Google confirmed it will adopt multiple generations of Intel’s central processing units (CPUs) across its AI data centres - expanding a relationship that dates back nearly three decades.

The move will see Intel’s latest Xeon 6 chips deployed for both AI training and inference workloads.

The announcement signals a notable shift in the AI hardware landscape, where graphics processing units (GPUs), particularly from Nvidia, have dominated.

Increasingly, however, CPUs are becoming critical as AI systems scale in complexity.

Amin Vahdat, Google’s chief technologist for AI infrastructure, said in a statement that Intel’s roadmap gives the company “confidence that we can continue to meet the growing performance and efficiency demands of our workloads".

The partnership comes at a time when industry leaders are rethinking how AI systems are built.

Intel CEO Lip-Bu Tan also said in a statement: “Scaling AI requires more than accelerators — it requires balanced systems.”

Alongside CPUs, the two companies are continuing work on infrastructure processing units (IPUs), a specialised chip designed to "offload networking, storage and security tasks from host CPUs".

First developed together in 2022, the IPU helps improve efficiency in data centres by handling background operations such as routing traffic, encrypting data and managing virtualisation.

Intel is manufacturing its newest Xeon processors using its advanced 18A process at its Arizona fabrication plant, part of a broader strategy to regain technological leadership.

In addition to Intel chips, Google has developed its own AI accelerators, known as tensor processing units (TPUs), and recently introduced an Arm-based CPU called Axion.