Nscale, the London-based cloud provider that exists to house and run AI models, has filed to go public in the United States. The filing pairs one of the steepest growth rates the AI boom has produced with a loss that dwarfs it, and it takes both numbers to the one stage where they have to survive public scrutiny.
Watch Desk analysis
What happened
The company has filed for a US IPO and, per CNBC's report, disclosed first-half 2026 revenue of $140.6m, up 1,252% year on year, against a net loss of $1.02bn, up from $368.9m a year earlier. That growth rate implies the prior-year half earned roughly $10m. The loss now runs at more than seven times revenue.
The filing caps a fortnight of private-market positioning. Earlier this month, Bloomberg reported that Nscale was seeking up to $3.5bn in pre-IPO financing. The company has been courting private capital while, on these figures, spending far more on building capacity than customers are currently paying for it.
The numbers
- Revenue: $140.6m
H1 2026, up 1,252% year on year, per the figures disclosed in the filing. - Net loss: $1.02bn
More than seven times the half-year's revenue, up from $368.9m a year earlier. - Prior-year base: about $10m
The growth multiple implies H1 2025 revenue near $10m, so the boom is young as well as steep. - $3.5bn in play
Bloomberg reported this month that Nscale was chasing up to $3.5bn in pre-IPO financing.
Why it matters
An IPO filing converts private AI economics into public ones. Until now, the cost of the AI infrastructure build-out has mostly been narrated through funding rounds and confident interviews. A prospectus brings accounts, risk factors and a customer list, so anyone buying into the listing gets the real books rather than the brochure. Nscale arrives just as investors have been anticipating a wave of AI listings, which makes it an early public-market read on what the build-out actually costs and earns.
The two numbers together are the story. Revenue growing more than thirteen-fold in a year shows that demand for AI compute is real and accelerating. A loss seven times revenue shows that capturing that demand still costs more than it earns. Public markets will have to price both at once, and they are historically less forgiving than private backers about the second number.
Our read
This is a genuine business with a genuinely enormous bill, and the filing forces the two into one document for the first time. Treat the 1,252% as the marketing and the $1.02bn as the due diligence. If the listing tempts you, the questions that matter are in the prospectus: how much revenue is locked into multi-year contracts, how concentrated the big customers are, and how much of the loss is depreciation on data centres that must be paid for whether AI demand holds or fades. Growth rates make headlines; customer lists make investment cases. Nscale has just handed the market the worksheet.
What to watch
- The prospectus
Customer concentration, contract quality and how the $1.02bn loss splits between running costs and capital spending. - The private round
Whether the reported $3.5bn pre-IPO financing lands alongside the listing or is replaced by it. - The pricing
The valuation the market eventually puts on 1,252% growth wearing a $1bn half-year loss. - The queue
Whether other AI infrastructure firms follow Nscale into filings and turn the anticipated IPO wave into arithmetic.
Discussion spark: Nscale grew revenue more than thirteen-fold in a year and still lost about $7 for every $1 earned. Is that a rational price for a land grab in AI infrastructure, or the moment the sector's maths starts to show?
Sources and evidence
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.