Editorial correction · November 3, 2026: An earlier version equated quadrupling with a 400% increase. Quadrupling is a 300% increase; this correction does not independently verify the underlying adoption figures.
Anthropic passed OpenAI in enterprise adoption in April 2026. By May, it claimed 41% of business AI subscription spending versus OpenAI’s 39.5%, according to Ramp data covering more than 70,000 companies. Anthropic’s annualized revenue reached $47 billion, compared with OpenAI’s $25 to $33 billion.
These figures describe spending and revenue. They cannot establish every buyer’s reason for choosing a vendor. Deployment effort is one explanation worth examining, especially for teams deciding how to put a capable model into everyday use.
Look at the working environment
Anthropic launched Claude Code in May 2025. Within three months it was generating $500 million in annualized revenue. That figure reached $2.5 billion by February 2026 and $8 billion by May.
Claude Code works inside a developer’s environment, where it can read the codebase, write code, run tests, and make changes. For an evaluation, inspect how those operations fit the team’s review and release process. Measure the setup required before a developer can use it safely on a real task.
Pricing also belongs in the test. Compare the cost of completing the same assignment to an acceptable standard, including retries and review time. A lower per-call price is useful only in relation to the work the model can complete.
Claude’s million-token context window gives teams another feature to evaluate with their own code or documents. A large window may change how much material they can provide at once. They still need to test retrieval, output quality, and cost on the actual task before deciding how much supporting infrastructure to build.
Keep adoption figures separate from the explanation
Anthropic quadrupled its business adoption in a single year, while OpenAI’s grew by 0.3% over the same period. A fourfold level is a 300% increase; 0.3% is less than 1%. Those comparisons need consistent definitions of business adoption before they can support a vendor evaluation.
OpenAI expanded into consumer products, advertising, hardware partnerships, and consulting. Its enterprise offering included ChatGPT Enterprise, the API platform, custom GPTs that were deprecated and replaced, and workspace agents launched in April 2026. Anthropic’s offering centered on Claude, the API, and Claude Code.
A product list can help a buyer map the options. It does not show how much friction a particular customer encountered, or whether that friction caused a purchase decision. The practical test is to follow the purchase, configuration, integration, and approval steps for the workflow your team needs.
Include deployment effort in the trial
Begin with a task the model must perform reliably. Then record the work required to make it usable: integration, training, permissions, review, and maintenance. If the proposal includes a six-week training program or a dedicated integration team, put that effort in the business case.
Compare accepted output, total cost, and reviewer effort under the same conditions. Keep benchmark results as one input, then test the candidate in the environment where people will use it.
Anthropic’s reported growth makes its deployment path worth examining. Your own trial can establish whether that path works for your team, with a clearer basis than a revenue comparison alone.