SpaceX said June 16 it's acquiring Anysphere, the company behind Cursor, for about $60 billion in stock — four days after its own blockbuster IPO, with the deal set to close in the third quarter. Cursor is the AI coding tool a huge share of professional developers standardized on. A lot of them picked it for one reason that's now an open question: it was neutral. It let you route a task to Claude, GPT, or Gemini and didn't care which model you used. The company that just bought it owns a model.
The neutrality was the product. Cursor's pitch was never a better model — it was that you didn't have to bet on one. Send the task to whatever's best, keep your data under a zero-retention policy, switch the moment something better ships. That promise carried Cursor from roughly $100 million in annualized revenue in early 2025 to more than $4 billion by June 2026, one of the steepest growth curves software has ever produced. Then SpaceX, which merged with Elon Musk's xAI and its Grok models back in February, agreed to pay $60 billion for it. SpaceX hasn't committed to keeping Cursor model-agnostic, and it has every reason to make Grok the default.
Whether they change the default tomorrow or never touch it doesn't change the lesson. The neutrality was a marketing position; nobody signed a contract guaranteeing it. Developers built a workflow — in plenty of cases a whole engineering org's daily habits — on a middle layer whose independence was a promise the owner could revoke or sell. The owner sold it.
What “model-agnostic” is actually worth
This is the second time this month the same shape showed up. Qualcomm paid about $4 billion for Modularto give customers a way out of Nvidia's lock-in. Now SpaceX pays $60 billion for the most neutral tool in the developer stack, and the question is whether it stays neutral. The neutral middle is valuable precisely because it sits between the giants, which is also why a giant ends up owning it. The position is the asset, and the position is exactly what gets bought.
So “independent” and “open” and “works with everyone” turn out to be features with no warranty. They're true until the cap table changes. Anywhere your stack leans on a tool whose main virtue is that it doesn't take sides, you're holding a promise, not a guarantee, and this month that promise traded hands at $60 billion.
If you sell
If your product's edge is “we're the neutral layer” or “we work with every model,” know exactly what you're holding. Neutrality is a strong wedge and a weak moat. It wins deals against the platforms because buyers are scared of betting everything on one model, and it makes you an acquisition target for those same platforms. The day you get bought, your whole pitch inverts — the thing you sold against becomes the thing you are. Sell the neutrality while it's real, and build something underneath it that survives if the neutrality goes: proprietary workflow, data the customer can't easily recreate, an outcome they can name out loud. Then a change of ownership doesn't hollow out your value the way it's about to test Cursor's.
If you buy
Ask the neutrality question before you standardize a team on any independent tool. Not “is it model-agnostic today” — it is, that's why you're looking at it — but “what happens to me if this company gets bought by someone with a model to push.” Keep an exit. Don't let a tool's independence talk you into the deep integration you'd refuse from a model maker, because it can become one overnight. With Cursor specifically, nothing has changed yet: your routing to Claude and GPT still works, and the deal hasn't closed. Use the months before it does to confirm you could actually move if Grok becomes the default and the other models start getting the slower lane.
SpaceX paid $60 billion for a tool whose main selling point was that it didn't take sides. The buyer takes a side. That's not a contradiction anyone needs to resolve — it's the reason the tool was worth buying. Treat neutral, open, and independent as claims with an expiration date you don't control, and build the part of your business that doesn't depend on them holding.