Apple is reportedly partnering with Broadcom to develop its own AI chip. By doing so, Apple wants to reduce its dependence on Nvidia.
Apple is developing its own AI chip with Broadcom. That’s what The Information knows. The project is codenamed Baltra, and the chip should be ready for mass production by 2026. Apple reportedly plans to build the AI chip on TSMC’s N3P (3 nm) baking process.
Apple, like just about every technology company today, wants to develop AI applications. Apple collects that functionality under the name Apple Intelligence. To build out their own AI functionality, companies need to train and modify models. That requires a lot of powerful hardware, which de facto comes from Nvidia. However, Nvidia’s chips are pricey and available in insufficiently large volumes.
Ecosystem
Apple wants to build its own AI chip with Broadcom in order to be less dependent on Nvidia. Dependence, however, comes from more than just the chip. Nvidia is successful not only because of the sheer computing power of its Hopper and Blackwell GPUs, but also because of the extensive software ecosystem that has been created around them.
If Apple wants to get started with its own chips, it will also have to provide the framework. Cupertino often chooses to take as many things into its own hands as possible, and this move fits into that vision. With chips , meanwhile, Apple has the necessary experience.