SpaceX Can Buy Cursor for $60 Billion. Or Pay $10 Billion Not To.

SpaceX and Anysphere, the company behind Cursor, announced a partnership today that comes with one of the stranger deal structures in recent tech history. SpaceX now holds an option to acquire Cursor outright for $60 billion before the end of 2026. If SpaceX decides not to exercise that option, it pays Anysphere $10 billion for the work they’ve done together. Either way, Anysphere walks away with a massive payday.

The deal was confirmed by Bloomberg and CNBC on April 21. SpaceX posted about the collaboration on X, stating that “SpaceXAI and @cursor_ai are now working closely together to create the world’s best coding and knowledge work AI.”

What They’re Actually Building

The partnership pairs Cursor’s product team with SpaceX’s compute infrastructure. Specifically, Cursor will train its upcoming “Composer 2.5” model on thousands of GPUs provided by xAI, SpaceX’s AI subsidiary. The hardware in question is Colossus, SpaceX’s AI training supercomputer, which packs the equivalent of roughly one million Nvidia H100 chips.

Cursor CEO Michael Truell posted on X that he was “excited to partner with the SpaceX team to scale up Composer.” Anysphere’s president Oskar Schulz was more specific in his praise, telling reporters that “SpaceX is basically the best company in the world when it comes to building out compute.”

The goal, per both companies, is building “the world’s most useful models” for coding and knowledge work. That’s a broad mandate, but the immediate focus is clear: Composer 2.5 trained at a scale that Anysphere couldn’t achieve on its own infrastructure.

The Companies Involved

Cursor (Anysphere) was founded in 2022 by four MIT students: Michael Truell (CEO), Sualeh Asif, Aman Sanger, and Arvid Lunnemark. The company built the most popular AI code editor on the market, competing directly with GitHub Copilot, Windsurf, and every other AI-assisted development tool. Before this deal, Anysphere was reportedly in talks to raise $2 billion at a valuation above $50 billion.

SpaceX needs less introduction, but the relevant context is the AI angle. In February 2026, Elon Musk merged SpaceX with xAI in a deal valued at $1.25 trillion. That merger turned SpaceX into an AI company in addition to a rocket company. xAI brought Grok, the Colossus supercomputer, and a massive GPU fleet. SpaceX brought the revenue, the launch infrastructure, and the path to what could be the largest IPO in history.

Why the Deal Structure Is Unusual

Most acquisition talks in tech follow a predictable pattern: due diligence, negotiation, signing, regulatory review, closing. This deal flips the script. SpaceX gets to work with Cursor first, evaluate the partnership, and then decide whether to buy. The $10 billion “break fee” is extraordinary by any standard. In traditional M&A, break fees typically run 1 to 3 percent of deal value. Here, the non-acquisition payment is nearly 17 percent of the full purchase price.

For Anysphere, the structure makes sense. They get access to Colossus-scale compute that no amount of fundraising could match, plus either a $60 billion exit or a $10 billion cash infusion that dwarfs any venture round they could raise. The downside risk is essentially zero.

For SpaceX, the option gives them time to evaluate whether Cursor’s team and technology are worth integrating before committing. If Composer 2.5 trained on Colossus delivers transformative results, $60 billion might look like a bargain relative to building the same capability in-house. If the partnership produces good-but-not-great results, $10 billion buys them a trained model and a relationship without the complexity of a full acquisition.

The Bigger Picture

This deal sits at the intersection of several trends that have been building for months.

Compute is the new currency. The companies with the largest GPU clusters are making deals that look more like sovereign partnerships than vendor contracts. Google has DeepMind. Microsoft has OpenAI. Now SpaceX/xAI has (or may soon have) Cursor. The ability to offer a million-GPU training run is leverage that no amount of cash can replicate on a short timeline.

AI coding tools are consolidating. The field went from zero to dozens of competitors in three years. Now the acqui-hire and acquisition phase is starting. Cursor is the biggest target by valuation, but it won’t be the last. Companies that built popular AI developer tools on top of third-party models are natural acquisition targets for the companies that own the compute and the foundation models.

Musk’s AI strategy is becoming clearer. The xAI-SpaceX merger in February looked like a financial engineering play to some observers. The Cursor partnership suggests something more concrete: Musk wants to own the AI development toolchain, from compute (Colossus) to models (Grok, potentially Composer) to developer tools (Cursor). Add in his plans for solar-powered, satellite-based data centers and you get a vertically integrated AI infrastructure stack that runs from orbit to your editor.

What This Means for Cursor Users

If you’re using Cursor today, the immediate impact is probably positive. More compute means better models, and Composer 2.5 trained on Colossus should outperform anything Anysphere could build on rented cloud GPUs.

The longer-term question is what happens if SpaceX exercises the acquisition option. Does Cursor stay model-agnostic, letting you use Claude and GPT alongside Cursor’s own models? Or does it become an xAI-first tool that pushes Grok and Composer while deprioritizing competitors? The history of big tech acquisitions suggests the latter is more likely eventually, even if the initial messaging promises otherwise.

For now, Cursor remains independent and the partnership is focused on model training. If you’re building workflows around Cursor, nothing changes today. But it’s worth watching how the Composer 2.5 launch plays out and whether the xAI integration stays limited to training or expands into the product itself.