Nvidia-backed AI cloud infrastructure provider Nscale announced plans for US public listing with estimated $35 billion valuation.
Nvidia-backed UK AI cloud provider Nscale has launched its US listing process, filing an S-1 registration statement with the Securities and Exchange Commission on September 18 and applying to list on the New York Stock Exchange under the ticker "NSCL." Goldman Sachs, JPMorgan, and Morgan Stanley will serve as lead underwriters. According to the Financial Times, Nscale could seek a valuation of up to $35 billion in this IPO. The company also plans to open subscription channels to UK individual investors through RetailBook, allowing local retail investors to participate.
**The Business**
Nscale is an AI infrastructure company headquartered in London, with operations spanning data centers, power, GPU compute capacity, and cloud software. Rather than providing traditional enterprise cloud computing, it focuses on serving AI model training and inference demands by constructing or leasing data centers, deploying large-scale GPU clusters, and leasing compute capacity to AI companies, technology firms, and government clients.
The company was spun off in 2024 from Australian cryptocurrency infrastructure firm Arkon Energy, shifting its focus from Bitcoin mining to AI compute power. As of August, Nscale operates about 7 megawatts of compute capacity in self-owned data centers, with another 48 megawatts in leased facilities representing approximately $2.6 billion in contract value. An additional 1.3 gigawatts is in planning and construction stages, with future revenue growth contingent on successful deployment.
In March 2026, Nscale completed a $2 billion funding round at a $14.6 billion valuation, with investors including Nvidia, Dell, and Nokia. The company has raised approximately $3.7 billion in equity financing and over $5 billion in debt. An IPO at $35 billion would more than double its value in six months.
**Nvidia's Multifaceted Role**
Nvidia operates simultaneously as Nscale's core chip supplier, investor, computing customer, and credit provider. The company has invested more than $2 billion and signed computing lease agreements worth approximately $1.2 billion. Nvidia also guarantees $860 million in lease obligations for a Texas facility. Additionally, Nvidia has committed £500 million to support deployment of up to 300,000 GPUs as part of UK AI infrastructure initiatives.
Nscale plans to collaborate with Nvidia and OpenAI to expand UK computing power, and with Microsoft to build a large-scale AI supercomputer in Loughton, England. These partnerships provide Nscale with predictable customer demand and strengthen its position in capital raising and facility acquisition, while enabling Nvidia to extend GPU adoption across UK and European markets.
**The Anthropic Contract and Capital Intensity**
Nscale's most significant project is in West Virginia, where it has signed a six-year computing lease agreement valued at up to $45 billion with Anthropic. The associated data center campus is expected to begin phased operations by end of 2027, accompanied by construction of a 2-gigawatt natural gas power facility, with total costs potentially reaching approximately $70 billion.
While this contract provides substantial long-term revenue visibility, it underscores the capital-intensive nature of the business. Data centers require massive upfront investment before generating revenue, creating risk if construction faces delays, power supplies fall short, or customer demand shifts. Financial results reflect these pressures: Nscale reported approximately $141 million in revenue for the first half of 2026, with net losses approaching $1 billion driven partly by non-cash charges. Relative to its $35 billion valuation target, current revenue remains modest—the company is effectively being valued on unbuilt capacity, long-term contracts, and future expansion potential.
**Outlook and Uncertainties**
Recent comments from Anthropic CEO Dario Amodei advocating caution in frontier AI development, along with concerns voiced by OpenAI's Sam Altman and Elon Musk about AI safety, briefly triggered declines in AI chip and data center stocks. However, even if frontier model development pace moderates, commercialized AI products will require substantial inference compute capacity.
Nscale's success ultimately hinges on whether it can deliver its 1.3 gigawatts of capacity under construction on schedule, translate partnerships with Nvidia, Microsoft, OpenAI, and Anthropic into sustained cash flow, and manage the debt and financing risks accompanying rapid expansion.