Claude Code, Doubled!
Published on
If you follow the AI world, you know Anthropic had been going through a rough patch over the last month. There was a serious GPU availability crisis that made the usage limits on Claude Code plummet, leaving the Opus model on the Pro plan practically unusable and a lot of people feeling wronged as customers.
Because of that, we saw a lot of people migrating to OpenAI’s Codex (even more so with the launch of the EXCELLENT GPT-5.5 model).
To start fixing that, today Anthropic fired back with a really interesting announcement. They signed a partnership with Elon Musk’s SpaceX (which has already absorbed xAI, his AI company) to increase their compute capacity.
The deal is to use ALL of the capacity of SpaceX’s Colossus data center, which has more than 220,000 NVIDIA GPUs. That potentially adds more than 300 megawatts of new capacity and goes live this very month.
With this absurd injection of infrastructure, they announced three immediate changes for those who use the tools:
- They doubled Claude Code’s “5-hour limits” for the Pro, Max, Team and Enterprise plans (WOO HOO!)
- They removed the limit reduction during peak hours for Claude Code on the Pro and Max plans (WOO HOO[2]!)
- They considerably increased the API rate limits for Claude Opus (WOO HOO[3]!)

Another thing that was, to say the least… curious, and caught my attention, was a detail at the very end of the announcement: they talked about partnering with SpaceX to develop orbital AI compute. Translation: that crazy idea (or is it a little less crazy now?) of Elon Musk’s of having data centers in space. Seeing what some physicists have said on the subject, I still think it’s nuts (for today), but who knows about 5/10 years from now?
On top of that, Anthropic had recently closed deals to add up to 5 gigawatts with Amazon, 5 gigawatts with Google and Broadcom, a 30 billion dollar strategic partnership with Microsoft and NVIDIA, and a 50 billion dollar investment in American infrastructure with Fluidstack. They also announced they’ll expand part of that capacity internationally (in Asia and Europe) to meet corporate customers’ compliance and data residency requirements.
Everybody in the market knows Anthropic is the most fragile company, in terms of the ability to sustain itself in the long run, among the big American AI players (Google, Meta and xAI have other revenue, OpenAI doesn’t, but it’s great at raising infinite funding). However, with this move and with the Pentagon taking them off the blacklist of companies because of Claude Mythos (we talked about that recently in an episode of IA Sob Controle), it’s possible that Anthropic has really come back into the game in full force.