If you're using OpenClaw to build AI agents, you just got an expensive wake-up call. Earlier this month, Anthropic severely restricted how the viral tool can access Claude, forcing millions of users to either pay up or find alternatives.
The reason? AI labs are bleeding money and need to start turning a profit. Anthropic's Boris Cherny, head of Claude Code, put it bluntly on X: their subscription plans weren't designed for how third-party tools actually use their APIs. Translation: agent builders were getting premium AI access at consumer prices.
This is the beginning of a broader trend. As AI companies face mounting pressure from investors and infrastructure costs, the generous access that fueled the agent boom is getting squeezed. OpenClaw became a phenomenon precisely because it made powerful AI accessible, but that accessibility was never sustainable at scale.
For anyone building workflows around AI agents, this is your signal to budget differently. The era of cheap, unlimited AI access is ending. Labs are drawing lines between casual users and power users, and if your tools fall into the latter category, expect to pay accordingly.
The timing matters too. OpenClaw exploded in popularity this year, meaning Anthropic is making this move right as users are most dependent on the integration. It's a calculated bet that enough users are locked in to accept the new pricing rather than rebuild their workflows.
Bottom line: if AI agents are core to how you work, start planning for higher costs. The labs are done subsidizing third-party innovation.
