SpaceX Locks In $60 Billion Option to Acquire AI Coding Giant Cursor

When SpaceX acquired xAI in February for a reported $250 billion, the move was widely read as Elon Musk consolidating his AI bets under the rocket company’s balance sheet ahead of its IPO. Two months later, SpaceX has expanded its AI ambitions again — this time by securing an option to acquire Anysphere’s Cursor, the AI coding assistant that has become the dominant tool for professional software engineers, for $60 billion.

The deal, confirmed by multiple outlets including Fortune and CNBC, has a dual structure: SpaceX is paying $10 billion now for what it describes as a collaboration to develop “coding and knowledge work” artificial intelligence; the arrangement also grants SpaceX the right to acquire Cursor outright “later this year” for $60 billion. If SpaceX doesn’t exercise the option, it has still paid $10 billion to access Cursor’s product, user base, and technical expertise.

What SpaceX Is Actually Buying

Cursor, built by San Francisco startup Anysphere, is a fork of Microsoft’s VS Code that embeds AI assistance throughout the development workflow — not just autocomplete, but full-context refactoring, test generation, multi-file reasoning, and an agentic mode that can autonomously implement features from a specification. By February 2026, Cursor had crossed $2 billion in annualized recurring revenue, making it the fastest B2B software company to reach that milestone from zero in roughly three years.

Crucially, Cursor’s user base skews toward exactly the engineers SpaceX needs most: it’s strongest in systems programming, aerospace software, and complex embedded development — the exact domains relevant to rocket guidance, Starlink software, and the industrial robotics work SpaceX has been quietly expanding into. In other words, this isn’t purely a financial bet; there’s an internal customer thesis.

The Colossus Angle

SpaceX framed the partnership as combining Cursor’s “product and distribution to expert software engineers” with its Colossus supercomputer, which the company claims has the equivalent compute power of one million NVIDIA H100 chips. The pairing suggests SpaceX intends to train proprietary coding AI models on Colossus — with Cursor as both the deployment vehicle and the source of high-quality training data drawn from millions of professional engineering interactions.

Colossus is currently shared with xAI, which used it to train the Grok 4.3 model released in beta earlier this month. Adding Cursor’s workflow data to that training mix would give SpaceX-xAI a dataset of genuine expert engineering reasoning at a scale that no competitor has published access to. GitHub Copilot draws on open-source repositories; Cursor data, by contrast, comes from closed, proprietary, commercially sensitive codebases that engineers have explicitly chosen to work with AI on.

IPO Backdrop

Timing matters here. SpaceX filed confidentially for a Nasdaq listing targeting a $1.75 trillion valuation and a raise of up to $75 billion, with the listing expected as early as June 2026. A $60 billion option over the world’s fastest-growing AI developer tool adds a compelling narrative to that prospectus regardless of whether the option is ultimately exercised.

The financial logic cuts both ways for Cursor’s founders and investors. The company raised $2 billion from Andreessen Horowitz, Nvidia, and Thrive Capital at a $50 billion valuation just days before the SpaceX deal was announced, suggesting the founding team and investors are comfortable with at least two exit paths: a $60 billion acquisition or continued independent growth toward a potential standalone IPO at a higher multiple.

The Competitive Fallout

The deal reshuffles competitive dynamics across the AI developer tools market — a space that has seen more venture capital than perhaps any sector in tech over the past 18 months.

GitHub Copilot, owned by Microsoft, remains the market leader by install base but has been losing enterprise accounts to Cursor on quality grounds. The prospect of Cursor gaining exclusive access to Colossus training runs — potentially producing models that no competitor can match on complex engineering tasks — changes the calculus for any enterprise evaluating which platform to standardize on.

Replit, which raised $600 million in January, targets a different segment: web-first, collaborative development and vibe-coding for less technical users. JetBrains, with its IDE installed base, has been integrating AI but without a comparable training compute story.

The deeper implication is that AI coding tools are converging toward a model where training compute, user data, and deployment surface must all be owned by the same entity to be truly competitive. SpaceX, through xAI and now Cursor, would own all three if the option is exercised.

What It Means for the Developer Ecosystem

For the roughly 4 million engineers using Cursor today, the deal raises immediate questions about product trajectory and data practices. Enterprise customers in particular will want clarity on whether their codebase interactions — currently used to improve Cursor’s local model context — might feed SpaceX-xAI training pipelines and, if so, on what terms.

Cursor’s team has historically been careful about data boundaries, and maintaining that trust will be essential whether the company is acquired or not. The $10 billion collaboration agreement running through 2026 gives both sides time to work out those details before any acquisition is finalized — but those assurances will need to be explicit, not implicit, for enterprise accounts to stay.

The broader message for the developer tools market: at $60 billion, Cursor would be the most expensive software acquisition in history, surpassing even the largest enterprise software deals. That premium reflects not just current revenue, but the growing belief that whoever owns the interface between human engineers and AI systems owns a strategic chokepoint in how software is built for the next decade.