At its ‘Advancing AI 2026’ conference on July 22, AMD will unveil roadmaps for its 2nm server CPU ‘EPYC Venice’ and rackscale AI system ‘Helios.’ Just before the event, AMD secured Microsoft as a new Helios customer, challenging Nvidia’s data center GPU dominance. Helios carries a price tag roughly 40% higher than Nvidia’s Vera Rubin, but its total cost of ownership competitiveness has attracted Meta, OpenAI, Oracle, and Microsoft. EPYC Venice, based on the Zen 6 architecture, is the first HPC processor using TSMC’s 2-nanometer process, featuring up to 256 cores, a 33% increase over the 5th-gen ‘Turin,’ and a new SP7 socket with PCIe 6.0 and 16-channel memory achieving up to 1.6TB/s memory bandwidth. Nvidia claimed on July 21 that its Vera CPU delivers up to 1.9 times the performance of AMD’s EPYC Turin in agentic AI benchmarks. On July 20, AMD announced Microsoft will adopt the Helios rack system for Azure, integrating 72 next-generation ‘Instinct MI455X GPUs’ with EPYC Venice CPUs. Overcoming Nvidia’s CUDA software ecosystem remains the critical hurdle.