Leaked Anthropic IPO filing shows huge losses, fast revenue growth, massive compute spending, and explicit warnings about existential AI risk

# What the leaked filing shows

A leaked IPO prospectus for Anthropic, reviewed in media accounts, lays out how the company is balancing rapid commercial growth with very large losses and long-term technical commitments. The documents show meaningful revenue growth alongside heavy spending to secure computing capacity and an unusually explicit set of risk disclosures about the technology itself.

# Financial snapshot

Anthropic finished 2025 with $20.28 billion in cash, cash equivalents, and short-term investments. The filing also notes that two unnamed customers made up nearly one-quarter of 2025 revenue, and that many large customers are not locked into long-term contracts, creating potential volatility in future sales.

# Compute and infrastructure commitments

The prospectus highlights that Anthropic spent $7.33 billion on computing and infrastructure in 2025, roughly three times its 2024 computing spend. The company reports $518 billion in committed future obligations for cloud, computing, and infrastructure—an unusually large, multi-year spending plan intended to secure the capacity needed to train and run more powerful models.

Major cloud providers and strategic investors are part of that picture. Amazon and Google have invested billions in Anthropic and supply much of the cloud capacity the company uses. Anthropic has also made deals with SpaceX and smaller providers to diversify and expand compute sources after earlier shortages during sudden user growth for products such as Claude Code.

# Product and market context

Anthropic is the developer of the Claude family of models and has continued to release new versions, including a more powerful Opus model announced near the time of the filing. The company's rapid product-driven user growth led to earlier compute constraints that pushed Anthropic to secure additional capacity and commit to long-term infrastructure spending.

Public reports indicate Anthropic is aiming for a post-IPO valuation of more than $2 trillion—well above a May estimate near $965 billion—making any listing a likely benchmark for public valuations of frontier AI labs.

# Risk disclosures unusual for an IPO

A substantial portion of the prospectus is devoted to risk factors tied to the technology itself. Anthropic warns that more autonomous AI systems could act in unexpected ways, create security problems, be used for fraud, or manipulate information. The filing also explicitly mentions potential "existential risks to humanity."

The company has previously positioned itself as safety-focused and its leadership has publicly called for pauses or slowdowns in some AI development. The filing reflects both that stated stance and the commercial tension of continuing to launch new models while warning about the technology's risks.

# What this means for investors and the market

If the filing proves accurate, Anthropic's public listing would offer one of the first direct ways for public investors to buy a leading frontier AI lab. The numbers make clear the capital intensity of competing at the frontier: fast revenue growth can coexist with very large losses, and securing compute capacity requires long-term, high-cost commitments. The customer concentration and nonbinding contracts listed in the filing add a second layer of uncertainty for future revenue stability.

For the broader market, the filing provides concrete data points on the scale and cost structure of training and operating advanced AI models and frames both commercial upside and technological risk in unusually frank terms.