Anthropic's annualized revenue run rate crossed $9 billion in March, according to the company's investor materials shared during its February funding round at a $185 billion valuation. That figure is up from $4.5 billion at year end 2025 and roughly $1 billion at year end 2024. The growth split is what matters. Approximately 72 percent of that $9 billion is enterprise and developer API revenue. Approximately 18 percent is Claude.ai consumer subscriptions. Approximately 10 percent is Claude Code, the company's coding product, which crossed $1 billion in annualized revenue in February after launching in May 2024.
The enterprise traction has been the part of the story that Anthropic has been less public about. Internal customer counts shared with investors show 47 of the Fortune 100 now have enterprise contracts with Anthropic. Twenty three of those are exclusive or near exclusive on Claude. The category killers include JPMorgan Chase which moved its internal coding assistant deployment to Claude in October, Pfizer which uses Claude for clinical trial document review, Salesforce which embedded Claude across the Einstein platform, and a wave of consulting firms including Bain, McKinsey, and Deloitte that have built Claude into client workflows. Boeing announced a multi year contract in March covering all engineering documentation systems.
The product set behind the growth is narrower than people assume. Claude Sonnet 4.6, released in early April, has become the workhorse model. It scores 81 percent on SWE-Bench Verified for coding, 73 percent on TAU-Bench for tool use, and is priced at $3 per million input tokens and $15 per million output tokens. Claude Opus 4.6 sits above it for the hardest reasoning workloads at $15 input and $75 output. Claude Haiku 4.5 is the fast tier for routing and simple tasks at $1 and $5. The enterprise pricing typically negotiates a 30 to 40 percent discount on those rates above $10 million annual commit, with prepay incentives at $25 million and $100 million tiers.
Claude Code has been the breakout. The product crossed 700 thousand weekly active users in March and is the fastest growing product in Anthropic's history. The base subscription is $20 per month per developer for individual use. The team and enterprise tiers run $80 to $250 per month with usage caps. Code conversion at large engineering organizations has been the surprising part. Stripe announced in February that it had moved roughly 4,500 of its 5,200 engineers to Claude Code as the primary AI assistant. Block, Atlassian, Datadog, and Snowflake have made similar deployments. Cursor, the IDE-first competitor at $9 billion valuation, has 1.4 million users and is growing fast but is increasingly being positioned as a complement rather than a substitute, with many engineering teams running Claude Code in the terminal alongside Cursor in the editor.
The competitive picture has tightened around three players. OpenAI sits at roughly $19 billion in annualized revenue with the bulk on consumer ChatGPT subscriptions and enterprise GPT-4o and the GPT-5 family. Google's Gemini business is harder to break out of cloud revenue but is growing through Gemini Code Assist and the Vertex AI platform, with Gemini 3 expected to be announced at Google I/O May 12-14. Anthropic's pure play position on enterprise has given it a clearer story to CIOs, particularly on data isolation and the company's emphasis on safety research. The model card and behavior documentation that Anthropic publishes for each release has become a procurement tool that risk and compliance teams reference directly.
Cloud distribution is the leverage point. Claude is available natively on Amazon Bedrock and on Google Cloud Vertex AI. Microsoft Azure does not currently host Claude as a first party model, which has been the largest gap. Amazon's $4 billion investment in Anthropic in 2023 and an additional $4 billion in November 2024 made AWS the primary cloud distribution partner. Bedrock revenue tied to Claude was $2.1 billion in 2025, according to AWS earnings supplements, and is on pace to roughly double in 2026. Google's Gemini still leads on Vertex but the gap has narrowed quarter over quarter through 2025.
The hiring and capacity story is more complicated. Anthropic ended Q1 with approximately 2,800 employees, up from 1,800 at the end of 2024. The company has been hiring aggressively in research, infrastructure, and enterprise sales. Compensation packages have been benchmarked at the top of the market with research scientists routinely earning $750 thousand to $1.4 million in total comp and senior research positions running into the $5 million range. Compute remains the binding constraint. Anthropic announced a multi year compute commitment with Amazon worth approximately $11 billion in 2024 and a similar order of magnitude commitment with Google in 2025. Internal capacity utilization has been above 90 percent on Sonnet for most of 2026, which has occasionally limited new enterprise onboarding.
The IPO question is open. CFO Krishna Rao has told investors the company is not actively planning a public offering in 2026 but expects one in the 2027 to 2028 window. The strategic question for buyers is whether to keep accumulating Claude usage or wait for what is expected to be a Claude Opus 5 release in late summer with substantially better long horizon agent performance. Most enterprise procurement officers interviewed for this story said they are buying now and budgeting for the upgrade later, which is the same pattern that drove the original SaaS adoption cycle a decade ago.
The next milestone for the product is the planned Claude for Operations release in May, which will package coding, research, and document workflows into a single enterprise-managed environment.