Nscale lands a $900M revolving credit facility for AI buildout

Nscale closed a $900 million revolving credit facility on July 7, 2026, giving the UK-based AI infrastructure company flexible liquidity to fund data center expansion across North America, Europe, and the Asia-Pacific region. The facility was syndicated by a twelve-bank consortium led by J.P. Morgan, Goldman Sachs, and Morgan Stanley.

The structure is a revolver rather than a term loan, which means Nscale can draw funds and repay them as project capital calls land rather than sitting on a fixed draw schedule. The company's pitch is that AI infrastructure has outgrown the financing templates of the previous generation of cloud buildouts. "The closing of this revolving credit facility with key global investment banks reflects real institutional confidence in our platform, capital structure, and team," Josh Payne, Nscale's CEO and Founder, said in the PR Newswire announcement. "We are building the infrastructure that the world's largest technology companies depend on to train, deploy, and scale AI, and this facility increases our flexibility to do that at speed and at scale." The lender roster reads like a list of who underwrites large infrastructure deals, and the structure says the bank group expects Nscale to draw and repay on a regular cadence rather than treat the facility as a single fundraise.

What Nscale is building and where

Nscale describes itself as a full-stack AI cloud platform that pulls software, compute, and power into a single vertically integrated offering. The pitch spans a unified cloud for AI training and inference, purpose-built data centers, and low-cost power contracts designed for high-density compute. The company's footprint is concentrated in three priority markets, and the revolver is sized to give the company enough optionality to land anchor tenant deals in each of them without having to refinance for every project. Nscale has been operating in the United Kingdom, but the facility is explicitly structured to fund expansion in the US, Europe, and APAC.

The most concrete deployed assets sit in Europe. In May 2026, Nscale and Microsoft announced a partnership to develop data center capacity at the Sines Data Campus in Portugal, where Nscale has already installed more than 12,600 Nvidia Blackwell Ultra GPUs in the first phase of the build. In April, the two companies also confirmed a separate data center project in Norway. The Sines site is one of the densest single-site AI training deployments publicly disclosed outside the hyperscaler-owned campuses, and the Norwegian site is positioned as a sovereign-AI capacity anchor for Nordic customers. The Microsoft agreements gave Nscale a marquee anchor customer and a publicly disclosed GPU count, which is the kind of proof point that helps a debt syndicate price a revolving facility rather than a term loan with milestone-based tranches.

The expansion is funded against a customer pipeline that includes hyperscalers, large enterprise model training programs, and government-backed sovereign AI deployments. The revolver gives Nscale a way to deploy capital against signed contracts without diluting equity or waiting for the next equity round. The Bank of America and Goldman Sachs credit teams that ran the syndication are the same groups that have underwritten several other large AI infrastructure deals in the last 12 months, including the recent Nvidia-backed AI cloud operator financing round at Sharon AI and Firmus, and the syndicate composition suggests that the AI infrastructure sector is now a recognized debt-financing category rather than a one-off venture story.

The lender list and what it signals

The twelve-bank roster is one of the strongest signals in the announcement. J.P. Morgan, Goldman Sachs, and Morgan Stanley sit at the top of the AI infrastructure debt book, with the relationship-banking capacity to underwrite ten- and twelve-bank syndicates. MUFG, Mizuho, SMBC, and the European banks (Crédit Agricole CIB, Deutsche Bank, RBC, TD, and KeyBank) round out a roster that covers North America, Europe, and Japan, which is roughly the geographic footprint of Nscale's pipeline. The mix of US money-center banks, Japanese megabanks, and European universal banks is the right syndicate shape for a project that will draw in multiple currencies and book capacity in multiple jurisdictions.

The structure is also worth noting. A revolving credit facility is a working-capital-style instrument, not a project finance loan. Project finance for data centers typically includes milestone-based tranches, completion guarantees, and offtake contracts as collateral. A revolver treats the company itself as the credit, with covenants tied to balance sheet metrics rather than specific site completions. The shift from project finance to corporate revolver is a sign that Nscale has reached the scale where the lenders underwrite the platform rather than each individual site, and it is the same shift that the larger AI infrastructure operators made when they crossed the multi-hundred-megawatt deployment threshold.

Nscale is one of a handful of Nvidia-backed AI cloud operators outside the hyperscaler tier, and the company has been positioning itself as a sovereign-AI capacity provider for European and APAC customers that cannot or do not want to run training workloads on US hyperscaler infrastructure. The Microsoft deals in Portugal and Norway gave the company two anchor deployments, and the revolver gives the balance sheet to convert those anchors into a multi-region footprint. For enterprise buyers evaluating AI training capacity outside the hyperscaler tier, the syndication is a positive signal that the underlying operator has reached the credit profile required to deploy capital at scale. For the broader AI infrastructure market, the deal is another data point that the financing category has matured past the equity-only stage and into a phase where the major banks will underwrite AI data center buildouts on the same terms they use for other large infrastructure assets. The AI Infrastructure in 2026 resource page covers the broader trend of sovereign AI and GPU-dense sites for buyers evaluating the same set of options.

The first draw against the facility is expected to support the next phases of the Sines Data Campus and the Norwegian site, with subsequent draws allocated to North American and APAC sites as anchor contracts close. The next milestone to watch is the second-phase Sines deployment, which is expected to push the campus past 30,000 GPUs once it is fully built out, and the Norwegian capacity that Nscale and Microsoft have signalled will follow a similar density profile.