump deepens: CoreWeave -46%, Nebius -43% as data-center listings stall
CoreWeave and Nebius have fallen 46% and 43% in the second half of 2026, a planned $1.5 billion data-center REIT listing has been postponed, and an IPO was withdrawn — as markets ask whether AI can pay for its own buildout.
The News
Doubts over whether AI can pay for its own infrastructure are weighing on the leveraged newcomers of the cloud world, according to a UOB Kay Hian report covered November 2 by Singapore Business Review. Shares of CoreWeave and Nebius have fallen 46% and 43% respectively so far in the second half of 2026. Australia's Firmus withdrew its planned IPO in October, Blackstone-backed AirTrunk postponed its proposed $1.5 billion Singapore REIT listing — now reportedly targeting completion by early 2027 — and South Korean digital-infrastructure company Elice fell 32% on its first trading day on October 20.
The report's argument is straightforward: neoclouds spend first and earn later, borrowing heavily to buy chips and build data centers before customers sign long contracts. When financing tightens, power proves scarce, or local opposition slows construction, that model strains. Power shortages, tighter financing and environmental opposition are all slowing new development, the report says — which, paradoxically, raises the value of data centers that already exist and of vacant capacity waiting to be filled. Established hyperscalers remain better positioned, the report argues, because diversified earnings and stronger cash flows let them absorb delays that would strain a pure-play borrower.
The context sharpens the point: credit spreads — the extra interest risky borrowers pay over safe ones — have widened most at leveraged names, with Oracle and CoreWeave singled out in November 2 market coverage, and hyperscaler capital spending is eating into free cash flow across the board. None of this means AI demand has vanished; it means the market is repricing who can afford to build for it. The question has shifted from 'how fast can we build?' to 'who pays, and on what terms?'
The AI buildout's question has changed: not how fast, but who can afford to build.
Why It Matters
This matters because it marks the moment the AI trade split in two. For two years, almost any company attached to AI infrastructure rose together. Now lenders and equity investors are distinguishing between companies that can fund a buildout from cash flow and companies that must borrow to build. That distinction decides which data centers get built, which get delayed, and which never happen — and it decides it faster than any technology debate, because financing is the binding constraint on construction.
It matters for the hyperscalers because 'better placed' is not the same as 'unaffected.' Microsoft, Amazon and Google carry the same power shortages, the same permitting fights and the same capital-spending surge — they simply carry them on bigger balance sheets. Their stocks have also felt the October rout. The neocloud slump is therefore a warning light for the whole sector: if leveraged builders cannot fund capacity, the industry's total build slows, and the hyperscalers' own timelines slip with it.
The third reason is what stalled listings signal about private markets. An IPO withdrawal and a postponed REIT are not stock-market noise; they are decisions by bankers and sponsors who concluded that public investors would not pay their price. When the exit door narrows for data-center assets, private funding for the next wave of builds gets more expensive too. That feedback loop — public skepticism raising private costs — is how a valuation reset becomes a construction slowdown, and it is the mechanism to watch into 2027.
Analyst's Take
The verdict is Mixed. The neocloud slump is real and the financing pressure is documented, but the same report argues the shakeout favors owners of existing capacity — including the established hyperscalers named here — rather than signaling an AI collapse. That is a rotation story, not a demand-collapse story: money moves from leveraged builders to balance-sheet-strong operators, while total AI demand keeps growing. Bearish for the leveraged fringe, constructive for the incumbents — hence Mixed, not Bearish.
What to watch next is financing, not headlines. If CoreWeave and Nebius stabilize their funding — through customer prepayments, secured debt or slower build plans — the slump reads as a valuation reset. If credit spreads widen further or another listing is pulled, the stress spreads from the fringe toward the core. The AirTrunk timetable, targeting early 2027, is the first checkpoint. Reader poll: neocloud dip — buy, avoid, or watch?
By The Numbers
FAQ
What is a neocloud?
A neocloud is a newer, smaller cloud provider built specifically to rent out AI computing — rows of graphics processors (GPUs) that customers use to train and run AI models — as opposed to the big, diversified hyperscalers like Microsoft, Amazon and Google, which also sell software, advertising and shopping.
Why are neocloud stocks falling?
A UOB Kay Hian report covered November 2 argues investors doubt whether AI revenue can justify heavy upfront spending on chips, data centers and power. Neoclouds borrow heavily to build, so when financing tightens or power runs short, they feel it first.
What listings stalled?
Australia's Firmus withdrew its planned IPO in October; Blackstone-backed AirTrunk postponed a proposed $1.5 billion Singapore REIT listing, now reportedly targeting completion by early 2027; and South Korea's Elice fell 32% on its first trading day on October 20.
What is a REIT?
A real estate investment trust: a company that owns income-producing property — here, data centers — and lets investors buy shares in that property portfolio rather than in an operating business.
Are the big hyperscalers safe, then?
Relatively, not absolutely. The same report says established hyperscalers are better placed because of diversified earnings and stronger cash flows — but the same financing and power constraints apply to their buildouts, just with stronger balance sheets behind them.
👁
Source: Singapore Business Review