Elon Musk’s SpaceX will now power Anthropic’s AI models. The aerospace firm has signed an agreement with Anthropic that allows the latter to use the Colossus 1 supercomputer to run its AI models. Do note that this deal comes at a time when Musk is suing Anthropic’s chief AI rival, OpenAI. Here is what the tech billionaire said about this deal.

Before we get into the details, you should know that Anthropic is also a direct rival to Elon Musk’s AI startup, xAI. Musk has previously called the Dario Amodei-led firm evil.

Elon Musk says Anthropic didn’t trigger his evil detector

On X, Musk replied to a post from Anthropic’s co-founder Tom Brown. The SpaceX CEO explained the reason for coming to this agreement. He stated that he had spent some time with Anthropic’s senior officials recently “to understand what they do to ensure Claude is good for humanity and was impressed.”

According to the tech billionaire, Anthropic’s team convinced him that they “cared a great deal about doing the right thing.” Elon Musk went as far as claiming that he didn’t find any of them evil, and was confident that Claude was going to be safe for humanity. He explained, “No one set off my evil detector. So long as they engage in critical self-examination, Claude will probably be good.”

Now, why does this matter? Elon Musk’s lawsuit against OpenAI alleges that the AI startup abandoned its original non-profit mission to develop safe, open-source artificial general intelligence (AGI) for the benefit of humanity.

Elon Musk believes that Claude can be good for humanity.
Musk further stated that he was fine with giving Anthropic access to Colossus 1 since “SpaceXAI had already moved training to Colossus 2.”SpaceX describes Colossus 1 as “one of the world’s largest and fastest-deployed AI supercomputers.” It said the system was designed for AI training, fine-tuning, inference, or the use of trained models to generate responses, and other high-performance computing workloads. Colossus 1 is said to have more than 220,000 NVIDIA GPUs, including H100, H200 and GB200 accelerators.

Internet reacts to Elon Musk’s Anthropic deal

Before we discuss how Anthropic users benefit from this deal, let’s look at how the internet reacted. Users on X seem to believe that Musk had reached this agreement, perhaps with the goal of helping Anthropic defeat OpenAI.

One user shared a meme from the film Thor: Ragnarok. The meme shows OpenAI telling Elon Musk and xAI (now merged with SpaceX), “You can’t defeat me.” To this, xAI points to Anthropic and replies, “I know, but he can.”

The internet was full of reactions to this deal.

Another person shared a similar meme. The user shows Elon Musk and xAI say, “I don’t care if Anthropic wins, I just need OpenAI to lose.”

Some joked that Musk may just want OpenAI to fall behind.

Do note that Anthropic is believed to have the edge against OpenAI, particularly in the entreprise market. However, the Claude Code maker is struggling with compute resources, which has led to reduced usage limits for users. This is something that Sam Altman and OpenAI have tried to capitalise on by doubling down on its own coding agent, Codex.

Anthropic doubles Claude limits, removes peak hour restrictions

Now that Anthropic had access to more computing power, the AI startup quickly changed its user limits. The company announced that it was doubling Claude Code’s five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans, while also removing peak-hours limit reduction on Claude Code for Pro and Max accounts.

Anthropic also stated that it was also “substantially raising” API rate limits for Opus models.