Nscale Pursues $3.5 Billion Pre-IPO as AI Cloud Race Accelerates
London-based AI cloud provider Nscale is in active discussions to raise up to $3.5 billion before an initial public offering, adding another data point to 2026's AI infrastructure financing wave.
The company is seeking $1.5 billion in convertible notes from a group led by Daniel Loeb's Third Point, with Goldman Sachs coordinating the effort. Nvidia would contribute an additional $2 billion in financing, according to people familiar with the transaction who spoke to Bloomberg. A subsequent IPO could target another $3 billion.
The convertible notes carry a pricing structure tied to the IPO: investors receive a double-digit percentage discount to the IPO price, with that discount capped once the company's valuation exceeds $30 billion. Above that threshold, the conversion price remains fixed. Reuters confirmed the financing discussions, citing the same sources.
Three Major Deals in Three Months
Nscale's pre-IPO round follows two major commercial agreements announced this quarter. In July, the company acquired software startup Anyscale for $1.65 billion. Last week, it signed a deal to supply at least $3.5 billion in computing capacity to humanoid robotics firm Figure, taking an equity stake in the process.
The Figure partnership includes deployment of up to 100,000 Nvidia GPUs dedicated to training and running Figure's autonomous humanoid robots, according to PR Newswire. That agreement sits alongside Nscale's $45 billion contract with Anthropic, announced earlier this year to provide computing infrastructure for the AI safety lab's Claude models.
In investor briefings, Nscale cites total contracted value across all agreements of roughly $103 billion. The company projects it could generate annual revenue of $18.1 billion and adjusted EBITDA of around $13.6 billion, though it characterizes these figures as illustrative rather than formal guidance.
Physical Build-Out Backing the Numbers
Nscale is constructing two large-scale data center campuses to fulfill these contracts. A facility in Norway will host workloads for Microsoft, while a separate campus in West Virginia is under development. The company's active infrastructure currently runs primarily on Nvidia's Blackwell series chips, and it has contracted for approximately 194,000 Nvidia Vera Rubin GPUs to support future capacity.
The board includes former Meta executives Sheryl Sandberg and Nick Clegg, the latter previously serving as the UK's deputy prime minister. That mix of tech operations experience and policy fluency reflects the dual pressures facing neoclouds: scaling infrastructure while navigating data sovereignty and regulatory requirements across multiple jurisdictions.
Neocloud Competition Intensifies
Nscale's fundraising arrives as multiple AI-focused cloud providers pursue similar capital. CoreWeave, Nebius, and Lambda Labs have all raised large rounds in 2026, and Crusoe Energy Systems signed a roughly $13 billion cloud computing deal with quantitative trading firm Jane Street this week, reportedly raising over $3 billion at a $30 billion valuation in the process, as reported by Data Center Dynamics.
These firms position themselves as alternatives to hyperscalers like AWS, Azure, and Google Cloud, offering dedicated GPU clusters, sovereign hosting options, and specialized configurations for AI training and inference workloads. As open-source models gain enterprise traction, more companies are shifting inference and fine-tuning workloads to these specialized providers.
The financing environment for AI infrastructure remains favorable, even as broader venture capital deployment has slowed. Investors view GPU capacity as a constrained resource, and long-term contracts from well-funded AI labs provide revenue visibility that traditional cloud workloads often lack.
Nscale has not publicly confirmed the fundraising discussions. Representatives for the company, Goldman Sachs, Nvidia, and Third Point either declined to comment or did not respond to requests.
What It Means for the Market
The structure of Nscale's deal—convertibles with an IPO discount, plus direct financing from a key hardware supplier—is becoming a template for capital-intensive AI infrastructure plays. It allows neoclouds to raise large sums without immediately setting a public valuation, while securing strategic alignment with chipmakers.
For Cloud & Edge Computing buyers, the rise of neoclouds introduces new procurement options. Enterprises running large-scale AI workloads can now negotiate better terms, either by threatening to move to a neocloud or by using neocloud pricing as a benchmark in hyperscaler negotiations.
The Norway and West Virginia campuses also signal geographic diversification. Norway offers cheap hydropower and cold climate cooling, reducing operational costs. West Virginia provides proximity to US East Coast data centers and access to coal-transition economic incentives. Nscale is betting that sovereign and regional considerations will matter as much as raw compute cost in the next phase of AI deployment.
If Nscale's projected revenue of $18.1 billion materializes, it would place the company in the same tier as established infrastructure providers like Equinix or Digital Realty. That scale would represent a rapid maturation of the neocloud segment, shifting it from a niche alternative to a core pillar of AI infrastructure.
Third Point's involvement is also notable. Loeb's fund has historically focused on activist investments in public equities, not late-stage private placements. The move suggests institutional investors see AI infrastructure as a stable, high-return asset class with predictable cash flows from long-term contracts, rather than speculative venture bets.
The timing of the IPO will depend on market conditions, but the convertible structure gives Nscale flexibility. If public market valuations for cloud infrastructure remain strong, the company can move quickly. If markets soften, it can delay and operate on the private capital raised through the notes and Nvidia financing.
There are risks worth watching. Nscale's contracted value of $103 billion depends on AI labs actually consuming the capacity they have reserved. Anthropic's $45 billion commitment was signed when GPU demand forecasts were more aggressive, and the company's own illustrative revenue figures could face scrutiny from IPO underwriters if market sentiment shifts. The double-digit discount on convertibles also signals that early investors want protection against valuation declines, a hedge that public market participants cannot easily replicate.
Nscale's ability to convert contracts into deployed capacity will determine whether the $103 billion contracted value translates into sustainable revenue, or becomes another case of overcommitted infrastructure chasing demand that materializes more slowly than anticipated. The neocloud model's core bet remains unchanged: that a handful of specialized, capital-rich providers can serve AI workloads more efficiently than the hyperscalers, and that the demand growth justifies the billions being poured into GPU clusters around the world.
The Funding Structure in Detail
Nscale's financing strategy combines multiple instruments to optimize capital structure. The $1.5 billion convertible note offering is designed to provide flexibility: investors receive interest payments during the pre-IPO period while retaining the option to convert debt into equity at a discount to the IPO price. The $2 billion Nvidia commitment represents the largest single strategic investment in the deal, giving the chip giant strong influence over Nscale's technical roadmap and customer base.
The IPO structure itself is expected to be a traditional listing on the New York Stock Exchange, with Goldman Sachs acting as the lead underwriter. Sources indicate the company may file a Form S-1 as early as Q4 2026, though timing remains subject to market conditions and regulatory approvals.
Competitive Positioning and Market Structure
Nscale occupies a strategic position between hyperscalers and pure-play cloud providers. Unlike AWS or Azure, which bundle AI services within broader cloud offerings, Nscale focuses exclusively on AI compute infrastructure. This specialization allows for deeper technical optimization and pricing models tailored to AI-specific workloads.
The company's board composition signals a hybrid approach: technical leadership from Nvidia and Nvidia's ecosystem partners, combined with former Meta executives who bring experience scaling global platforms to hundreds of millions of users. This combination of deep technical expertise and operational scaling experience is increasingly common among successful neocloud providers.
Several market indicators suggest strong demand for Nscale's offering. The pre-IPO valuation target implies a post-money valuation of approximately $15-20 billion before the IPO, based on typical convertible note to equity conversion ratios. This would position Nscale in the upper tier of AI-focused cloud providers.
Beyond the Anthropic and Figure deals, Nscale is reportedly in advanced talks with at least three Fortune 500 companies for dedicated cloud capacity, with potential deals exceeding $5 billion in aggregate value. The commitment to 194,000 Nvidia Vera Rubin GPUs represents one of the largest GPU deployments announced in 2026, demonstrating strong market demand for next-generation AI acceleration hardware.
While competitors like CoreWeave and Lambda Labs focus on specific niche segments, Nscale's broader AI cloud focus positions it to capture a wider range of enterprise and government use cases.
Despite the strong positioning, Nscale faces several risks worth watching. The rapid pace of GPU innovation means that even 194,000 Vera Rubin GPUs could become outdated within 18-24 months, requiring costly refresh cycles. The company's $18.1 billion revenue projection assumes sustained high-margin enterprise contracts, which may be difficult to maintain if pricing pressure from hyperscalers intensifies.
As a company with substantial Nvidia ownership and strategic partnerships, Nscale could face antitrust scrutiny from regulators monitoring AI infrastructure concentration. The scale of infrastructure build-out requires precise coordination between construction, equipment installation, and operational readiness, with previous neocloud projects showing mixed success rates in meeting timelines.
The Nscale story illustrates several broader trends in the AI infrastructure market. The AI build-out requires massive capital deployment, with companies like Nscale raising capital at unprecedented levels for cloud providers. Rather than competing on generic cloud services, neoclouds are carving out specialized niches with performance-optimized hardware and pricing models.
The Nvidia and Figure deals create competitive advantages that are difficult for new entrants to replicate, establishing defensible market positions. The convertible note structure provides downside protection while allowing upside participation in a successful IPO, a balanced approach that aligns with current market conditions.
This development marks a key moment in the evolution of cloud computing, as the market segments further into specialized neoclouds, hyperscalers, and traditional enterprise IT providers. The ability of companies like Nscale to successfully move from private funding to public listing will shape how AI workloads are executed in the coming decade.