AI Infrastructure Firm Nscale Targets $35B IPO; High School Dropout Founder Earned $23.2M

For Nscale, the real test may have only just begun.

A company founded just over two years ago has filed for an initial public offering. Nscale, headquartered in London, plans to list on the New York Stock Exchange under the ticker symbol "NSCL."

According to the Financial Times, the IPO targets a valuation of up to approximately $35 billion (roughly ¥237 billion). In March of this year, Nscale completed a $2 billion Series C funding round at a valuation of $14.6 billion.

Financial Snapshot

• Revenue (H1 2026): $140.6 million, up 1,252% year-over-year
• Net Loss (H1 2026): $1.02 billion
• Total Long-Term Contracts (as of end of August): $103.4 billion
• Note: This figure does not represent realized revenue. Contracts currently in execution total approximately $2.6 billion; the remainder represents future projects.

The bulk of these contracts comes from Microsoft and Anthropic. The maximum disclosed contract values are approximately $43.8 billion and $44.6 billion, respectively, totaling roughly $88.4 billion—about 85% of Nscale's total $103.4 billion contract value.

The Founder: A High School Dropout Turned Billionaire

32-year-old Payne is an Australian who dropped out of high school. He worked in a coal mine, then as a construction worker, a miner, and a Bitcoin miner. In 2024, Nscale was spun out from a company Payne founded in 2020. In 2025, Payne's total compensation from Nscale reached $23.207 million (approximately ¥155 million), the vast majority from stock options.

Ahead of Nscale's IPO, Nvidia committed to a $1 billion convertible note financing. Nscale is also a major customer of Nvidia's GPUs.

The Core Strategy: Power

Payne's entrepreneurial journey did not begin with AI.

Born in 1993 in Newcastle, New South Wales, Australia, Payne left high school to work in the coal mines of the Hunter Valley. During long shifts, he read entrepreneurship books. He later moved to construction, working on large-scale development projects in Sydney. In 2017, he co-founded a mining labor company.

Around 2019, Payne entered the renewable energy sector, founding a Bitcoin mining company later renamed Arkon Energy. Bitcoin mining is a capital-intensive business, with electricity as a primary cost. Arkon focused on Norway, acquiring a Bitcoin data center in the north to leverage abundant, low-cost hydropower. This experience became the foundation for Nscale's business model.

With the rise of large AI models, GPU demand surged. But GPUs are only part of the equation. Running tens or hundreds of thousands of GPUs requires land, power, data centers, cooling systems, networks, and software. Payne's insight was to integrate the entire chain.

Nscale describes its model as "power to token"—from electricity to AI services. Around 2023, Payne began securing land and power for a new AI infrastructure business. In May 2024, Nscale was spun out from Arkon Energy.

Early on, Payne sent a cold LinkedIn message to Øyvind Eriksen, CEO of Norwegian industrial group Aker, pitching his idea for an AI data center in Norway. This led to a partnership; Aker became a major shareholder, and Eriksen joined Nscale's board.

Rapid Funding Growth

• December 2024: $155 million Series A
• September 2025: $1.1 billion Series B (led by Aker, with participation from Nvidia, Dell, and others)
• March 2026: $2 billion Series C at a $14.6 billion valuation (led by Aker and 8090 Industries, with continued participation from Nvidia, Dell, Nokia, etc.)
• September 15, 2026 (three days before IPO filing): At least $3.1 billion in financing, including a $1 billion commitment from Nvidia

In just two years, Nscale has reached the IPO stage.

Nscale's Business Model

Nscale is an AI-era infrastructure company. Large AI firms need massive GPU capacity but face challenges in building data centers, including procuring GPUs, land, power, facilities, networks, and operations. Nscale integrates these elements, offering a complete infrastructure stack from power and data centers to GPU compute and cloud services.

Nscale has two main business segments:

• Nscale Infrastructure: Power, data centers, and GPU infrastructure
• Nscale Cloud: Cloud computing, inference, and model deployment

Infrastructure remains the core of the business.

As of August 31, Nscale had approximately $2.6 billion in contracts under "take-or-pay" agreements currently in execution. Including contracts that have taken effect but not yet commenced, the total contract value reaches $103.4 billion.

"Take-or-pay" means clients pre-commit to a certain compute capacity and pay according to the contract, not solely on actual usage. The weighted average term of these contracts is approximately 5.7 years.

Key Clients

• Microsoft: Multiple agreements since late 2025, with a maximum total value of approximately $43.8 billion, with services extending to the end of 2033.
• Anthropic: A service agreement signed on August 25, 2026, with a maximum total value of approximately $44.6 billion. Nscale will deploy Nvidia Vera Rubin NVL72 GPUs at its Monarch Compute Campus in West Virginia to provide dedicated GPU infrastructure for Anthropic.

Together, these two contracts account for approximately $88.4 billion, or about 85% of Nscale's total $103.4 billion contract value. The vast majority of these contracts require future construction, deployment, and delivery.

Capacity

As of August 31, Nscale had approximately 55 MW of operational compute capacity (7 MW owned + 48 MW leased). Meanwhile, approximately 1.3 GW of capacity is under development, primarily to fulfill signed future contracts.

The Challenge Ahead

While Nscale holds a massive order book, only a small portion has been converted into operational data centers, deployed GPUs, and revenue.

The company is in a high-growth phase, with H1 2026 revenue of $140.6 million (up 1,252%). However, it posted a net loss of $1.02 billion for the same period. Revenue cost alone was $189.6 million, exceeding revenue. The company has not yet achieved profitability at the gross margin level.

The key concern is Nscale's future capital expenditure requirements. Data centers, GPUs, and power infrastructure are capital-intensive assets. Most of Nscale's long-term contracts require continued construction and deployment to be realized. The company explicitly warns in its prospectus that it will need substantial capital to support operations and expansion.

Notably, Anthropic's $44.6 billion contract includes financing conditions. Nscale must secure qualifying financing for the related GPU and data center infrastructure, which is still in progress.

Nscale's IPO narrative is a dual story: order-driven growth expectations on one hand, and sustained, massive capital expenditure pressure on the other.

Two Paths in the Same Compute Race

Nscale is not alone. In recent years, several companies have pivoted from cryptocurrency mining to AI infrastructure:

• CoreWeave: Started with Ethereum mining, shifted to GPU cloud computing.
• IREN: Moved from Bitcoin mining to AI data centers and compute services.
• Crusoe: Initially used energy resources for Bitcoin mining, now focused on AI infrastructure.

While their paths differ, the logic is similar: mining and AI computing both depend heavily on power, land, data centers, and large-scale computing equipment. The AI boom has given these existing assets a new purpose.

The Chinese Approach

China is following a different path. Traditional data center (IDC) companies are extending from IDC to AI data centers (AIDC) and compute services. As AI demand grows, data centers are no longer just providing racks and cabinets but expanding into high-density intelligent computing infrastructure and compute services.

Two main directions are emerging:

1. Intelligent Computing Infrastructure: Traditional IDCs now need to support higher power density AI servers, requiring upgrades in power supply, liquid cooling, and networking.
2. Compute Services: Companies procure and operate AI servers, directly offering training and inference compute to clients. This model is closer to Nscale and CoreWeave, requiring higher capital investment and operational demands.

Several leading Chinese data center companies are accelerating in both directions:

• One company announced plans to invest ¥30-50 billion in domestic AI infrastructure over the next three years.
• Another is advancing data center and intelligent computing projects in Langfang, Zhangjiakou, and Wuxi, with some projects designed from the start for AI computing power requirements.
• A third began its AIDC compute business in 2023. By H1 2026, this segment generated ¥1.995 billion in revenue, up 126.24% year-over-year, accounting for 53.26% of total revenue—surpassing its traditional IDC cabinet business.
• Data centers are also adding intelligent computing capabilities on top of their existing data center operations.

While paths differ, the direction is consistent: data centers are evolving from simply hosting servers to hosting and providing compute power.

Key Differences

• Chip and Compute Ecosystem: Nscale is heavily built around Nvidia GPUs, with Nvidia also as a shareholder and financing participant. China's AI infrastructure uses more diverse chip sources and technical routes, with domestic AI chip deployments expanding.
• Client and Financing Structure: Nscale locks in demand through multi-year contracts with clients like Microsoft and Anthropic, then secures equity, debt, and equipment financing. In China, large intelligent computing projects involve internet giants, cloud providers, telecom operators, data center companies, and local state capital, with more diverse financing methods.

China is unlikely to produce a "Chinese Nscale" or "Chinese CoreWeave." Instead, the likely landscape is: data center companies extending into compute services, compute companies locking in upstream facilities and power, and cloud providers and telecom operators continuing to move downstream into infrastructure. These forces are converging at the AI compute layer.

Previously, an IDC company was measured by rack count, utilization rates, and power costs. In the AI era, power, chips, data centers, and customer demand are all interconnected. Competition has shifted from single resources to the ability to organize these elements together.

Conclusion

From an Australian coal mine to AI data centers, Payne's journey has followed a consistent thread: find cheap energy, then convert that energy into computing power.

Today, Nscale holds $103.4 billion in long-term contracts, but only about $2.6 billion is in active execution. In H1 2026, the company generated $140.6 million in revenue while posting a net loss of $1.02 billion.

For Nscale, the real test may have only just begun.