At its Advancing AI 2026 event in San Francisco, AMD unveiled its latest server and AI infrastructure portfolio, projecting a clear roadmap through 2030 and redefining performance metrics for artificial intelligence workloads. The company launched sixth-generation EPYC Venice processors, the Instinct MI400 series GPUs, and Helios rack-scale systems, marking a coordinated push to capture share in the rapidly expanding data center market. AMD also expanded its annual release cadence for CPUs, GPUs, networking hardware, and rack solutions through 2030, projecting a total addressable market of approximately two trillion dollars by that year. A central focus of the announcement was AMD’s architectural strategy for the next generation. The Zen 7 platform, slated for a 2028 release, will be deployed across three distinct EPYC families designated Florence, Ferrara, and Fidenza, rather than a single unified line. These codenames will be applied systematically across AMD’s AI-focused product stack to address varied enterprise and cloud computing requirements. To align with evolving AI infrastructure demands, AMD introduced a novel competitive benchmark measuring processing capacity in terms of AI agents per watt, per dollar, and per rack. While the company’s supplementary documentation notes that these agent counts are estimated using CPU thread resources as a proxy, the metric underscores a strategic shift toward quantifying efficiency in agentic and generative AI deployments. The announcement reinforces AMD’s commitment to delivering scalable, high-performance computing solutions while positioning its architecture to capture sustained growth in the enterprise AI sector.