Ampere and Qualcomm Collaborate on Advanced AI Server and 256-Core CPU

The partnership aims to deliver power-efficient AI inferencing solutions while unveiling Ampere's latest high-performance processor.

Ampere introduces its 256-core AmpereOne CPU, built on TSMC's 3nm process technology, promising 40% more performance.
The new CPU will feature 12-channel DDR5 memory, enhancing memory access for data center customers.
Ampere and Qualcomm's collaboration focuses on AI inferencing servers using Qualcomm's Cloud AI 100 Ultra accelerators.
The joint server solution aims to efficiently run pre-trained AI models, not compete directly with Nvidia's training-focused chips.
Ampere also forms a UCIe working group to integrate customer-developed IPs into future custom silicon.