Musk Hails Anthropic’s AI Lead After Massive SpaceX Compute Deal

Elon Musk is now publicly praising Anthropic and says he would not intentionally cut the AI company off from SpaceX infrastructure, despite earlier dismissive comments about the rival. The shift matters because Anthropic is already one of SpaceX’s biggest compute customers under a multibillion-dollar deal that ties the two companies together through 2029.

The reversal underscores how quickly the AI market is forcing old competitors into uncomfortable partnerships, with enormous financial stakes and strategic risks on both sides.

What changed between Musk and Anthropic?

The relationship appears to have moved from open skepticism to reluctant admiration. Musk had previously mocked Anthropic’s prospects, but after users on X suggested he might use SpaceX servers to squeeze the company, he responded that doing so would not be his “style.”

In a post on Thursday, Musk said he had been wrong about Anthropic. He also described the company as the current leader in artificial intelligence and argued that no rival had yet produced a model comparable to Mythos/Fable, the systems he referenced in his remarks.

The comments are notable not just for their tone, but for the backdrop: Anthropic is now deeply embedded in infrastructure tied to Musk’s business empire.

How big is the SpaceX-Anthropic deal?

The deal is enormous by any standard. Anthropic signed an agreement in May to purchase 300 megawatts of compute, effectively taking the full output of xAI’s Colossus 1 data center near Memphis, Tennessee, which is now part of the broader SpaceX-linked infrastructure arrangement after Musk’s xAI merged with SpaceX in February.

Under the agreement, Anthropic is expected to pay $1.25 billion per month through May 2029. Over the life of the contract, that works out to roughly $40 billion in revenue for SpaceX’s xAI unit.

Google also signed a separate deal to rent SpaceX infrastructure, with payments of $920 million per month through June 2029.

Why would a rival host a competitor’s AI?

The answer is simple: money, capacity and leverage. SpaceX and the xAI business linked to it gain massive revenue from the contract, while Anthropic gets access to the compute it needs to train and serve frontier models at scale.

In the short term, that arrangement is mutually beneficial. SpaceX earns recurring cash from some of the largest technology companies in the world, and Anthropic gets access to scarce infrastructure at a time when demand for high-end AI compute remains intense.

There is also a strategic dimension. Hosting a major AI customer can give the infrastructure provider a better understanding of how large model builders deploy systems, optimize workloads and manage growth. That doesn’t mean the host sees the models themselves, but it does mean closer operational exposure than a standard arms-length vendor relationship.

How is Musk defending his position?

Musk says his history proves that he does not try to sabotage competitors through infrastructure leverage. In his response, he pointed to Tesla’s patent pledge from 2014, which encouraged outside companies to use Tesla technology in good faith without facing lawsuits.

He also cited Tesla’s decision to open its Supercharger network and charging port design to other automakers. In a similar vein, he said SpaceX launches competing satellite systems without imposing higher prices or discriminatory terms.

Those examples are meant to show a pattern of openness, even if the reality is more complicated.

Musk argued that even his fiercest critics can still use the platform and services he controls, and he said he would not deliberately cripple Anthropic while it is a customer.

That claim comes with obvious caveats. Musk has often used aggressive tactics against opponents, including legal action against OpenAI. Still, his point is that business relationships and competitive rivalries can coexist when the financial incentives are large enough.

What risks does Anthropic face?

Anthropic is not relying on goodwill alone. A contract of this size would typically include provisions governing service continuity, termination rights, performance obligations and financial penalties. If a provider were to cut off service without cause, the consequences would likely be severe.

Beyond the legal protections, SpaceX has a powerful incentive to keep the arrangement intact. Losing a customer paying more than a billion dollars a month would be difficult to justify, especially when demand for compute remains high and alternative customers can be hard to replace.

There are, however, broader strategic concerns. Anthropic is depending on infrastructure controlled by a major competitor in the AI ecosystem. That creates obvious questions about trust, operational transparency and the long-term stability of the relationship.

Could SpaceX learn from Anthropic?

Yes. Even without direct access to model weights or proprietary systems, an infrastructure provider can still glean meaningful clues from how a customer behaves.

By supporting Anthropic at scale, SpaceX’s engineers may learn how frontier AI companies structure workloads, handle reliability challenges and deploy rapidly expanding systems. That kind of operational knowledge can be valuable in a market where execution often matters as much as raw model capability.

The source material also notes Musk’s own acknowledgment in a separate legal context that “distillation” among AI companies is real — a practice in which one model provider tries to infer another’s behavior by flooding it with prompts through fake accounts. Anthropic has accused other model makers of doing this before, and the possibility highlights how fiercely companies guard their systems.

Why does this deal matter for the AI industry?

The Anthropic-SpaceX arrangement shows how AI infrastructure has become a strategic commodity. The biggest model developers need enormous, reliable and expensive compute capacity, and the companies that can supply it gain both revenue and influence.

It also reveals a market where rivals may end up depending on one another even while competing head-to-head. That interdependence can soften hostilities in public, but it can also create tension beneath the surface.

For investors and enterprise customers, the deal is a reminder that the AI race is no longer just about model quality. It is also about power availability, data center scale, contract structure and who controls the underlying infrastructure.

Timeline of the key events

The recent sequence of events helps explain why Musk’s comments drew attention.

How should the market read Musk’s comments?

The safest reading is that Musk is balancing candid admiration with hard business reality. Anthropic is a heavyweight in AI, and SpaceX benefits from a lucrative long-term contract that is difficult to replace.

That does not erase competitive tensions. Musk’s companies remain rivals to Anthropic in the broader race to build and deploy advanced AI systems, and his history shows he is willing to fight aggressively when he believes it serves his interests.

But the size of the deal suggests that both sides have reasons to keep the peace. Anthropic gets the infrastructure it needs, while SpaceX and the xAI unit gain a huge source of revenue and exposure to one of the most important companies in the field.

In practical terms, Musk’s public praise may be less about sentiment and more about signaling that a profitable arrangement will continue.

What happens next?

For now, the arrangement appears to be stable, and both companies have strong incentives to make it work. The real test will come over the remaining years of the contract, as the AI market evolves, model training requirements increase and new competitors attempt to close the gap.

If the partnership holds, it may become a case study in how the most powerful names in tech can be both rivals and business partners at the same time. If it breaks down, it could expose the limits of trust in a market defined by speed, secrecy and scale.

Either way, the story is a reminder that in artificial intelligence, infrastructure is strategy — and sometimes, the enemy becomes the customer.