Apple may buy AI chip firms as server plans slip

Apple is reportedly considering acquisitions in the AI chip sector as it tries to strengthen server infrastructure that has struggled with some of its own workloads. According to The Information, Apple has been talking with semiconductor companies and bankers about possible deals after running into performance issues with servers based on the M2 Ultra.

Those servers are used for some AI tasks, but the heavier work — including the Gemini model behind Siri AI — is reportedly handled by NVIDIA chips on Google Cloud. Apple is said to have tried shifting that work onto its own servers, but the current setup has not been enough.

A fix does not appear imminent. Bloomberg reported this week that a server chip based on the M7 Ultra will not be ready until 2029, though Apple is expected to upgrade its infrastructure sooner with M5 Ultra chips. Apple had also reportedly planned to launch a next-generation server chip, code-named “Baltra,” this year, but that schedule now appears to have slipped.

Recommended reading

Mira Murati’s lab unveils Inkling, a 975B open model

The company has already made one major chip-related commitment. Last week, Apple struck a deal with Broadcom to purchase $30 billion worth of chips made in the US.

Apple has deep chip design experience, but mostly on the consumer side. Its move into custom silicon began with the $278 million acquisition of PA Semi in 2008. Large takeovers are still unusual for Apple, although that may be changing: it bought AI startup Q.ai for almost $2 billion this year, its second-biggest acquisition after the $3 billion Beats deal more than a decade ago. As of the end of March, Apple had $45.6 billion in cash and cash equivalents.