Elon Musk publicly reversed his long-standing criticism of Anthropic on July 9, declaring he was "clearly wrong" about the AI startup and acknowledging it as the current leader in artificial intelligence. The admission marks a dramatic shift from his September prediction that "winning was never in the set of possible outcomes for Anthropic." Musk's candid reversal comes as Anthropic relies on computing resources leased from SpaceXAI, Musk's AI infrastructure venture, raising questions about competitive dynamics in the rapidly evolving AI sector.

What Changed Musk's Mind About Anthropic?

Musk's reassessment centers on Anthropic's technical achievements. He specifically praised the company's Claude Mythos and Fable models, stating that "no company has released a model as good as Mythos/Fable" and expressing confidence that "Mythos 2" would arrive soon. This assessment came even as SpaceXAI launched its own Grok 4.5 model on Wednesday and OpenAI released its GPT-5.6 series on Thursday, both positioning themselves as competitive alternatives.

The timing of Musk's praise is significant given the business relationship between the two companies. In May 2026, Anthropic signed a major compute lease with SpaceXAI for access to the Colossus 1 supercomputer in Memphis, Tennessee. The facility houses over 220,000 Nvidia graphics processing units (GPUs) and operates at 300 megawatts of computing capacity. Under the agreement, Anthropic committed to paying approximately $1.25 billion per month through May 2029, generating an estimated $40 billion in revenue for SpaceX's xAI unit.

Could Musk Leverage This Dependency as a Competitive Weapon?

The compute lease structure raised immediate concerns among observers. The agreement includes a 90-day termination clause, meaning either party could end the arrangement with roughly three months' notice. Some commentators noted that if Musk wanted to disadvantage a rival, he theoretically held the power to cut off Anthropic's access to critical computing resources.

"I would never cut them off in a way that hurt them badly, even as a competitor. That's not my style," Musk stated.

Elon Musk, CEO of SpaceX and Tesla

To support his commitment to fair competition, Musk cited examples from his other companies demonstrating a pattern of openness over competitive advantage:

• Tesla Patents: The electric vehicle manufacturer open-sourced its entire patent portfolio in 2014 to accelerate global adoption of sustainable transportation technology rather than protect proprietary advantages.
• Supercharger Network: Tesla made its exclusive Supercharger charging network available to competing electric vehicle manufacturers, transforming what could have remained a closed ecosystem into shared infrastructure that reduces barriers for EV adoption.
• SpaceX Launch Services: SpaceX launches satellites for competing commercial systems "with no increase in price or use of unfair terms," Musk noted.
• X Platform Discourse: Musk observed that "even my worst enemies attack me on this platform," underscoring his preference for open discourse over retaliation.

These examples illustrate Musk's stated philosophy that long-term technological progress is best served by open competition and infrastructure sharing rather than leveraging market power to stifle rivals.

How Does This Fit Into Musk's Broader AI Strategy?

Musk's public acknowledgment of Anthropic's leadership comes as his own AI ventures advance. SpaceXAI unveiled Grok 4.5 on Wednesday, positioning it as a coding-focused model designed to compete with Anthropic's Claude Code offering. The model entered private testing at SpaceX and Tesla facilities. Meanwhile, Musk confirmed that Grok voice commands will integrate with Tesla's Full Self-Driving (FSD) system by September 2026, allowing drivers to give spoken instructions that the vehicle will remember for future trips.

This integration represents a significant expansion of Grok's role within Tesla's autonomous driving ecosystem. Previously, Grok served as an in-vehicle assistant with location-based reminders and natural-language navigation capabilities, but it had no authority over FSD's actual driving decisions. The planned September update will push Grok into a supervisory role, translating spoken intent directly into driving commands. Tesla VP of AI Software Ashok Elluswamy acknowledged that this capability "opens up an entire area of testing that we have to do," including safety guardrails to prevent drivers from instructing the vehicle to crash.

How Will Grok Voice Integration Improve Tesla's AI Training?

The deeper significance of Grok's FSD integration lies in Tesla's AI training advantage. Every time an owner corrects FSD with a spoken instruction and the vehicle learns and remembers it, that interaction becomes a data point covering an edge case that simulation or scripted testing could not generate. A fleet of millions of Tesla vehicles crowdsourcing hyper-local contextual knowledge builds a layer of geographic and behavioral intelligence that competitors without a comparable fleet cannot replicate at the same speed or scale.

• Hyperlocal Learning: Drivers can teach their vehicle which driveway to pull into, which gate entrance to use, and specific neighborhood details that maps cannot capture, creating personalized navigation profiles for each location.
• Edge Case Coverage: Real-world corrections from millions of vehicles generate training data for scenarios that engineers could never simulate or script manually, accelerating the development of safer autonomous systems.
• Rider Preference Data: Tesla's expanding Cybercab and robotaxi operations in Miami and Austin collect rider profile preferences, allowing vehicles to anticipate user behavior before arrival and optimize the final hundred feet of any trip.

This crowdsourced intelligence advantage compounds over time, making Tesla's autonomous systems increasingly difficult for competitors to match without access to a similarly sized fleet.

What Does Musk's Reversal Signal About AI Competition?

Musk's admission arrives amid intense competition in the AI sector, where compute resources and model capabilities determine market leadership. His public correction of his earlier assessment, combined with his reaffirmation of fair-play principles, suggests a willingness to compete on innovation and performance alone rather than leveraging structural advantages. This stance contrasts sharply with his contentious relationship with OpenAI, where he sued the company in 2024 over its transition from nonprofit to capped-profit status. That lawsuit concluded in March 2026 with a federal court ruling against Musk.

The broader context matters: Musk co-founded OpenAI in 2015 alongside CEO Sam Altman, but their relationship deteriorated over the following decade. Musk left OpenAI's board in 2018 after failed negotiations about the company's structure and path forward, including a rejected proposal to merge it with Tesla. He subsequently founded xAI in 2023, which merged with SpaceX earlier in 2026.

Musk's public praise of Anthropic also comes as he postponed a scheduled CNBC interview on July 10 where he was expected to discuss Tesla, SpaceX's recent initial public offering, and Grok 4.5. SpaceX began trading on June 12, 2026, debuting at $150 per share and raising $85.7 billion in the record IPO.