Claude Borrows a Giant Engine From Its Rival
12:50, 07.05.2026
Anthropic chose an unusual solution for one of Claude Code’s most pressing issues. At the “Code with Claude” event, the company announced that paid users would be granted twice as many hours as the five-hour limit imposed by Claude Code, under the Pro, Max, Team, and Enterprise plans (which have limits on the number of seats). API limits for Claude Opus have also been increased.
The extra capacity reportedly comes from Colossus, the massive xAI supercomputer linked to Elon Musk’s AI empire and SpaceX infrastructure. That makes the move especially strange. Anthropic is not just buying more compute. It is borrowing power from a direct competitor.
Why This Deal Feels Bigger Than a Limit Increase
You can read this as a simple win for developers. Many Claude Code users have complained that paid plans, including the costly Max tier, could run out far too quickly during serious coding sessions. Anthropic also had to admit that bugs in prompt caching made token use worse than expected.
But the bigger story is about pressure. AI labs now need compute as much as they need talent, models, and customers. When users hit limits too often, they do not wait patiently. They test other tools.
That is why Colossus matters. The same data center may support xAI’s own ambitions, Cursor’s future plans, and now Claude Code’s daily workload. Rivalry still exists, but capacity has become a currency.
Our Take
For you, this should mean fewer blocked sessions and more stable work inside Claude Code. It also signals a new AI market where companies may compete in public and rent from each other in private.
Our expert view is simple. Better limits help users today, but they also reveal how fragile the AI boom still is. The winners may not be only the smartest model makers. They may be the ones who can secure enough machines to keep your workflow alive.
Share this article with colleagues who use AI coding tools, follow our social channels for more AI analysis, and read our other stories on how the AI infrastructure race affects your work.