Google and Intel launch co-designed chip amid semiconductor shifts

Google and Intel have co-designed a new custom infrastructure processing unit (IPU) called the E2000. This chip is intended to reduce the load on primary CPUs and handle data-intensive cloud workloads more efficiently. The E2000 will be utilized in Google Cloud’s new C3 Virtual Machines, which run on 4th Gen Intel Xeon Scalable processors. Early testing by Snap Inc. reported a 20% performance increase compared to previous C2 VMs.

Separately, Google’s new Tensor G6 chip, powering the Pixel 11 series, has shown mixed benchmark results. While some 3DMark tests suggested lower performance than the previous generation due to thermal issues in retail demo units, other Geekbench tests indicated significant improvements in specific GPU tasks. The Tensor G6 features a transition to Arm’s latest mobile cores and a new MediaTek modem to improve connectivity.

In the broader semiconductor landscape, Apple is facing supply chain pressures. Commerce Secretary Howard Lutnick has urged the company to avoid using Chinese memory chips, such as those from ChangXin Memory Technologies (CXMT), to mitigate AI-driven chip shortages. Additionally, reports suggest Apple may face hardware shipment reductions due to a global DRAM supply shortage as manufacturers prioritize high-margin AI data center components.