Search

Nscale Adds Two Data Centers and 25,000 GPUs in Under Three Months Ahead of IPO

By: IDCNOVARegion: Europe
British AI infrastructure provider Nscale has significantly accelerated its buildout, bringing two additional data centers and roughly 25,000 GPUs online since the end of August 2026, according to a newly published IPO prospectus. The disclosure underscores the rapid pace at which AI cloud providers are racing to secure compute capacity amid surging demand from model developers and enterprises.

The updated prospectus reveals that Nscale now operates 50,000 active GPUs out of a planned fleet of 461,000, doubling the 25,000 active GPUs reported in its September prospectus dated August 31, 2026, which represented only about five percent of the company's target at the time. Alongside the GPU expansion, Nscale has grown its active data center footprint from five sites totaling 55MW to seven active locations, with ten more in the pipeline. The company did not disclose the aggregate capacity across the seven operational sites in the latest filing.

In total, Nscale says it holds 1.37GW of active and contracted capacity, comprising 1GW at its owned sites, 200MW at leased facilities, and 165MW at colocation data centers. The company claims a "line of sight" to 10GW of potential power capacity across its footprint, signaling ambitions to become a major player in AI infrastructure provisioning.

The prospectus also reflects a change in how Nscale reports its financials. The previous filing cited revenue of $140.6 million for the six months ending June 2026, up 1,252 percent year-on-year, alongside losses that ballooned to $1.02 billion from $368.9 million in the same period of 2025. The new prospectus instead focuses on the full year 2025 ending December 31, reporting revenue of $33 million, a 73 percent increase from $19.1 million in 2024. As of the prospectus date, Nscale reported approximately $6.4 billion of active total contract value and $105.8 billion of active and contracted TCV under long-term take-or-pay agreements, compared with $0.5 billion and $38.0 billion respectively as of December 31, 2025. These contracts underpin the deployment of approximately 461,000 GPUs that were active or contracted as of that date.

The filing also provides additional detail on Nscale's agreements with Anthropic. In August, the two companies signed a $45 billion compute agreement covering Nscale's upcoming Monarch Compute Campus in West Virginia. The new prospectus confirms that on September 29, 2026, Anthropic exercised its right to contract for additional compute at Building 2 on the Monarch campus, with both parties having 60 days, subject to mutual extension, to negotiate an agreement in good faith. Anthropic may also exercise its right with respect to Building 3 within 15 business days after the Building 2 agreement is signed, after which the parties would have 60 days to finalize a written agreement for Building 3.

Nscale said it expects to fund the Anthropic deployment through external financing and cash flow from operations. While the company acknowledged it has not yet secured any financing agreements, it stated that it has tested market depth based on discussions with market participants and has a track record of securing data center and GPU financings, expressing confidence in obtaining qualifying financing within the required period. The prospectus also notes that should Nscale fail to provide GPU infrastructure to Anthropic on schedule, including due to supply chain delays, Anthropic may be entitled to terminate that tranche without liability.

Nscale is reportedly hoping to raise billions of dollars through its IPO to fund its AI data center buildout, targeting a valuation of approximately $30 billion, though no official date for the offering has been provided. The accelerated GPU and site deployment, combined with the massive Anthropic contract, positions Nscale among the more aggressive challengers in the AI infrastructure market, where securing power, land, and chips ahead of rivals has become a defining competitive advantage.