Intel inks Google cloud deal
Intel and Google announced a multiyear collaboration to deploy Intel Xeon–based platforms for Google’s next generation of cloud and AI infrastructure, signalling continued enterprise investment in tried-and-tested hardware stacks. The partnership is as much about operational reliability and compatibility as raw model performance, which matters to IT buyers who prioritise predictable rollouts. (networkworld.com) (verdict.co.uk)
Google just renewed a long bet on Intel at the exact moment the cloud industry has been talking nonstop about custom chips and graphics processors. On April 9, Intel and Google said Google will keep deploying future generations of Intel Xeon processors in its next wave of cloud and artificial intelligence infrastructure. (intel.com) (bloomberg.com) This is not just about the main processor that runs a server. Intel and Google also said they will expand work on custom infrastructure processing units, which are chips that handle plumbing jobs like networking, security, and storage so the main processors are not doing every task themselves. (intel.com) (bloomberg.com) Google already uses Intel chips in its cloud fleet today. Network World reported that Google Cloud’s C4 and N4 virtual machine families already run on Intel Xeon 6 processors, and those instances are aimed at jobs ranging from general computing to artificial intelligence inference, which is the step where a trained model answers real user requests. (networkworld.com) That detail matters because Google has also been building its own alternatives. In April 2024, Google Cloud introduced Axion, its own Arm-based central processing unit, and said Axion would be used for services including BigQuery, Spanner, Bigtable, Pub/Sub, and the YouTube Ads platform. (cloud.google.com) So the picture inside a modern Google data center is not “one winner takes all.” Google is using a mixed stack: its own Axion chips for some general-purpose workloads, Tensor Processing Units for many artificial intelligence jobs, and now a renewed multiyear commitment to Intel Xeon processors plus custom Intel-Google infrastructure chips. (cloud.google.com) (intel.com) (verdict.co.uk) Intel’s pitch here is that artificial intelligence systems still need a lot more than the chips that train giant models. Intel said Xeon processors will keep handling coordination, inference, and general-purpose workloads, while the custom infrastructure chips are meant to improve efficiency and utilization across Google’s systems at scale. (intel.com) (businesswire.com) For Intel, the timing is important. Bloomberg reported that the agreement gives Intel a commitment from one of the world’s biggest data-center buyers to use future Xeon generations, which helps Intel defend a business that has been under pressure from Arm-based designs and from the rush toward graphics processors for artificial intelligence. (bloomberg.com) For Google, the deal looks less like a dramatic change than a decision to keep proven hardware in the parts of the stack where surprise is expensive. Verdict said Google plans to keep deploying multiple generations of Xeon processors across its global infrastructure, which suggests the company wants continuity across several product cycles rather than a one-off purchase. (verdict.co.uk) (intel.com) That is why this announcement landed even without a giant performance claim or a dollar figure. In cloud computing, the chips that win are not always the newest or flashiest ones; they are often the ones that fit the software, the supply chain, and the rollout plan of a company that operates data centers in many regions at once. (networkworld.com) (techcrunch.com)
