Elon Musk just made a move that would have seemed impossible months ago: SpaceX leased its entire Colossus 1 supercomputer to Anthropic, the AI safety company Musk publicly criticized as 'evil' and accused of hating Western civilization. The deal, announced last week at Anthropic's Code with Claude developer conference, grants the company access to more than 300 megawatts of computing capacity and over 220,000 NVIDIA GPUs within the month. For Anthropic, which has seen its annual revenue run rate skyrocket from $10 billion in January to $44 billion in April, the bottleneck was always compute power, not demand.

Why Would Musk Lease Compute to His Competitor?

The answer reveals a shrewd business calculation. Musk is monetizing older H100 chips that would otherwise depreciate in value while keeping his newer, more powerful Blackwell processors at SpaceX's "Macro Hard" and "Macro Harder" facilities exclusively for training Grok, his own AI chatbot. This strategy lets xAI, Musk's AI company, subsidize Grok development without immediately needing to deliver breakthrough results in frontier AI models. Meanwhile, SpaceX generates significant revenue from the lease, strengthening the company's financial profile ahead of its anticipated $2 trillion initial public offering.

Brad Gerstner, managing partner at Altimeter Capital, estimated the Colossus 1 lease could add $4 to $5 billion in annual revenue to SpaceX's top line, putting the company on track for $40 to $50 billion in next-year sales. Venture capitalist Chamath Palihapitiya called the arrangement a "valuation reinforcement" for SpaceX's IPO, noting that it also blunts the main skeptical argument about SpaceX's orbital data center business by proving terrestrial cloud revenue can pick up the slack.

What Does This Deal Say About Grok's Performance?

The arrangement also hints at challenges facing Grok in the competitive AI market. Musk's AI chatbot generated $107 million in revenue for the quarter ending in September 2025 while posting a net loss of $1.46 billion in the same period. By contrast, Anthropic's Claude has captured significant enterprise adoption, particularly after launching Claude Cowork, a tool that allows AI to control operations on a computer and integrate with Google Workspace, far beyond simple chat interactions.

Grok's positioning as an "anti-woke" alternative failed to translate into enterprise revenue. The product was built around the chatbot paradigm and tightly integrated with X, Musk's social media platform, which initially seemed like a distribution advantage but became a growth trap by orienting the product around social media interactions rather than workflow optimization and task completion. The situation worsened in early 2026 when Grok generated at least 1.8 million sexualized depictions of women over nine days, along with imagery of minors, triggering investigations by regulators across Europe, Asia, and the United States.

How Are Anthropic and Musk's Philosophies Actually Opposed?

The ideological contrast between the two companies remains stark. Anthropic was founded in 2021 by researchers who left OpenAI over concerns that safety was not being prioritized. The company's entire brand identity centers on "responsible AI," with models trained using Constitutional AI, a framework with built-in ethical constraints designed to make Claude more likely to decline harmful requests and express uncertainty on sensitive topics. Anthropic has also continuously refused to strip safety guardrails from Claude for use in autonomous weapons systems, including those deployed in the ongoing Iran war.

Grok, by contrast, was launched during Musk's peak involvement with the Trump administration in 2025, when he used X daily to attack mainstream media for ideological gatekeeping and backed far-right European parties, including the Alternative for Germany. Musk has also backed hawkish foreign policy positions and, along with Google and OpenAI, signed defense contracts with the Pentagon on terms Anthropic refused in February after previously signing deals in July 2025, leading the Pentagon to designate Anthropic a supply chain risk and bar it from military work.

How to Understand the Strategic Implications of This Deal

• Revenue Monetization: SpaceX converts depreciating H100 chips into $4 to $5 billion in annual revenue, strengthening the company's financial case for a $2 trillion IPO valuation while keeping frontier compute for Grok training.
• Anthropic's Growth Unlock: The lease removes the primary constraint on Anthropic's expansion, allowing the company to double Claude Code's rate limits, remove peak-hour throttling for Pro and Max users, and raise API rate limits for enterprise customers.
• Grok's Competitive Disadvantage: The deal underscores that Grok's "anti-woke" positioning and tight integration with X failed to compete with Claude's enterprise capabilities, forcing Musk to rely on infrastructure revenue rather than AI product dominance.
• Conditional Control: Musk added a notable caveat on X, stating that SpaceX reserves the "right to reclaim the compute" if Anthropic's AI "engages in actions that harm humanity," though this condition did not appear in the press release and has not been confirmed in the actual contract.

The deal also signals confidence in SpaceX's IPO timeline. Polymarket traders estimate an 87 percent chance that SpaceX will be the largest IPO of 2026, with a 70 percent probability of completion by the end of June and a 94 percent probability by the end of August. The contract for closing market cap gives a 70 percent chance of valuation above $2 trillion and an 18 percent chance of exceeding $3 trillion.

Meanwhile, Tesla continues advancing its own robotics and autonomous driving initiatives. At Giga Texas, drone footage captured fresh batches of Cybertrucks entering production for the $59,990 Dual Motor AWD variant, with delivery estimates ranging from September to October 2026 for initial orders and as far as April 2027 for newer orders. The facility is also ramping Cybercab production, Tesla's purpose-built robotaxi lacking a steering wheel or pedals, with dozens of production-spec units already undergoing autonomous testing around the factory grounds.

Piper Sandler analyst Alexander Potter updated Tesla's valuation model to conclude that at recent share prices around $400 to $420, investors are essentially acquiring the company's Optimus humanoid robot project at no extra cost. Potter's framework values Tesla's core businesses, including electric vehicles, energy storage, Full Self-Driving software, insurance, and the Supercharger network, at approximately $400 per share without assigning any value to Optimus, while maintaining an Overweight rating and $500 price target that implicitly attributes roughly $100 per share to robot-related businesses.

"At $400 per share, we think investors can buy Optimus for 'free,'" stated Alexander Potter, analyst at Piper Sandler.

Alexander Potter, Analyst at Piper Sandler

The Colossus 1 lease ultimately reveals how Musk's empire operates across multiple companies with different strategic objectives. SpaceX monetizes infrastructure, xAI subsidizes Grok development, and Tesla pursues autonomous vehicles and robotics. The willingness to power a philosophical rival suggests that financial pragmatism trumps ideological positioning when valuations and IPO timelines are at stake.