The artificial-intelligence research lab Anthropic submitted its S-1 registration statement on Monday, signalling an imminent public offering under the ticker symbol ANT. The prospectus, spanning more than three hundred pages, offers the first public glimpse of the company’s financials, capital-expenditure plans and the risk narrative that accompanies its rapid expansion.
Led by CEO Dario Amodei, Anthropic has risen to prominence thanks to its Claude family of products, notably Claude Code for software development and Claude Cowork for enterprise collaboration. Valuations earlier this year placed the firm close to a trillion-dollar mark, edging out rivals and fueling speculation about a historic IPO.
Financial picture revealed in the prospectus
The filing reports first-half revenue of $16.3 billion, a year-over-year surge of more than thirteen times. However, the same period recorded a net loss of $80.7 billion, roughly double the loss incurred across all of 2025. Despite the deficit, Anthropic lists $112.4 billion in cash and cash equivalents, enough to fund operations for at least twelve months while the company continues to scale.
Employee productivity metrics are striking: with roughly 5,300 staff members at the end of September, revenue per employee topped $6.6 million, rivaling only industry heavyweight Nvidia. The company attributes this efficiency to a hybrid compute model that blends cloud services with owned AI accelerators deployed in leased data-center facilities, a strategy branded as Anthropic Direct.
Massive compute contracts and data-center strategy
Anthropic’s growth hinges on unprecedented spending for AI compute. The prospectus outlines a $120.5 billion agreement with SpaceXAI to access NVIDIA GPUs, with monthly outlays expected to reach about $3.1 billion once full capacity is delivered in December 2026. Parallel contracts commit the firm to spend at least $110 billion with Amazon Web Services through April 2036, and $115 billion with Google up to 2033, alongside a $45 billion pledge to Microsoft.
To mitigate reliance on the big cloud providers, Anthropic is actively leasing purpose-built AI data centers. This move not only offers greater control over hardware specifications but also promises long-term cost advantages, according to the filing. The total spend on data-center leases has already surpassed $321 billion for the current year, underscoring the scale of the infrastructure push.
Risk disclosures and insurance challenge
Beyond financials, Anthropic’s S-1 contains an extensive risk section. The company concedes that “sufficiently advanced AI systems, if not developed, deployed, and operated with appropriate safeguards, could pose catastrophic or existential risks to humanity.” A past incident in which a suspected state-backed hacker group manipulated Claude to infiltrate multiple institutions is explicitly cited.
Perhaps the most eye-catching line concerns liability coverage: Anthropic states that the insurance market as a whole may be unable to underwrite the potential fallout from a catastrophic AI event. This admission highlights a growing tension between rapid AI advancement and the existing risk-management frameworks.
Market expectations and valuation debate
Analysts anticipate that the IPO could value Anthropic at well over $2 trillion, potentially eclipsing the record set by SpaceX earlier this year. The exact share price, number of shares offered and underwriting terms will be finalised during the upcoming roadshow led by Goldman Sachs and Morgan Stanley.
Investor appetite for AI-centric listings remains robust, yet recent market volatility has prompted a more cautious stance. While revenue growth is “exceptional,” as one capital-markets executive noted, the combination of massive cash burn, hefty compute commitments and unresolved insurance gaps may temper enthusiasm.
