Report: Anthropic to Invest Over $600 Billion in Computing Power. The company generated $16.3 billion in revenue in the first half of the year, with approximately $9 billion spent on computing power.

Anthropic is burning cash and expanding its computational power at an unprecedented pace, while also accelerating what could be the largest IPO in history.

On October 20, according to The Wall Street Journal citing a draft of Anthropic's IPO prospectus, the AI startup reported a net loss of over $80.7 billion in the first half of 2026, with $72 billion of that resulting from accounting charges related to valuation changes in convertible notes and other financing instruments, a sharp increase compared to the $13 billion net loss in the same period last year.

Meanwhile, the company achieved revenue of $16.3 billion in the first half of this year, with Q3 revenue growing 45% quarter-on-quarter to $16.8 billion.

The scale of Anthropic’s strategic bets on computing power is also catching the market’s attention. According to The Wall Street Journal, the company’s total committed spend on computational power will exceed $600 billion, with commitments for compute-related agreements totaling over $518 billion over the next ten years, and negotiations for further expansion of computing resources are underway.

Analysts have noted that such spending levels represent significant valuation variables for potential public investors. Anthropic is reportedly seeking a valuation of $1.5 to $2 trillion for a public listing in November.

Anthropic’s investment in computational infrastructure is approaching its upper limits.

According to The Wall Street Journal, the company’s total potential spend on computational power could reach as high as $750 billion, with up to $86 billion devoted just to leasing NVIDIA chips. In the first half of 2026, Anthropic had already spent roughly $9 billion on compute, with its total computing and infrastructure spend projected to hit $7.33 billion for the full year of 2025.

Reportedly, Anthropic is currently negotiating with UK-based emerging cloud provider Nscale to expand its compute partnership, planning to lease additional resources from the second building at the Monarch Data Center campus in West Virginia, which would add about 460 megawatts of computing capacity using NVIDIA’s latest generation Vera Rubin chips, and retaining priority negotiation rights for a planned third building.

If the deal is finalized, Anthropic’s total contract commitment with Nscale would rise to about $90 billion.

In terms of cloud computing agreements, according to The Wall Street Journal, Anthropic will spend about $111 billion on the Google Cloud agreement, approximately $110 billion on Amazon AWS, and around $31 billion on the Microsoft cloud contract.

Despite staggering losses, Anthropic’s revenue growth curve is equally steep.

According to Reuters, a source who attended Anthropic’s pre-IPO investor meeting in San Francisco revealed that full-year revenue for 2025 is expected to grow 12-fold to about $4.6 billion; Q3 2026 revenue increased 45% quarter-over-quarter to $16.8 billion.

However, high growth comes with continued operating losses. In 2025, the company’s operating loss exceeded $8 billion (excluding various debt reductions related to previous financing rounds).

The company states it achieved adjusted positive profit in Q2 of 2026, but the exact definition of “adjusted” profit remains unclear until the full audited financial data is released.

Anthropic is reportedly preparing for what could be a record-breaking IPO, targeting a valuation between $1.5 trillion and $2 trillion, aiming for a listing window in November this year, though specific dates have yet to be announced. By comparison, the SpaceX IPO previously set the highest market valuation record.

Analysts believe there are several factors in the balance sheet warranting careful consideration by potential investors:

• First, although most of the $80.7 billion net loss is due to accounting charges rather than direct cash outflow, it reflects significant valuation fluctuations in the company’s prior financing instruments;
• Second, over $600 billion in compute-related commitments implies rigid capital demand in the coming years;
• Third, the specific definition of “adjusted positive profit” lacks transparency, and full audited financials have yet to be disclosed.

As the AI compute arms race heats up, the market is subjecting the financial structure of such companies to increasing scrutiny.