Nscale files the paperwork for a September IPO with $51 billion in contracts
AI infrastructure is heading to public markets fast, and the neoclouds are leading the charge. Nscale, the London-based AI data center builder, is reportedly preparing a U.S. IPO that could raise up to $3 billion as soon as next month.
Nscale is targeting up to $3 billion in a U.S. IPO that could land as soon as September, according to a Bloomberg report citing people familiar with the matter. Goldman Sachs and JPMorgan are managing the offering. The company has not set a target valuation; it was valued at $14.6 billion in its March funding round, which counted Nvidia and Nokia among its backers.
The pitch to investors rests on contracted revenue rather than hope: Bloomberg's sources put total contracted revenue at about $51 billion. The anchor is Microsoft, which commissioned 1.35 gigawatts of capacity at Nscale's 2,250-acre West Virginia campus in March — to be delivered on Nvidia's latest Rubin-based systems — plus a reported $14 billion contract to host 300,000 Blackwell Ultra chips across four other locations. All told, Nscale operates roughly 831 megawatts today and plans to grow toward 11 gigawatts, anchored by a West Virginia site it says could theoretically hold more than eight gigawatts of capacity.
Nscale also spent July buying its way up the software stack, acquiring Anyscale for $1.65 billion to fold in a commercial version of Ray, the open-source tool for orchestrating AI clusters. That matters for the IPO story: hardware-only neoclouds compete on price per GPU-hour, while a software layer gives Nscale something closer to recurring revenue and stickier margins.
Why it matters: the AI build-out is starting to test public-market appetite directly. Nscale would be the first pure-play AI data center builder to list, ahead of Switch (a confidential filing reportedly targeting a $50 billion valuation) and Vantage Data Centers (exploring options at around $100 billion). We covered the IPO wave building earlier this month — Vantage explores $100B IPO as data center listings stack up. A $51 billion backlog makes Nscale's file one of the strongest on the pile; whether public investors pay growth-multiple prices for power-hungry, customer-concentrated infrastructure is exactly what this listing will test. If you want one number to watch, it's the final valuation against that $14.6 billion private mark from March.
China's first national standard for AI customer service takes effect September 1, and its core demand is simple: the exit to a human must be easy to find. The standard, issued by China's market regulator (GB/T 47746—2026), requires platforms to show a clear "transfer to human" option in their service menus — no hiding it behind loops of canned answers — and to hand users to a person automatically when they explicitly reject the bot or safety is at stake. A companion position from the China Consumers Association closes the accountability loophole: companies own what their bots say, and "the AI's answer doesn't represent the company" no longer works as a defense.
The rule targets an economic distortion, not just bad manners. A human agent costs roughly 3,000 yuan a month while AI service plans start around 99 yuan, and Chinese media report some operators grade their bots on "interception rate" — how few callers reach a person — instead of how many problems get solved. When the incentive is to filter out customers, "answers that miss the question" isn't a bug, it's the product. If enforcement has teeth, expect versions of this rule elsewhere: the EU mandates human oversight for high-risk AI systems, but China is regulating the mundane case first, exactly where AI touches the most people.
What happens when the first company gets fined for hiding the human option? Tell us in the comments.
