• AMD began Helios rack production: up to 72 Instinct MI455X GPUs, 6th‑Gen EPYC Venice CPUs, 31 TB HBM4 per rack, Pensando networking. Shipments due by end‑Q3 2026; Azure deployments H2 2026.
  • AMD launched 6th‑gen EPYC 'Venice' CPUs for data‑center AI, offering higher thread density and improved throughput per watt, per dollar and per rack; CEO Lisa Su says Venice is in full production.
  • AMD is seeking 1–2+ year supply deals with Chinese data‑center buyers for server CPUs to secure volumes amid ongoing server CPU shortages and rising AI-driven data‑center demand.