SpaceX Is Overpaying Wildly for Cursor

BySarah Fraser
$60 billion: Software engineer working at a terminal on a dark desk

SpaceX’s reported deal for the right to acquire Cursor at a $60 billion price looks drastically overvalued because it prices a still-unprofitable, roughly 300-employee coding-tool company as if its growth, margins, and competitive moat will all hold at once. They probably will not.

The best support for the price is Cursor’s revenue growth: secondary-market tracker Sacra estimated about $4 million ARR in 2023, $100 million by late 2024, and roughly $500 million by early 2026, while TechCrunch reported the company was in talks to raise capital at a $50 billion valuation as enterprise growth surged. But even taking those growth numbers at face value, $60 billion is about 120 times $500 million in annualized revenue. That is venture-capital fantasy pricing, not normal software math.

Cursor’s price, size, and revenue

Bloomberg reported that SpaceX had an agreement for the right to acquire Cursor for $60 billion. TechCrunch separately reported that Cursor parent Anysphere was already discussing a financing round at more than $50 billion. The market was clearly willing to suspend disbelief before SpaceX arrived.

The size mismatch is the first red flag. CompWorth puts Cursor at roughly 300 employees. A $60 billion price against 300 people works out to about $200 million per employee, using Bloomberg’s reported deal value and CompWorth’s headcount estimate. That is not a sign of hidden efficiency. It is a sign that buyers are paying for a story.

Revenue is the story, and it is a real one. Sacra’s estimates show Cursor ramping from $100 million ARR in late 2024 to around $500 million by early 2026. TechCrunch also described strong enterprise demand and named rising competition from Claude Code and OpenAI Codex.

But revenue is not profit, which is where a lot of AI-tool valuation arguments become magical thinking. Cursor sells a product whose core experience depends on expensive frontier-model inference. If usage rises quickly, costs can rise with it. That is the awkward part of the AI application stack: a company can grow fast and still fail to capture software-like margins because its suppliers own too much of the value.

This is also why “it’s growing insanely fast” is not enough. A coding assistant that sits on top of Anthropic or OpenAI models is not obviously building the kind of durable moat that justifies tens of billions in strategic value. It may be renting one.

Why the market is turning against the IDE

Cursor became important by making the AI-native IDE feel better than the incumbents. That was a real advantage, and our earlier coverage of Cursor leads AI coding agents on workflow laid out why developers cared. The problem is that the battlefield has moved.



OpenAI’s own product trail shows the direction clearly. It introduced Codex as a coding system, then expanded into the Codex app with a more agentic workflow, then pushed Codex for (almost) everything beyond narrow IDE assistance, and then framed it even more broadly as a productivity tool for general knowledge work. That sequence matters. OpenAI is not treating the IDE as the final destination. It is treating it as one surface among several.

That makes Cursor’s center of gravity weaker than it looks. If developer workflows keep moving toward agents, background tasks, terminals, chat-driven repos, and app-level orchestration, then an IDE-centric company is defending a shrinking choke point. The bet starts to look like paying luxury pricing for a very nice plugin layer.

Claude Code sharpens the same problem from another angle. TechCrunch’s reporting did not portray Cursor as running away with the market; it explicitly highlighted competition from Claude Code and Codex. That matters because the moat was supposed to be workflow quality. Once the model vendors ship the workflow too, the wrapper premium gets thinner fast.

Cursor has kept shipping, including features discussed in our coverage of Cursor Composer 2, but shipping features is not the same thing as owning the category. In AI tools, the category leader can become the feature testbed for its suppliers.

The bull case, and why it still fails

The bull case is simple:

  • Cursor has very fast revenue growth.
  • Enterprise buyers appear willing to pay for it, per TechCrunch’s reporting.
  • AI coding remains one of the clearest commercial uses of frontier models.
  • A strategic buyer could imagine pairing it with a broader software or platform push.

That is the best version of the argument, and it still falls apart on price.

A $60 billion acquisition can work if at least one of three things is true: the company has huge profits, a locked-in distribution moat, or infrastructure-level strategic control. Cursor has none of those in the public reporting here. It has strong growth, yes. It also has supplier dependence, active model-vendor competition, and a core surface that is becoming less central.

The 12-month crash thesis is plausible for a plain reason: the valuation has no room for ordinary disappointment. If revenue growth slows, if inference costs stay stubborn, if Codex and Claude Code keep improving, or if buyers decide agentic coding belongs in terminals and multi-agent apps instead of an IDE tab, the multiple compresses brutally. When a business is priced for perfection and positioned in the blast radius of its own suppliers, “pretty good” can look catastrophic.

That does not mean Cursor is a bad product. It means a $60 billion price assumes a level of durability the market has not earned. For a fast-growing but still structurally exposed AI coding company, that is not aggressive. It is reckless.

Key Takeaways

Further Reading