Fidji Simo가 GPU 스타트업 Nscale의 이사회에 합류했으며, 회사의 가을 IPO를 불과 며칠 앞두고 있습니다.

Fidji Simo is joining Nscale's board at exactly the moment the London AI infrastructure startup needs seasoned operators: before a possible US IPO, with enormous contracts, heavy debt, and public investors ready to scrutinize the fundamentals.

Nscale announced on September 11 that Simo has joined its board of directors, effective immediately. The move carries significant weight. Simo led Instacart through its 2023 listing, joined OpenAI in 2025 as CEO of Applications reporting to Sam Altman, and stepped back from her full-time OpenAI role in July following medical leave related to Postural Orthostatic Tachycardia Syndrome, the chronic condition she has discussed publicly. She will remain as a part-time adviser to OpenAI and has agreed to recuse herself from Nscale board matters involving OpenAI.

That recusal matters because Nscale is not selling productivity software or another chatbot wrapper. It is selling access to the compute that companies like OpenAI, Anthropic and Microsoft need before any AI product reaches customers.

Simo is not walking into an empty boardroom. Sheryl Sandberg, Nick Clegg and Susan Decker joined Nscale's board earlier this year, after the company raised $2 billion in Series C funding. Those appointments signal what Nscale is trying to become before an IPO filing: not merely a GPU rental startup, but a public-market infrastructure company staffed with Big Tech operators.

Nscale was founded in 2024, growing out of the Arkon Energy orbit—the crypto-mining and data-center infrastructure business associated with Josh Payne. Its pitch is straightforward but demanding: build data centers, secure power supply, acquire or finance advanced Nvidia hardware, then rent capacity to AI labs and large customers unable to source sufficient compute from conventional cloud providers.

The scale of customer commitments is already substantial. Anthropic has agreed to pay Nscale roughly $45 billion over six years for AI computing capacity from a new West Virginia data center campus, using Nvidia's unreleased Vera Rubin chips. Microsoft committed about $23 billion to Nscale for access to roughly 200,000 Nvidia GB300 chips across sites in the UK, Norway, Portugal and Texas. These are not typical startup contracts; they are infrastructure commitments measured like power plants.

In March, Nscale announced its $2 billion Series C valuation of $14.6 billion, led by Aker ASA and 8090 Industries with Nvidia participating. According to Bloomberg, Nscale is now pursuing a US IPO that could raise as much as $3 billion, with Goldman Sachs and JPMorgan advising on the process.

The tension lies in the gap between booked ambition and operating reality. Reuters reported that Nscale has been telling prospective investors it has about $103 billion in total contracted revenue after the Anthropic deal, with contracts averaging 5.7 years—implying roughly $18 billion annually on paper. However, sources cautioned that this figure was illustrative rather than formal revenue guidance.

Recognized revenue tells a different story. Leverage Shares noted in an investor note that Nscale generated roughly $33 million in recognized revenue in 2025, about $37 million in the first quarter of 2026 and just over $100 million in the second quarter. The company also carries substantial financing obligations: a $1.4 billion GPU-backed loan, a $790 million Norway facility and a $900 million credit line signed in July. These obligations mount quickly for a business still demonstrating how smoothly long-term contracts convert into actual cash flow.

Here is what Simo brings that a purely infrastructure executive does not: experience navigating public-market scrutiny through a growth story. Instacart went public in 2023, then spent extended periods trading below its IPO price as investors examined margins and the durability of demand. Nscale is making a different pitch, but the discipline is identical: demonstrate what is real and acknowledge what could break if market conditions shift.

Nscale's expanded board suggests the company understands the examination ahead. Sandberg brings scaled operations and advertising experience. Clegg knows policy pressure. Decker brings public-company finance expertise from Yahoo and other boards. Simo adds consumer product experience and IPO readiness. None of this lightens the debt load or reduces Microsoft exposure, but it signals serious preparation for public markets.

A strong board, however, does not guarantee a solved business. CoreWeave's public listing demonstrated how eager the market is for AI infrastructure and how quickly investor appetite can curdle when underlying economics are questioned. Nscale faces a sharper challenge: asking public investors to bet on enormous future contracts before years of actual operating history validate the model.

Still, Simo's appointment deserves attention. OpenAI alumni could easily chase another model lab, another assistant, another application. Instead, Simo chose a board seat at the compute layer. In this cycle, that may prove the more durable position—provided the data centers get built, the chips remain useful, and customers continue paying long after the IPO roadshow concludes.