SpaceX filed its S-1 for an initial public offering this week, and the numbers tell a story far beyond rockets. The 36-page risk factors section buried the headline: Anthropic has agreed to pay $1.25 billion per month — $15 billion annually — for access to SpaceX's Colossus I and Colossus II AI training centers in Memphis, Tennessee.
That sum would nearly double SpaceX's total 2025 revenue of $18.7 billion. The contract runs through May 2029 with a 90-day termination clause — a sign of how fast the AI landscape can shift. "SpaceX is offering AI compute as a service at significant scale," Elon Musk posted on X.
Why this matters
The S-1 filing paints a stark picture of where the money is flowing in the AI industry. SpaceX spent $12.7 billion on AI capital expenditures in 2025 — 61% of its total spend. In Q1 2026 alone, AI capex hit $7.7 billion, compared to just $1 billion on the space division.
Yet SpaceX's AI division is still deep in the red: a $6.3 billion operating loss on $3.2 billion in revenue last year, and a $2.5 billion loss on $818 million in Q1 2026. On the other side, Anthropic is cruising toward its first quarterly operating profit, with revenue expected to hit at least $10.9 billion — more than double Q1's $4.8 billion.
Microsoft pulls back from Claude Code
In a parallel move, Microsoft is planning to remove most Claude Code licenses from its developers and push them toward GitHub Copilot CLI instead. An internal memo from Rajesh Jha, Microsoft's EVP of Experiences + Devices, confirmed the shift, with the cutoff set for the end of June — the last day of Microsoft's fiscal year.
The decison is both strategic and financial. While Claude Code proved popular among Microsoft engineers, it had been competing with Microsoft's own Copilot CLI. "Claude Code was an important part of that learning," Jha wrote. "At the same time, Copilot CLI has given us something especially important: a product we can help shape directly with GitHub."
The irony is that despite cutting Claude Code, Microsoft has been increasing its overall Anthropic usage. Reports indicate the company is in early talks to rent Azure servers powered by its Maia 200 chips to run Claude models.
Government regulation struggles to keep pace
President Trump postponed signing an AI executive order last week, reportedly saying he "didn't want to do anything that's going to get in the way of" America's AI lead over China. The order, which would have established government oversight and access rules for new AI models, is now in limbo.
Meanwhile, the White House has approved a $9 billion request to buy advanced AI chips for the CIA and NSA, which lack the computing capacity to run the latest models. Congress still needs to approve the funds.
Pope Francis issued his first major manifesto on Monday, calling for a new legal and ethical framework to govern AI — the first papal document to address artificial intelligence directly.
Anthropic expands security access
Anthropic has made its Mythos Preview security tools more broadly available through Project Glasswing. The company says Mythos has found vulnerabilities "in every major operating system and web browser." However, CISA — America's central cybersecurity agency — has reportedly been left out of the initial access, raising questions about the administration's priorities.
The bottom line
Two days worth of headlines tell a coherent story: the AI industry is in an unprecedented infrastructure arms race. SpaceX has emerged as a compute superpower, Anthropic is sprinting toward profitability, Microsoft is navigating complex partner-competitor dynamics, and regulators are still playing catch-up.
