At a time when AI tools are beginning to take over parts of a developer’s job, an interesting twist is playing out behind the scenes. Cursor, which is a startup building systems that can write and manage code with minimal human input, is now turning to Elon Musk for help to move faster. The company is reportedly planning to use infrastructure from Musk’s AI venture, xAI, to train its next big coding model. It is a move that shows how even companies trying to reduce human effort are still heavily dependent on something else – raw computing power.
From replacing coders to relying on computing giants: Is Cursor struggling?
People familiar with the matter told Business Insider that Cursor’s upcoming model, Composer 2.5, will be trained using tens of thousands of GPUs from xAI. These chips are the backbone of modern AI systems, and getting access to them at scale is no longer easy. In fact, in today’s AI race, computing power has quietly become as important as the models themselves.
For xAI, this is more than just lending infrastructure. It suggests a change in direction. The company, which has been rapidly building out its data centres under a project called Colossus, is now starting to behave like a cloud provide, something companies like Amazon, Microsoft, and Google have turned into a highly profitable business over the years. Newer firms such as CoreWeave and Lambda are also riding this wave by supplying GPUs to AI developers.
The timing is not random. xAI has built one of the largest AI computing setups in a short span, with around 200,000 Nvidia GPUs already in place and plans to scale much further. But having that kind of power comes with its own pressure — it needs to be used efficiently. Internally, the company is still working on improving how effectively its systems are utilised. Partnering with companies like Cursor could help solve that, while also bringing in revenue.
For Cursor, the stakes are equally high. The startup is already being talked about in the same breath as larger AI players and is reportedly eyeing a valuation of around $50 billion. But competition is getting tougher, with companies like OpenAI and Anthropic pushing aggressively into AI coding tools.
At the same time, Cursor is trying to redefine what a developer actually does. Its recently released interface, Cursor 3, moves away from the traditional idea of writing code line by line. Instead, developers can assign tasks to AI agents — one can build a feature, another can fix bugs, while a third handles a separate part of the project. The human role moves to overseeing all of this. This sounds efficient, but it also creates a new kind of complexity. Instead of writing code, developers now have to manage multiple AI agents working at once. Keeping track of what each system is doing — and making sure nothing breaks — becomes the new challenge.
Cursor is trying to address this with features like Automations, where AI agents can start tasks on their own when certain conditions are met. The idea is to reduce the constant cycle of prompting and reviewing that many AI tools still rely on. In simple terms, the company is not just building AI that writes code — it is building AI that knows when to act.
There is also a deeper connection between Cursor and xAI. Earlier this year, xAI hired two former Cursor engineering leaders into key roles, bringing in people who understand how AI coding tools are evolving. That overlap adds another layer to this reported partnership.
Besides, neither company has officially commented on the partnership arrangement yet. So, readers are advised to take the details with a pinch of salt.
