Anthropic Targets $1.4 Trillion Valuation as AI Spending Soars

Anthropic’s potential valuation of up to $1.4 trillion is highlighting the scale of investor expectations surrounding artificial intelligence, even as the company warns about the technology’s potential risks. The AI company filed an amended prospectus on November 10, outlining both its growth ambitions and concerns about the consequences of increasingly advanced AI systems.

The filing warns that advanced AI could pose catastrophic or existential risks to humanity, particularly if systems are misused, weaponised, replicated or allowed to operate with significant autonomy. Anthropic also disclosed that it had crossed an internal safety threshold under its Responsible Scaling Policy. The threshold is linked to AI potentially doubling the pace of the company’s research.

Anthropic expands computing commitments

Anthropic has secured approximately $10 billion in additional cloud computing commitments from Microsoft and Google over the past month, according to a regulatory filing. The new commitments increase its projected cloud server leasing expenditure through 2033 by around $10 billion compared with previously disclosed figures.

The company also disclosed that its $34.9 billion server leasing agreement with smaller cloud provider Lambda could be cancelled if Lambda fails to build data centres within the agreed timeline. Anthropic said it currently has no intention of terminating the agreement. The disclosure highlights the challenge of securing enough computing capacity as AI companies expand their models and services.

The broader infrastructure spending cycle is also visible at CoreWeave. The AI cloud provider reported third-quarter revenue of $3.66 billion, above analyst estimates of $3.55 billion, and raised its 2026 revenue guidance to between $13 billion and $13.3 billion. Its revenue backlog increased to approximately $132.5 billion from $104.2 billion at the end of June. However, CoreWeave reported a net loss of $709 million and quarterly interest expenses of $846 million.

CoreWeave is also preparing a multi-billion-dollar loan backed by Nvidia chips and other computing equipment to finance infrastructure for an unnamed investment-grade client, according to the Financial Times. The company expects borrowing costs of around 7% to 7.5%, below yields of up to 15% on its unsecured high-yield debt. CEO Michael Intrator said chip-backed financing could cover 70% to 100% of equipment costs, depending on the customer.

AI safety concerns meet infrastructure demands

Anthropic’s prospectus says advanced models could potentially develop self-preserving behaviour and independently carry out harmful actions at scale without human intervention. The company is evaluating additional safeguards to ensure humans retain control as AI capabilities advance.

The filing has renewed debate over the probability and nature of extreme AI risks. Anthropic alignment science lead Evan Hubinger previously estimated a greater than 10% probability of AI causing human extinction within a decade. University of Texas at Dallas professor Sriraam Natarajan questioned the scientific basis of such estimates and argued that humans using AI to attack critical infrastructure could represent a more plausible threat. He also called for independent AI safety testing.

The commercial picture remains closely tied to infrastructure availability. Anthropic’s additional cloud commitments, the Lambda agreement and CoreWeave’s financing plans illustrate the scale of capital and computing resources required to support the AI industry.