The AI competition is getting more intense almost every week now, and Anthropic just made a move that could seriously strengthen Claude’s position against rivals like OpenAI and Google. The company behind Claude AI has officially partnered with Elon Musk’s SpaceX, giving Anthropic access to a massive amount of computing power through the Colossus 1 data centre. For users who were frustrated with Claude’s limits recently, this announcement probably comes at the perfect time.
Anthropic confirmed on Wednesday that the partnership will allow the company to tap into more than 300 megawatts of compute capacity from SpaceX’s infrastructure. That is a huge jump in processing power, and the company says users will start seeing the impact very soon. Claude Pro and Claude Max subscribers are expected to benefit directly as the extra capacity rolls out over the next few weeks. According to the company, this should improve overall availability and performance across Claude products, especially during heavy usage periods.
The interesting part here is that Anthropic is not just increasing capacity quietly behind the scenes. The company also announced major changes to Claude Code usage limits almost immediately after securing the deal. The five-hour rate limits for Claude Code have now been doubled for Pro, Max, Team, and Enterprise users. On top of that, Anthropic is removing the peak-hour reduction system that previously slowed things down for many subscribers during busy times. However, the weekly quota itself is staying the same for now, so users will still have an overall cap despite the improved short-term usage flexibility.
This move shows how badly AI companies now need infrastructure to survive in the current market. Training and running advanced AI systems is becoming insanely expensive, and firms are racing to secure GPUs and data centre partnerships before competitors grab all the supply. Anthropic seems to be building a very aggressive long-term strategy here. Apart from the new SpaceX collaboration, the company already has large compute partnerships lined up with Amazon, Google, Broadcom, Microsoft, and Nvidia.
The Colossus 1 facility itself has already become a major talking point in the tech world over the last year. Located in Memphis, Tennessee, the data centre was reportedly built by xAI in only 122 days, which is unusually fast for a project of this scale. The facility currently houses over 220,000 Nvidia H100 GPUs, making it one of the most powerful AI-focused compute hubs in the world right now. Financial details of the Anthropic-SpaceX agreement were not shared publicly, but industry analysts believe deals like this are worth billions considering the insane demand for AI hardware.
Anthropic is also increasing API rate limits for Claude Opus models, which is another big signal that the company expects much heavier developer usage going forward. In the highest API tier, input tokens per minute are jumping from 2 million to 10 million. Output token limits are also being increased from 400,000 to 800,000 tokens. That may sound technical for regular users, but for developers building apps and AI tools on Claude, it basically means they can process much larger workloads with fewer interruptions.
The timing of this announcement is important too. OpenAI, Google, xAI, Meta, and Anthropic are all fighting for dominance in both consumer AI products and enterprise services. Claude has gained popularity for its coding abilities and longer context handling, but users have also complained about usage restrictions and inconsistent availability during peak hours. This latest compute expansion feels like Anthropic responding directly to those complaints while also preparing for even larger AI models in the future.
What makes this story even more interesting is the unusual connection between competing AI ecosystems. Elon Musk’s xAI is technically competing in the same industry with Grok, yet infrastructure partnerships like this show how interconnected the AI business has become behind the scenes. Companies may compete publicly, but many of them still depend on shared hardware suppliers, cloud services, and data centre networks to keep their AI systems alive.
According to Anthropic’s official statement, the additional compute resources should start becoming available within this month. Source: Anthropic newsroom announcement.
