Nscale Raises $3.36B in Pre-IPO Financing to Accelerate Full-Stack AI Cloud Expansion

Nscale, the London-based AI cloud infrastructure company, has raised $3.36 billion in convertible loan notes as it prepares for a planned IPO on the New York Stock Exchange.

The financing was led by Third Point, the New York hedge fund founded by Dan Loeb, with participation from NVIDIA, funds managed by Apollo, Citadel, Hudson Bay Capital, the Abu Dhabi Investment Council, 8090 Industries and a range of other institutional investors.

Nscale announced the financing on September 25, shortly after filing for a US initial public offering. The company plans to list on the New York Stock Exchange under the ticker NSCL.

The financing consists of $2.36 billion funded at closing and an additional $1 billion commitment from NVIDIA that is expected in mid-November.

A $3.36B financing ahead of an IPO

The new financing is structured as convertible loan notes.

The $2.36 billion funded at closing provides Nscale with capital ahead of its planned public-market debut, while NVIDIA has committed another $1 billion expected to fund in mid-November.

The notes are expected to automatically convert into ordinary shares when Nscale completes its IPO. NVIDIA's notes will convert into non-voting shares.

Goldman Sachs acted as placement agent for the financing and is also serving as a lead bookrunner for the IPO alongside J.P. Morgan and Morgan Stanley.

The financing announcement did not disclose the conversion price, interest rate or maturity of the notes.

The source notes that Hedgeweek previously reported that the notes were being marketed at a double-digit discount to the eventual IPO price, although Nscale has not confirmed that detail.

Nscale's IPO filing also leaves the eventual price range and share count blank.

Nscale's massive contracted backlog

One of the most notable figures in Nscale's IPO filing is its reported contracted value.

As of August 31, 2026, Nscale reported $103.4 billion in total contracted value, compared with $38 billion at the end of 2025.

The company defines contracted value as revenue contracted over the full term of signed customer agreements.

However, only about $2.6 billion of that contracted value was active at the time of the filing.

The company's first-half 2026 revenue was $140.6 million, while it reported a net loss of $1.02 billion.

That creates a significant difference between the scale of Nscale's contracted commitments and the revenue currently being recognized.

According to reporting cited in the source, approximately 85% of Nscale's contracted value comes from two customers.

One is a Microsoft agreement worth approximately $43.8 billion and running through 2033.

The other is an Anthropic agreement worth up to approximately $44.6 billion involving Nscale's Monarch Compute Campus in West Virginia.

The Anthropic agreement is subject to financing and construction milestones. According to the source, Anthropic can cancel the agreement if Nscale fails to meet certain milestones described in the IPO filing as stringent.

Building a full-stack AI cloud

Nscale was founded in 2024 as a spin-out of crypto-mining company Arkon Energy.

The company has developed a business model focused on controlling multiple layers of AI computing infrastructure.

That includes land, power and data centers, as well as the GPU capacity required to run large-scale AI workloads.

The company also uses behind-the-meter power, meaning electricity generated on-site rather than being entirely dependent on electricity purchased from the grid.

This vertically integrated approach is designed to give Nscale greater control over the infrastructure required to deploy AI computing capacity.

Nscale says it serves hyperscalers, major cloud providers, frontier AI laboratories, AI-native companies and enterprise customers.

From 25,000 active GPUs to 461,000 contracted

Nscale's filing provides a snapshot of the scale of its infrastructure expansion.

As of August 31, the company had approximately 25,000 active GPUs.

It had another 461,000 GPUs that were either active or contracted across its infrastructure footprint.

The company had five active data center sites and another 12 contracted sites.

That gap between active and contracted capacity is central to Nscale's expansion strategy.

The company has signed large customer commitments and is now building the physical infrastructure needed to support those agreements.

Its workforce has also expanded rapidly.

Nscale says its team has grown from approximately 40 people to more than 1,000.

A series of large financing deals

The latest $3.36 billion convertible financing follows several major capital transactions by Nscale during 2026.

The company raised $2 billion in a Series C in March at a $14.6 billion valuation.

In May, it secured $790 million of debt for its Narvik AI infrastructure campus in Norway.

In July, Nscale announced a $900 million revolving credit facility from J.P. Morgan and Goldman Sachs.

On August 31, the company announced approximately $3 billion of additional debt to finance GPU campuses in Texas and North Carolina.

Nscale has also agreed to provide at least $3.5 billion of AI cloud capacity to humanoid robotics company Figure, with the agreement potentially expanding beyond $6 billion.

In July, the company agreed to acquire AI software company Anyscale for approximately $1.65 billion.

Together, these transactions demonstrate how capital-intensive the current AI infrastructure buildout has become.

The AI cloud infrastructure race

Nscale is operating in an increasingly competitive AI cloud infrastructure market.

Other providers are also raising billions of dollars to secure GPUs, electricity and data center capacity.

Crusoe announced a $3.9 billion Series F at a $30.9 billion valuation in September.

Lambda has reportedly been exploring a potential pre-IPO financing of up to $3 billion at a valuation of $12 billion or more, according to reporting cited in the source.

Publicly listed AI infrastructure companies are also using large debt facilities to finance expansion.

CoreWeave secured an $8.5 billion investment-grade delayed-draw term loan, while Nebius closed $5.75 billion in convertible notes in August and secured an additional $775 million loan in July.

The financing environment reflects the enormous capital requirements associated with building infrastructure capable of supporting frontier AI models.

A potential $35B IPO valuation

Nscale's IPO filing does not yet specify the final share price or valuation.

However, the source cites Financial Times reporting of a valuation target of approximately $35 billion.

Bloomberg had previously reported a financing of up to $3 billion, according to TechCrunch.

A $35 billion valuation would represent a significant increase from the $14.6 billion valuation associated with Nscale's March Series C.

The actual IPO pricing will determine the company's final public-market valuation.

For now, the company has secured $2.36 billion through the initial closing of the convertible financing, with another $1 billion NVIDIA commitment expected in November.

The challenge of converting contracts into infrastructure

Nscale's next phase will involve turning its enormous contracted backlog into operating AI infrastructure.

The company reports $103.4 billion in total contracted value, but only $2.6 billion was active as of August 31.

At the same time, Nscale had approximately 25,000 active GPUs against 461,000 active or contracted GPUs.

That means significant construction, financing, power and deployment work remains before the company's contracted capacity can become operational.

The planned IPO would provide another source of capital and public-market access as Nscale continues building data centers and GPU infrastructure across multiple regions.

Its ability to execute those projects will be central to the company's next stage.

Preparing for the public markets

Nscale's $3.36 billion convertible financing is therefore more than another funding round.

It is part of a broader capital strategy surrounding the company's planned transition to the public markets.

With more than 1,000 employees, hundreds of thousands of GPUs either active or contracted, major agreements with customers including Microsoft and Anthropic, and a growing global data center footprint, Nscale is positioning itself as a full-stack infrastructure provider for AI.

The company now has significant capital commitments behind that strategy.

The remaining question is how quickly Nscale can translate its contracted demand into operational capacity while meeting the milestones attached to its largest agreements.

Its planned IPO will provide investors with another opportunity to assess that expansion as Nscale moves from private-market growth toward the public markets.