Intel–Google Infrastructure Tie

Intel and Google said on April 9 they expanded a multiyear deal to build more of Google Cloud’s next artificial intelligence infrastructure on Intel chips and jointly designed networking silicon. (intel.com) Intel said Google Cloud will keep using Intel Xeon processors for artificial intelligence inference, general cloud computing, and other workloads, and the companies will widen co-development of custom infrastructure processing units, or chips that offload networking and data-movement tasks from central processors. (intel.com) Google has already put Intel’s latest Xeon 6 processors into its C4 virtual machines, which it said were the first broadly available hyperscale cloud instances based on that chip generation. Google said those machines delivered up to 30% better general compute performance and up to 60% better machine-learning recommendation performance than the prior generation. (cloud.google.com) In a cloud data center, graphics processors handle the heaviest artificial intelligence math, while central processors and networking chips move data, schedule jobs, and keep the system fed. Intel and Google are betting that those support layers still matter enough to justify a deeper, multiyear buildout. (intel.com) That pitch lands as cloud companies spread spending across more than one supplier instead of relying on a single stack. Reuters reported Google and Intel framed the deal around rising demand for artificial intelligence-focused central processors and custom infrastructure chips as data centers become more heterogeneous, with more kinds of processors working together. (reuters.com) The partnership is not new. Google Cloud launched C3 machines in 2022 with what it called a custom Intel infrastructure processing unit, describing that chip as a way to move networking, security, and storage work off the main processor so customer applications get more of the server. (cloud.google.com) Google and Intel have worked together on cloud hardware for more than 20 years, according to Intel and Google Cloud product pages. CNBC reported the companies tied the new agreement to multiple future Intel chip generations for Google’s artificial intelligence data centers. (intel.com) (cloud.google.com) (cnbc.com) For Intel, the announcement gives the company a public hyperscale customer commitment one week before Google Cloud Next opens on April 22 in Las Vegas. For Google, it preserves a second path for core infrastructure as artificial intelligence capacity expands beyond graphics processors alone. (community.intel.com) (intel.com)