SpaceX confirmed completion in a regulatory filing. Bloomberg’s Carmen Arroyo and Natasha Mascarenhas reported the close.
The desk covered the binding agreement in June, when SpaceX filed an 8-K and said it expected to complete in the third quarter, subject to regulatory approval.
That approval has now happened, on the schedule the filing set out.
What Cursor says it bought into
Cursor published its own note the same day. It is two paragraphs of strategy and one sentence that matters.
“We will have access to the largest fleet of GPUs in the world, giving us the compute to build stronger models that are also more economical to run,” the company wrote.
The consequence it draws is commercial. That means more capable models at lower cost for customers.
It also dates the process earlier than the filings do. The acquisition completes something that started in April, when Cursor announced a partnership with SpaceXAI to accelerate model training.
Cursor points at Grok 4.6, released on Wednesday, as an early look at what the two can build together.
Note what is missing from the post. There is no figure, no leadership change and no product commitment beyond that model.
The day before, Cursor was the buyer
On 13 August, one day before the deal closed, Cursor announced that Firetiger had joined the company. In December 2025 it bought Graphite.
A company being acquired for $60bn was still acquiring on the eve of completion. That is unusual, and it says something about how the two sides expect the arrangement to work.
Cursor keeps its name, its blog and its own transactions. The legal entity is Anysphere.
Its framing of the future is deliberately narrow. SpaceX builds the computing capacity, and Cursor will be “one place where that intelligence becomes useful”.
Why SpaceX wanted it
The target is coding, and the rivals are named. The deal is part of Musk’s bid to gain ground on Anthropic and OpenAI, per Bloomberg.
SpaceXAI needed the help. It previously had limited business adoption, and it has been through job cuts and restructurings.
Cursor arrives with what SpaceXAI lacked. Its assistant launched in 2023, became one of the fastest-growing startups on record, and sat at the centre of the vibe-coding era.
The two have already shipped together. Grok 4.5 landed in July as an Opus-class model aimed at coding, legal and finance work, and priced below rivals.
Grok Bot followed on 11 August, a product that works like a team of agents fielding assignments through the day.
The customer who is also the target
There is an oddity in the competitive framing worth stating plainly. Anthropic is one of the rivals this deal is meant to help SpaceX catch.
Anthropic is also a customer. It agreed in May to pay SpaceX close to $45bn for computing, which works out at roughly $15bn a year, Axios reported at the time.
Google has struck a similar arrangement for SpaceX compute. So the same infrastructure serves the company’s own models and its competitors’ models.
That is not a contradiction so much as the business SpaceX is now in. Selling compute pays for building models with it.
The number Musk is working to
Musk told SpaceX staff this month that the company “must succeed on the software front”. He also gave them a date.
The desk covered that all-hands, where he said AI revenue would out-earn rockets by September and significantly exceed the rest in the fourth quarter.
Set the $60bn against that timetable and the logic is visible. A rocket company is buying its way to a software business it says will be its largest within weeks.
Investors took the close well. SpaceX shares were up about 3.6% at $136.20.
The company has only had a share price to react with since this summer. Its listing was days old when it made the Cursor takeover official in June.
What the founders get
The four MIT co-founders become billionaires on completion. That is the part of the story that travels furthest and matters least.
The more consequential transfer is the team and the compute. Cursor engineers now have access to SpaceX’s stockpile of advanced AI chips.
Cursor has described the ambition as moving from completing the next few lines of code to building AI teammates that can be given real work.
Whether the compute makes that cheaper is the open question. Cursor says it will, and the company has an unusually direct way to find out.
Owning the chips changes the unit economics rather than the product. The models still have to be worth running.
The competitive squeeze underneath
Cursor is not buying itself out of a hard market. The value in AI coding has been shifting from the model to the layer that orchestrates it.
The desk reported enterprises moving to open-source harnesses that route work to cheaper models, with some claiming 97% cost reductions.
That is the argument for owning the GPUs. If the fight is on cost per task, the company that does not rent its compute has an advantage.
It is also the argument against paying $60bn for a product layer. Both cases rest on the same shift, and only one of them can be right.
What would settle it
Three things, and the first is September. Musk has put a month on AI passing every other SpaceX revenue line, which is checkable against the next results.
The second is pricing. Cursor has promised more capable models at lower cost, and its own price rises are what pushed customers towards harnesses in the first place.
The third is whether Cursor stays recognisable. It kept its name, its blog and its appetite for acquisitions through the close, and none of that is a guarantee about next year.
Get the TNW newsletter
Get the most important tech news in your inbox each week.