Sponsored by
I opened my editor this week the way I open it every morning, and somewhere between the coffee and the first failing test it occurred to me that the blinking cursor on my screen is now, technically, a SpaceX asset. On Tuesday SpaceX agreed to buy Anysphere, the company behind Cursor, for sixty billion dollars in stock. Days after the biggest IPO in history. A rocket company bought the IDE.
I keep saying that sentence to myself because it refuses to sound normal. A rocket company bought the IDE.
Nobody pays sixty billion for autocomplete.
So what did they actually buy. Not the text editor. You can clone the text editor in a weekend, and people have. SpaceX bought distribution, the millions of developers who already live in Cursor, plus the workflow data those developers generate every time they accept or reject a suggestion, plus the enterprise contracts that turn a beloved tool into a line item. They paid for it with stock that got real about ninety-six hours earlier, when the IPO priced at a hundred thirty-five and ran toward two hundred by the time the deal hit. Fresh currency, spent immediately.
And they wanted it badly. Back in April, SpaceX wrote itself an option: buy Cursor for sixty billion, or walk away and eat a ten billion dollar break-up fee. You do not put a ten-figure penalty on your own indecision unless you have already decided. This was never going to fall through.
The tell is what Cursor was running on.
Here's the part that made me put my coffee down. Cursor, the best-loved coding tool in the world, did not own the intelligence inside it. It ran on other people's models. Anthropic's, OpenAI's, the same labs it was nominally competing with. So Cursor was, in the most literal sense, paying its rivals for the brains that made its own product worth four billion dollars a year in revenue. A spectacular business, built on rented cognition.
That arrangement only ends one way. Either you raise enough to build your own model, which is a different and far more expensive company, or you sell to somebody who already has one. SpaceX, post-xAI-merger, has Grok and it has Colossus, a compute cluster the size of a small nation's grid. So the plan, openly stated, is to rip out the rented brain and bolt in their own. Vertical integration. The wrapper becomes the storefront for the model underneath.
Which tells you where the value actually lived. Not in the editor. Barely in the brand. It lived in the distribution, and distribution is exactly the thing a buyer can absorb.
Coding agents don't have moats. They have leads.
I say this as someone who switches tools roughly every time a better one ships. Copilot used to own this category, something like sixty-seven percent of it. It's down near fifty now. Most engineers I know run two or three of these things at once and feel zero loyalty to any of them. The switching cost is an afternoon. Maybe a long lunch.
In a market like that, being the best does not compound the way it does in software with real lock-in. Cursor was winning, genuinely winning, and the rational endgame was still to sell rather than try to outrun the next clever startup forever. When the lead is real but the moat is not, "best in category" isn't a fortress. It's a really good asking price.
One AI. Every Tool Your Store Actually Needs.
Most e-commerce sellers are paying for 6 to 8 separate tools that don't talk to each other — and spending hundreds of dollars a month just to keep up. StoreClaw replaces your entire stack with one autonomous AI engine that monitors competitors, optimizes listings, automates marketing, and tracks real profit across Shopify, Amazon, and beyond.
It doesn't wait for you to ask. It runs 24/7 in the background, so you wake up to a full dashboard instead of a list of things you forgot to check.
Connect your store, and StoreClaw gets to work — no prompts, no complex setup, no six-app stack.
Free to start. No credit card required.
This is the exit nobody underwrote.
From the investor seat, this is the uncomfortable part. The venture model wants durable franchises, companies that get more defensible as they scale, that go public and compound for a decade. Application-layer AI keeps refusing to be that. It keeps resolving to M&A, where the leader gets swallowed by whoever owns the models and the compute and the balance sheet, because those are the things that actually defend.
If you had pitched me Cursor two years ago, I'd have loved the product and failed to find the moat, and I'd have been right on both counts. The moat never showed up. The return came anyway, sixty billion of it, from a buyer who needed the distribution more than they needed to invent the tech. That's a fine outcome. It's also not the outcome the spreadsheet was modeling. We keep underwriting the next great software franchise and keep getting handed a very expensive takeout instead.
So tomorrow morning I'll open the same editor. It'll work exactly like it did this week, and that is the whole point. Nobody paid sixty billion for the code it writes. They paid because I open it without thinking, and so do millions of other people, and that reflex turned out to be the only durable thing in the building. I spent this entire piece hunting for the moat. It was the door I walk through every morning, not the room behind it.
— SWEdonym