Elon Musk called Anthropic "misanthropic" in February. Now he's handing the company the keys to one of SpaceX's most powerful data centers. The reversal illustrates how quickly ideology bends to the economics of artificial intelligence.
SpaceX will provide Anthropic access to the full capacity of Colossus 1, a facility containing over 220,000 Nvidia GPUs and more than 300 megawatts of computing power. The deal rolls out within weeks. Musk acquired xAI in a separate move and will operate that AI lab from Colossus 2, a different supercomputer complex.
The timing reveals the pressure points in Musk's empire. SpaceX is expected to file for an initial public offering next month. Colossus 1 sits underutilized, a multi-billion-dollar asset gathering dust as Grok, Musk's chatbot, failed to scale as anticipated. Rather than absorb a massive write-down on idle chip capacity before going public, Musk found a customer. Anthropic gets compute. SpaceX gets revenue.
Anthropic faces the opposite problem. The AI lab experienced 80 times revenue growth in the first quarter of 2026 when it planned for just 10x expansion. CEO Dario Amodei disclosed at a developer conference that the company had imposed usage limits on customers and acknowledged "difficulties with compute." The demand spike exposed a critical bottleneck at the worst possible moment.
Enter the convenience of mutual self-interest. Musk met with senior Anthropic leadership last week and emerged impressed, according to his own statements. Within days, the partnership moved from conversation to announcement.
The deal also carries a pointed subtext in Silicon Valley's ongoing power struggle. Musk is suing OpenAI, the company he co-founded. Anthropic, which was built by defectors from OpenAI, now competes directly with Sam Altman's operation. Altman is Musk's stated nemesis in AI development. The calculation is straightforward: Musk gains leverage against his rival by strengthening Anthropic, a thorn in Altman's side.
"Elon's enemy is Sam. Dario's enemy is Sam," wrote Ben Pouladian, a tech market researcher, on X. "Enemy of my enemy is a compute partner."
Anthropic's valuation has also climbed sharply. When Musk made his "misanthropic" jab in February, the company announced a $380 billion post-money valuation. That figure now appears dated. Current expectations place Anthropic closer to $900 billion, a massive jump that underscores the competitive intensity and capital velocity in the AI sector.
The arrangement includes language about future collaboration on orbital AI compute capacity spanning multiple gigawatts, though details remain sparse. SpaceX's blog post flagged this as part of the formal agreement, suggesting the partnership may expand beyond the immediate data center lease.
What remains unclear is how sustainable this relationship might be. Musk's infrastructure tends to outpace his product adoption curves. OpenAI capitalized on excess SpaceX capacity in 2024 as xAI struggled to populate earlier systems. If Grok eventually catches up to demand, or if Anthropic's growth plateaus, the partnership could dissolve just as quickly as it formed.
For now, both sides have what they need. Anthropic stops turning customers away. SpaceX puts idle assets to work before facing public markets. Musk gets a chance to complicate life for Altman. The AI business moves forward, driven as much by who hates whom as by innovation itself.
Author James Rodriguez: "Three months from calling your competitor evil to leasing them half a data center is pure Musk, and it tells you everything about where real power sits in AI right now."
Comments