Anthropic’s revenue has seen explosive growth in 2025–2026, heavily driven by its agentic AI offerings, particularly Claude Code, an autonomous coding agent and tool, and related features like Computer Use which allows Claude to interact with a user’s desktop, browser, files, and applications to complete tasks.
Revenue Trajectory (Annualized Run Rate/ARR)
January 2025: ~$1 billion ARR. Mid-2025: ~$4–5 billion. End of 2025: ~$9 billion. February 2026: ~$14 billion with Claude Code contributing significantly. March 2026: Surged to ~$19 billion. April 2026 recent reports: Exceeded $30 billion ARR, with some months adding billions in run-rate growth. This represents roughly 10x+ annual growth sustained over multiple years, and in some periods, multi-billion-dollar monthly increases.
Enterprise deals including via AWS Bedrock and Google Cloud make up the vast majority—often cited around 80%—with over 500 customers now spending $1M+ annually and eight of the Fortune 10 using Claude. Claude Code launched to the public around May 2025 and quickly became a major growth engine: Reached $1 billion ARR in about six months.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
Hit $2.5 billion+ ARR by February 2026, more than doubling since the start of 2026. Business and enterprise subscriptions quadrupled in the early months of 2026. Enterprise usage now accounts for over half of Claude Code revenue. It has driven a notable share of overall growth; some estimates put it at ~20% or more of total revenue at certain points, with reports of 4%+ of all public GitHub commits authored by it in early 2026.
Anthropic has also expanded agentic capabilities with: Computer use and browser agents; Claude controlling screens, apps, and workflows. Agent Teams and multi-agent coordination. Products like Claude Cowork for white-collar automation. These features emphasize tool use, long context, and autonomous execution, which boost usage-based pricing and enterprise adoption over pure chat interfaces.
Perplexity’s recent 50% monthly revenue jump to ~$450M ARR; post-agent pivot with Perplexity Computer and usage-based pricing is impressive for a smaller player but remains dwarfed by Anthropic’s scale. Anthropic’s agentic push especially in coding and computer control has fueled far larger absolute and relative gains, with Claude Code alone surpassing Perplexity’s total ARR in velocity.
This aligns with industry patterns: agentic systems command premium monetization because they deliver measurable productivity e.g., 20x faster development in some case studies, massive time savings in compliance or security workflows. However, challenges remain—high compute costs can lead to negative margins on heavy usage, and scaling safely while maintaining reliability is key for Anthropic’s constitutional AI focus.
Anthropic’s trends strongly validate the agents = higher revenue thesis seen at Perplexity. The company has moved from a safety-focused research outfit to a revenue powerhouse by productizing agentic tools that enterprises will pay heavily for, particularly in software engineering and workflow automation. Growth has been so steep that some analysts project even higher run rates potentially $100B+ annualized if momentum continues.
Anthropic’s own research on real-world Claude conversations estimates that AI assistance reduces task completion time by around 80% in many cases, with software developers seeing the largest contributions to overall labor productivity about 19% of AI-attributable gains.
Internal Anthropic data shows engineers using Claude in ~60% of their work, reporting a 50% productivity boost up from 20% the prior year, including more output volume and the ability to tackle tasks that wouldn’t have been done otherwise. Pull request merge rates have increased significantly in some cases.
Developer reports and case studies often cite: 40% faster task completion in controlled evaluations of Claude-based copilots; strongest for mid-complexity scaffolding, glue code, debugging, and business logic rather than novel architecture. Up to 2x developer velocity, doubled pull request rates, or 164% increases in story completion for individual users.
In enterprises: One fintech team achieved 2x velocity and +10% test coverage; another platform saw 95–99% R&D time reduction for non-technical users building tools. Broader estimates suggest current-generation models with adoption could boost U.S. labor productivity growth by 1.0–1.8% annually over a decade, with coding as a top area.