Elon Musk Reverses Course and Calls Anthropic the AI Leader - Memeburn

For South African businesses, the deal shows how access to AI tools increasingly depends on a small group of foreign infrastructure providers.

Elon Musk has made a sharp U-turn on Anthropic.

After months of attacking the Claude developer as "woke," hypocritical and unlikely to win the AI race, Musk now says Anthropic has become the industry's clear leader. He also praised its newest models and promised that he wouldn't use his control over computing infrastructure to seriously harm the company.

The warmer language matters because Anthropic isn't just one of Musk's competitors anymore. It's also a major customer.

Musk admits he got Anthropic wrong

Musk acknowledged the reversal in a post on X, writing that he was "clearly wrong about Anthropic."

"They are obviously currently the leader in AI," he added, while praising Anthropic's Mythos and Fable models. The comments represent a dramatic change from his earlier public attacks on the company, as detailed in Business Insider's report on Musk's Anthropic reversal. Earlier in 2026, Musk accused Anthropic of stealing training data, promoting political bias and acting hypocritically. He also dismissed the company's chances of beating rivals such as OpenAI, Google and his own AI operation.

Now, he's describing the company's technology as the best available.

That doesn't mean the rivalry has disappeared. Musk continues to promote Grok as a serious challenger, especially after the release of Grok 4.5. You can read our breakdown of Elon Musk's Grok 4.5 "Opus-class" claims for a closer look at how his model compares with Anthropic's systems.

The compute deal changed the relationship

The friendlier tone emerged after Anthropic signed a major computing agreement with SpaceX.

According to Anthropic's official announcement of the SpaceX compute partnership, the company gained access to more than 300 megawatts of capacity, representing over 220,000 Nvidia GPUs at the Colossus 1 data centre. Anthropic said the added infrastructure would increase usage limits for Claude subscribers and API customers. Reported contract documents indicate that Anthropic agreed to pay around $1.25 billion per month for capacity through May 2029. However, termination clauses may allow either party to leave the agreement with relatively short notice. That creates an unusual relationship.

Musk responded to suggestions that he could simply cut Anthropic off. He said he wouldn't end access in a way that seriously damaged the company, even though Claude competes with Grok.

Praise doesn't remove the business risk

Musk's assurance may calm some concerns, but Anthropic still faces a clear dependency.

A frontier AI company needs more than clever researchers and strong software. It needs enormous data centres, reliable electricity, advanced chips and enough cooling equipment to run those chips around the clock.

Only a small number of companies can provide that infrastructure at the required scale.

We think the real story here isn't Musk's change of heart. It's the growing power held by companies that control computing capacity.

Musk can compete against Anthropic through Grok while earning billions from Anthropic's demand for GPUs. In other words, he can benefit whether customers choose his AI model or one built by a rival.

That helps explain why public criticism may now matter less than commercial cooperation.

What it means for South African AI users

For South African companies, the dispute may feel distant. But the infrastructure behind Claude, Grok and other major models directly affects local pricing, reliability and availability.

A startup in Cape Town or Johannesburg might build its customer service, coding or research workflow around Claude. Yet the servers powering that service sit overseas and may depend on commercial agreements between a handful of American technology companies.

If those agreements change, local customers have limited influence.

This matters because African businesses often access AI as imported infrastructure, rather than technology they control themselves. The Anthropic-SpaceX relationship offers another reminder that model access can depend on corporate negotiations taking place thousands of kilometres away.

It also strengthens the argument for more African investment in data centres, energy capacity and locally hosted AI systems. South Africa has a growing cloud and data-centre sector, but training a frontier model still requires infrastructure on a completely different scale.

Musk's endorsement could shift again

Musk's latest position combines praise with competition.

He has acknowledged Anthropic's technical lead while continuing to argue that his younger AI business could catch up. The recent release of Grok 4.5 shows that Musk hasn't abandoned the race; he's simply recognising the strongest current opponent. What we're watching now is whether the cooperation survives the next major model launch.

Anthropic needs stable computing capacity. Musk wants Grok to win. Those goals can coexist while the infrastructure deal remains profitable, but a closer contest could test that arrangement.

So, is Musk genuinely reconsidering Anthropic, or has a billion-dollar customer simply become harder to criticise?

FAQs

Why did Elon Musk change his opinion about Anthropic?

Musk admitted that he had misjudged Anthropic's progress and now considers it a leader in artificial intelligence. The company's newer Claude models have performed strongly in coding, reasoning and business tasks. Anthropic's commercial relationship with Musk-controlled infrastructure may also have softened the rivalry.

Does Anthropic compete with Elon Musk's xAI?

Yes, Anthropic and xAI are direct competitors in the advanced AI model market. Anthropic develops Claude, while xAI offers Grok across X and other Musk-owned platforms. However, the companies can still cooperate on computing infrastructure while competing for users.

Why does Anthropic need so much computing power?

Training and operating advanced AI models requires thousands of powerful chips, large data centres and substantial electricity. More computing capacity allows Anthropic to improve Claude and support more users without severe usage limits. It also helps the company compete with OpenAI, Google and xAI.