Divisions

Kaplan · Olah · McCandlish · Leike

Spans pretraining, alignment science and interpretability. The S-1 says model research and inference engineering share planning cycles, reporting lines and targets.

Source: S-1 prospectus

Ami Vora

Owns Claude.ai, Claude Code, Claude Cowork (which is merging into Claude.ai), Claude Security and industry editions for finance, healthcare, life sciences, legal, education and government.

Source: S-1 prospectus

Rahul Patil

Covers inference, product engineering, infrastructure and security engineering.

Source: Anthropic leadership

Tom Brown

Secures and scales capacity across four chip platforms: AWS Trainium, Google TPUs, NVIDIA GPUs and AMD GPUs. Capacity is managed as one shared pool.

Source: S-1 prospectus

Mike Krieger & Ben Mann

An incubator for experimental products at the edge of Claude's capabilities, expanded in January 2026.

Source: Anthropic

Jack Clark

Combines the Frontier Red Team, Societal Impacts and Economic Research teams. New work includes forecasting AI progress and AI and the law.

Source: Anthropic

Paul Smith

Applied AI teams design custom deployments for large customers. Direct sales teams build on existing adoption inside companies.

Source: S-1 prospectus

Tino Cuéllar

Handles policy, international engagement and government relationships.

Source: Anthropic leadership

Brian Israel

Covers legal and compliance work, plus integrity work such as usage policy enforcement.

Source: Anthropic leadership

Wholly owned subsidiary

Received Class N shares equal to 5% of fully diluted shares, dedicated to the company's public-benefit purpose. The board may add up to 2% more each time the share price rises 50% past the last milestone.

Source: S-1 prospectus

AI services affiliate

An AI services company launched with private equity firms to build Claude solutions for mid-sized companies. It works with Anthropic's Applied AI teams.

Source: S-1 prospectus

Partner program

Covers training, certification and joint go-to-market. By September 30, 2026, more than 150,000 practitioners at over 4,000 organizations were certified.

Source: S-1 prospectus

Training program

Trains customer and partner engineers to deploy Claude in production, with a goal of 10,000 engineers by the end of 2027.

Source: S-1 prospectus

Gated model access

The program through which Claude Mythos Preview, and later Mythos 5, reached selected organizations for cybersecurity work.

Source: S-1 prospectus

Research grants

Funds research, policy work and data infrastructure on AI's economic effects. It has expanded to the UK and Europe.

Source: Anthropic

Headquarters

Headquarters, with international offices across Europe and Asia-Pacific.

Source: S-1 prospectus

Europe

New offices expanding Anthropic's European presence.

Source: Anthropic

Asia-Pacific office #3

Opened as Anthropic's third Asia-Pacific office.

Source: Anthropic

Asia-Pacific office #4

Opened with Theo Hourmouzis as general manager for Australia and New Zealand.

Source: Anthropic

Asia-Pacific office #5

Opened in October 2026.

Source: Anthropic

US · Australia · Canada

Leases US data centers and has capacity leases in Australia and Canada. More countries are being evaluated.

Source: S-1 prospectus