Intel and Google have intensified their collaboration to promote the use of central processing units (CPUs) optimized for artificial intelligence and to develop tailor-made infrastructure processors, as the evolution of AI uses stimulates new demand for traditional computer chips.
Companies are increasingly moving away from training AI models to deploying them, fueling the need for general-purpose CPU chips designed to support massive workloads.
Under the terms of the agreement announced Thursday, Google, an Alphabet subsidiary, will continue to deploy Intel’s Xeon processors, which support a wide range of tasks such as inference and general-purpose computing. The group will also use the latest Xeon 6 chips from Intel.
Intel and Google will also increase the co-development of custom infrastructure processing units (IPUs). These can take over tasks traditionally assigned to the CPU, thus enabling more efficient computing.
“Scaling AI requires more than just accelerators: it requires balanced systems. CPUs and IPUs are essential to delivering the performance, efficiency and flexibility that modern AI workloads demand,” said Lip-Bu Tan, CEO from Intel.
The explosion in demand for “agentic” AI systems – capable of performing complex, multi-step operations beyond simple chatbot functions – has dramatically increased the computing power requirements of processors.
This renewed demand for CPUs could help Intel consolidate its balance sheet and win over new customers, after the founder ceded market share to its competitors during the first years of the AI boom.
The company said Tuesday it would join Elon Musk’s “Terafab” AI chip complex project, in collaboration with SpaceX and Tesla, to support the billionaire’s ambitions in robotics and data centers.
Intel also plans to regain full control of its manufacturing plant in Ireland, where Xeon server processors are produced, by purchasing the stake previously sold to Apollo Global Management.
