i
News
News · 2026-09-22

Nscale takes its concentrated AI infrastructure bet public

@neuronium_ai @neuronium_ai

Nscale is preparing to list on the New York Stock Exchange at a reported $35 billion valuation, with Bloomberg saying it plans to raise $3 billion. The AI infrastructure company has assembled more than $103 billion in contracts since separating from Arkon Energy two years ago, but two customers account for about 85% of that total. The IPO will therefore test not just investor appetite for AI infrastructure, but how much concentration Wall Street is willing to underwrite.

Cover: Nscale takes its concentrated AI infrastructure bet public

Contracts with Microsoft and Anthropic

Nscale’s two largest agreements are:

Microsoft: $43.8 billion for computing capacity through 2033;
Anthropic: $44.6 billion for computing capacity.

The Anthropic agreement depends on Nscale securing financing. Anthropic can also walk away or terminate the contract if Nscale misses several milestones that the IPO filing calls “strict.”

That makes the headline contract value less secure than it first appears. Much of Nscale’s future revenue is tied to a small number of customers, and one of its biggest agreements is conditional on the company raising more capital and meeting operational targets.

The pattern is common across AI infrastructure. A recent study by credit hedge fund Sona Asset Management, reported by the Financial Times, found that many providers depend heavily on a limited group of customers.

Comparable exposures include:

CoreWeave: 67% of revenue from Microsoft;
Applied Digital: 67% from Oracle and another 30% from CoreWeave.

Sona says this interdependence is not necessarily harmful by itself. But a failed project or strategy change at a major player could move quickly through the industry.

The numbers behind the valuation

For the six months ended June 30, Nscale reported $140.6 million in revenue, up from $10.4 million a year earlier. Its net loss grew from $369 million to $1.02 billion over the same period.

Revenue$140.6 million for six months
Net loss$1.02 billion for six months

In early September, Nvidia, one of Nscale’s major investors, agreed to provide $1 billion in convertible debt. The financing was part of a larger $3.1 billion fundraising transaction.

Nscale previously raised $2 billion in a Series C round led by Aker ASA and 8090 Industries. That round valued the company at $14.6 billion.

The proposed IPO valuation would therefore represent a sharp step up from the Series C. But the operating figures complicate that increase: revenue is growing quickly, while the net loss is expanding even faster in absolute terms.

A concentrated bet on AI infrastructure

Nscale operates data centers in Norway, Portugal, Texas and West Virginia. Its competitors include:

CoreWeave;
Nebius;
Lambda;
Crusoe.

Crusoe reported last week that it had raised $3.9 billion at a $30.9 billion valuation.

Nscale’s board includes former Meta executives Sheryl Sandberg and Nick Clegg, as well as former OpenAI executive Fidji Simo.

I think the more revealing part of this IPO is not the $35 billion target but the structure underneath it. Nscale has turned customer commitments into a very large contract backlog, yet the same AI buildout that creates demand also links providers, customers and financiers tightly together. The result is growth with limited room for an error by any one major participant.

What I’d want to know is how much of the $103 billion can become revenue without further financing, and how much depends on Anthropic’s milestones and other customer decisions. The announcement is clear about the size of the opportunity, but quieter about how much capital and execution it will take to make that opportunity durable.

Nscale is asking public-market investors to value future AI demand while accepting that its current business is concentrated, loss-making and dependent on financing. That tension will matter more than the contract headline once the company has to perform in public.

Daily AI news

Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.

Only what matters — every day

Follow on X