UK AI infrastructure company Nscale has raised $3.36 billion through convertible loan notes ahead of its proposed stock-market listing, adding another large pool of private capital to a company expanding data centres and GPU capacity across several markets.

The financing is led by Third Point, with Nvidia, Apollo-managed funds, Citadel, Hudson Bay Capital, Abu Dhabi Investment Council, 8090 Industries, and other investors participating. Nvidia has committed $1 billion of the total.

The notes are structured to convert into equity around Nscale’s planned initial public offering, for which the company filed its registration statement on 18 September and has applied to list on the New York Stock Exchange under the ticker NSCL.

The transaction follows several other large financing commitments around Nscale’s infrastructure, demonstrating how the economics of AI cloud increasingly depend on the ability to raise capital well before the revenue attached to new sites is fully realised.

The funding follows the infrastructure bill

Nscale describes its platform as a full-stack AI cloud serving hyperscalers, model developers, AI companies, and enterprises. Its strategy extends beyond renting computing equipment to customers because the company is also involved in the data-centre, power, networking, cooling, and GPU infrastructure underneath those services.

That makes growth expensive. New capacity requires power, specialist buildings, cooling, networking, and accelerators before customers can consume it, creating a gap between committing to infrastructure and receiving the cash generated by the finished capacity.

Nscale says it has more than $103 billion in total contracted value. Contracted value is not equivalent to revenue recognised today, however, because infrastructure agreements are realised over time and depend on facilities and hardware being delivered.

The financing requirement can therefore grow alongside the order book. The more capacity a provider sells for future delivery, the more capital it may need to build the physical environment required to honour those contracts.

AI cloud becomes a financing contest

Specialist GPU providers have emerged beside the traditional hyperscale cloud market, promising faster access to accelerators and infrastructure designed specifically around AI training and inference.

The model can generate rapid growth when compute is scarce and highly utilised, but providers remain exposed to hardware cycles, electricity costs, construction schedules, financing terms, and the concentration of large customers.

Access to capital therefore becomes part of the competitive position. Demand is of limited value if the supplier cannot finance the next cluster, while operators capable of raising billions can reserve sites and hardware well ahead of utilisation.

Nvidia’s participation is notable because it sits upstream in the same market. New AI cloud capacity creates additional demand for its hardware, while infrastructure operators depend on continuing access to high-end accelerators.

The relationship highlights how closely capital, chips, power, and cloud services are becoming connected as AI infrastructure expands.

Contracted demand still has to become cash

Nscale’s listing preparations will expose that business model to public-market scrutiny. Investors will need to assess how quickly facilities become operational, how efficiently capacity is used, how customer concentration evolves, and how much additional investment is required to produce each increment of revenue.

The broader sector faces the same test. Demand forecasts can justify enormous projects, but data-centre and cloud operators eventually have to convert those forecasts into infrastructure that is powered, connected, occupied, and profitable for long enough to recover its cost.

The $3.36 billion convertible financing gives Nscale more room to build before its proposed listing. It does not make the infrastructure less capital intensive; it demonstrates the scale of financing now required to compete in a market where physical capacity has become one of the constraints on AI growth.