Anthropic Shifts Strategy to Curb High-Cost API Usage

Anthropic stops unlimited third-party automation on its Pro/Max plans, moving to pay-as-you-go to curb high costs. The change comes as revenue soars and an IPO is planned for 2026.

In a decisive move to protect its financial margins, AI developer Anthropic is ending the unlimited use of its subscription plans through third-party automation tools. The change, effective April 4, 2026, comes after the company discovered that some high-volume users were generating daily computational costs of up to $5,000 while paying flat monthly fees of just $20.

A New Pay-Per-Use Model

The core of the policy shift targets subscribers to the Claude Pro and Max plans. They will no longer be permitted to employ their flat-rate subscriptions for external AI agent frameworks, specifically through the use of OAuth tokens in tools like OpenClaw. Internal analyses revealed these automated processes were consuming substantial capacity, undermining the economics of the subscription model.

Boris Cherny, head of Claude Code at Anthropic, explained that the existing plans were never designed for the extreme workloads generated by external automation. To assist developers with the transition, the company is offering a one-time credit of up to $200 to affected accounts, which must be redeemed by April 17.

Future access to Claude via external interfaces will require a switch to an "Extra Usage" model. This pay-as-you-go system will mirror the company's standard API pricing. Anthropic's own products, including Claude Code and its desktop application, remain unaffected by these new restrictions.

Fueling Growth and a Prospective IPO

This stricter cost control coincides with a period of explosive revenue growth for Anthropic. In March 2026, the company achieved an annualized revenue run rate of $19 billion—a dramatic increase from the $1 billion mark recorded at the end of 2024. Following a funding round in February, Anthropic's current valuation stands at $380 billion.

Concurrently, the leadership team is laying the groundwork for an Initial Public Offering (IPO). Market observers anticipate a public listing as early as the fourth quarter of 2026. The company is also evaluating a dual listing in London, a strategy viewed as a potential workaround for regulatory challenges in the United States.

Building Political and Financial Foundations

Beyond commercial consolidation, Anthropic is expanding its political footprint. Last Friday, the company registered "AnthroPAC" with the Federal Election Commission. This political action committee is intended to support candidates in the 2026 midterm elections, where the regulation of artificial intelligence is expected to be a pivotal issue.

Management has set a clear financial target: achieving positive cash flow by 2027. The transition to a more disciplined commercial framework is designed to strengthen the balance sheet in advance of the anticipated market debut later this year.