British AI Firm Nscale Raises $3.36 Billion Before U.S. Stock Debut
Nscale, a British company that provides computing power for artificial intelligence, has raised $3.36 billion ahead of a U.S. stock listing planned for later in 2026. TechCrunch
The funding is a convertible loan led by Third Point. It starts as a loan and turns into shares at the IPO. $2.36 billion was available at once, with $1 billion more from chipmaker Nvidia due in mid-November 2026.
The loan converts automatically when the IPO is complete. It supplies cash to keep building without setting a share price yet.
Nscale filed its IPO papers in the week before September 25, 2026. It plans to list on the NYSE at a $35 billion value and raise $3 billion in the listing.
Based in London, it calls itself a full-stack AI cloud platform, a single provider of computers, chips and software for AI. Nscale Apollo, Citadel and Hudson Bay joined Third Point and Nvidia in the round. Bloomberg
Its filing lists over $103 billion in customer contracts. To meet them, it is building large data center sites in Norway and West Virginia, like warehouses filled with computers that use a lot of power.
Nscale was spun out of Australian crypto-mining firm Arkon Energy two years before September 2026. It began with computers built to run constantly and use large amounts of power, now used for AI.
These terms update earlier reports from early September. Then Nscale was discussing up to $3.5 billion, as $1.5 billion in Third Point-led loans plus about $2 billion from Nvidia. TNW The September 25 close was $3.36 billion, with $2.36 billion now and $1 billion in November.
The broader context here is plain. Nscale has promised a lot of future computing, but it must first get land, power and chips and switch the sites on. This kind of loan lets building continue before the listing.
In my view, construction matters more than the big numbers. The $103 billion only becomes sales when the centers are powered and accepted. If Nscale finishes them and lists cleanly, tech teams will have one more large option for AI computing outside the biggest providers.
