Google expands Intel tie‑up

Google and Intel said on April 9 that they are expanding a multiyear deal to use more Intel Xeon processors and jointly build more infrastructure chips for Google Cloud. (intel.com) The companies said Google Cloud will keep deploying Intel Xeon processors across artificial intelligence, inference, and general-purpose workloads, including Xeon 6 chips in its C4 and N4 instances. Intel also said the pair are expanding co-development of custom application-specific integrated circuit infrastructure processing units, or chips that handle networking, storage, and security jobs. (intel.com) A central processing unit is the general manager inside a server, while a graphics processing unit is the specialist that does the heavy math for model training. An infrastructure processing unit moves support work such as packet handling and storage traffic off the main processor so more of the machine can run customer software. (intel.com) (cloud.google.com) Google already uses Intel chips in several cloud products, and its current C4 and N4 virtual machines are built around Intel Xeon processors plus Google’s Titanium offload system. Google’s documentation says C4 systems can reach up to 200 gigabits per second of network bandwidth per virtual machine and up to 2.2 terabytes of DDR5 memory on Granite Rapids configurations. (cloud.google.com) That matters because artificial intelligence systems do not run on accelerators alone. Intel said CPUs are used for orchestration, data processing, and system-level work, and Google’s Amin Vahdat said central processing units and infrastructure acceleration remain a “cornerstone” from training orchestration to inference and deployment. (intel.com) The announcement lands in a market where Nvidia still dominates the most visible artificial intelligence hardware, but cloud operators are spending more attention on the rest of the rack. CNBC reported that Intel shares rose nearly 5% on April 9 after the news, while Alphabet closed little changed. (cnbc.com) Google and Intel did not disclose financial terms or a timetable for the expanded agreement. Reuters, as cited by other outlets, reported that Google plans to keep using Intel’s new Xeon 6 chips as demand shifts toward a broader mix of computing inside artificial intelligence data centers. (cnbc.com) (igorslab.de) Google has worked with Intel since its early server builds, and the newer cloud machines show how that relationship has changed. The pitch now is not just faster chips, but a server design where Xeon processors, offload hardware, and accelerators divide the work inside the same artificial intelligence system. (cnbc.com) (cloud.google.com)