Our Blog

Blog Index 

Nscale Seeks $3.5 Billion Pre-IPO Financing: Inside the British Neocloud Riding the AI Compute Boom

Posted on 5th Sep 2026 06:04:47 in Artificial Intelligence, Machine Learning

Tagged as: Nscale, AI Infrastructure, AI Cloud, Data Centers, NVIDIA, Anthropic, IPO

On September 4, 2026, a two-year-old British company with a name most people outside the AI industry have never heard became one of the most watched businesses in technology. Nscale, a London-based provider of AI cloud computing, is in talks with investors to raise about $3.5 billion in pre-IPO financing, according to reports from Reuters, Bloomberg and TechCrunch. The company, which may list on a U.S. exchange as early as this month, is simultaneously negotiating with Nvidia for roughly $2 billion in financing and selling up to $1.5 billion in convertible notes. The numbers are large enough to make Nscale the latest emblem of a new reality in the AI economy: compute has become the scarcest and most valuable currency, and the companies that sell it are racing to the public markets to fund an unprecedented build-out.

The $3.5 Billion Raise: Convertible Notes, Third Point and Nvidia

The financing under discussion is structured in two parts. According to a Reuters source familiar with the matter, Nscale intends to sell as much as $1.5 billion in convertible notes — debt that can later convert into equity — to a group of investors, with Goldman Sachs working on the fundraising and New York-based hedge fund Third Point set to lead the convertible note investment. The notes are reportedly being offered at a double-digit discount to Nscale's IPO price, with the conversion price adjusting upward to a $30 billion valuation cap, above which it would remain unchanged. The second leg is even more significant: Nscale aims to secure about $2 billion in financing from Nvidia, the chip giant that already participated in the company's earlier rounds.

Both Nvidia and Third Point declined to comment to Reuters, while Nscale and Goldman Sachs declined to discuss the talks. Bloomberg reported that the IPO itself could raise another $3 billion, which would make the combined pre-IPO and listing exercise one of the largest AI infrastructure capital events of 2026. Nscale has previously said it may go public as early as September, with Goldman Sachs and JPMorgan reportedly advising on the listing. The timing is aggressive by any standard: a company that raised its first institutional round in December 2024 would reach the public markets less than two years after its founding, a trajectory that would have been unthinkable in infrastructure businesses a decade ago.

From Crypto Spinoff to $14.6 Billion Neocloud

Nscale's origin story explains both its speed and its appetite for debt. The company was founded in 2024 by Josh Payne and spun out of Melbourne-based crypto miner Arkon Energy, inheriting the infrastructure skills and power-negotiation playbook of the crypto mining industry. In the AI era, that playbook translated directly: Nscale owns and operates its own data centers, GPUs and software stack to deliver large-scale, GPU-powered AI compute, a model sometimes called a "neocloud" to distinguish it from hyperscalers like AWS or Azure.

The company's funding history is a steep upward curve. Its Series A in December 2024 raised $155 million. In March 2025, Nvidia participated in a $1.1 billion Series B led by investment fund Aker, which Nscale called "the largest Series B in European history." Then came a $2 billion Series C in March 2026 that valued the company at $14.6 billion and delivered a listing-ready board, adding former Meta COO Sheryl Sandberg, former Yahoo president Susan Decker and former UK deputy prime minister Nick Clegg as directors. On the capacity side, the company is scaling toward 10 gigawatts of data center capacity and nearly 289,000 active and contracted GPUs, including roughly 194,000 Nvidia Vera Rubin chips. It has announced projects in the UK, Europe and West Virginia in the United States.

The Anthropic Megadeal and the $103 Billion Contract Backlog

The single event that transformed Nscale's IPO narrative came on August 26, 2026, when Anthropic signed a six-year agreement worth approximately $45 billion to rent AI computing capacity from Nscale's West Virginia data center campus. The deal instantly made Nscale one of the largest beneficiaries of the frontier-lab compute race, and it reshaped the numbers the company presents to investors. Before the Anthropic agreement, Nscale told prospective investors it had secured around $51 billion in contracted revenue. After the deal, reports emerged that the company was citing approximately $103 billion in projected revenue based on signed customer leases, as first detailed by The Information.

That headline number deserves careful reading, and Nscale's own disclosures make the distinction clear. The $103 billion is not current sales; it is a projection of future revenue based on multi-year contracts already signed. The company's actual reported revenue is far smaller: roughly $33 million in 2025, about $37 million in the first quarter of 2026, and just over $100 million in the second quarter, putting its current annualized run rate around $400 million to $500 million. That is rapid growth from a tiny base — but it is a fraction of the contract backlog figure and a small number next to a $14.6 billion valuation. For IPO investors, the gap between contracted future revenue and recognized current revenue will be one of the central questions of the listing.

The Risks: GPU-Backed Debt, Delays and Big-Tech Competition

Nscale's growth has been financed substantially with borrowed money secured against rapidly depreciating hardware. The company has taken on a $1.4 billion loan secured against its GPUs, a $790 million facility in Norway and a $900 million credit line signed in July 2026 — roughly $3 billion in financing against a business generating a few hundred million dollars in annual revenue. The lenders' assumptions rest on AI chips remaining valuable and productive for five to six years, a horizon that could prove optimistic given how quickly new generations of accelerators arrive. If GPU values fall faster than expected, the collateral behind the debt weakens precisely when the company needs to keep building.

Execution risk has already shown up in specific projects. Nscale announced in September 2025 that it planned to deploy more than 23,000 Nvidia chips at a site in Loughton, Essex, by early 2027, but a Guardian journalist who visited the site in March 2026 found construction had not yet begun, with delays to the grid connection cited as a factor. Separately, OpenAI paused Stargate UK, a major project in which Nscale was a partner, citing high UK energy costs and regulatory uncertainty. Neither setback is unique to Nscale — grid connection queues in the UK can run seven to thirteen years — but they underscore that contracted demand is not guaranteed demand. Finally, the competitive shadow is long: Microsoft, Google and Amazon are spending hundreds of billions on their own AI infrastructure, and if they build capacity faster than expected, specialist neoclouds face pressure on both utilization and pricing.

What Nscale's Sprint Says About the AI Economy

Beyond the deal mechanics, Nscale's dash to the public markets is a signal about where the AI boom has moved. The first wave of AI investment went to model builders; the second wave is flowing to the physical layer beneath them. GPUs, data centers, power purchase agreements and grid connections are now the bottlenecks, and the companies that control them can sign contracts worth tens of billions of dollars before their revenue catches up. Nscale is not alone: its peers in the neocloud sector have attracted similar attention as investors look for exposure to AI infrastructure outside the dominant chipmaker.

The IPO, if it proceeds this month, will be a live test of how public markets price that story. The bull case rests on the contract backlog and the Nvidia relationship; the bear case rests on debt, depreciation and execution risk. Both are visible in the same balance sheet. What is no longer in doubt is the underlying demand: with frontier labs signing multi-year compute leases worth $45 billion at a time, the infrastructure layer of AI has become a capital-hungry industry in its own right, and Nscale intends to be one of its first public pure-plays. For anyone tracking the next phase of the AI economy, the company's pre-IPO financing is less a fundraising story than a preview of how the market will value the builders of the boom.

Sources

whatsapp me