Technology

data processing units

1 synchronized story carry this tag.

AMD Helios challenges NVIDIA at the rack level

AMD announced Helios as a rack-scale platform built around MI455X accelerators and Venice CPUs, while Microsoft committed to deploy Helios at scale on Azure for frontier-model inference and AI services. OpenAI separately planned large deployments later in 2026 and 2027.

Why it matters: Accelerator competition is moving from individual chips to complete racks, networking, cloud integration and the surrounding software stack. A named hyperscaler commitment strengthens delivery evidence without proving customer availability or performance.

Read storyOriginal source · reuters.com