Apple explores AI chip acquisition as M2 Ultra falls short for advanced workloads

Apple is exploring a possible acquisition of a chip company to strengthen its push into server processors for artificial intelligence, as internal performance issues and delays in its AI chip efforts weigh on its in-house plans. Reuters reported on July 15, citing The Information, that Apple has approached chip startups to gauge their interest in a buyout and has spoken with bankers about possible deals. The company’s AI servers currently run on internally designed M2 Ultra chips, but those systems have struggled with more demanding workloads. The near-term roadmap uses M2 Ultra and M4 chips for AI server applications, while Apple is aiming longer term for custom processors designed specifically for AI workloads from the ground up. Earlier this year, Apple tested Google’s Gemini models on its own servers as part of a Siri overhaul, but the Mac-based chips could not handle the large model, pushing the company to run parts of the revamped assistant on Nvidia chips in Google’s cloud infrastructure. Apple had originally planned to ship a future AI server chip known internally as Baltra this year, but the project has been delayed. The strategy echoes Apple’s earlier chip playbook: its roughly $278 million acquisition of PA Semi in 2008 helped lay the foundation for the A-series chips used in the iPhone, and the company later applied a similar model in shifting Macs from Intel processors to its M-series. Potential deal targets could command premium valuations, while Apple’s effort adds to the list of large technology companies trying to curb dependence on Nvidia for the most demanding AI computing tasks.