AMD Unveils CPU Roadmap Through 2030, Teases Zen 7 and Zen 8 Architectures
At its Advancing AI 2026 event in San Francisco, AMD CEO Lisa Su unveiled a long-term CPU roadmap extending to Zen 8 in 2030. The company launched sixth-gen EPYC 'Venice' processors, Instinct MI400 GPUs, and Helios rack systems. AMD confirmed Zen 7 'Florence' will launch in 2028, split into three EPYC families (Florence, Ferrara, Fidenza), and Zen 8 'Ravenna' is in development for 2030. The company also detailed MI500-series accelerators targeting a 2,000x inference throughput increase.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page itself is projected from evidence records.
- Current automated evidence projection
Cross-source coverage
Wire timeline
AMD Splits Zen 7 into Three EPYC Families for 2028, Extends Product Cadence to 2030
At its Advancing AI 2026 event, AMD launched sixth-gen EPYC 'Venice' processors, Instinct MI400 GPUs, and Helios rack-scale systems. The company confirmed that its upcoming Zen 7 architecture, arriving in 2028, will be split into three distinct EPYC families: Florence (fresh Zen 7 cores with AI extensions), Ferrara (AI host node), and Fidenza (agentic sandbox product). AMD extended its annual CPU, GPU, networking, and rack cadence out to 2030, and estimated its total addressable market at roughly $2 trillion. The company introduced a new competitive metric claiming the most AI agents per watt, per dollar, and per rack, though these estimates are based on CPU thread resources as a proxy. AMD also replaced TDP with 'Default CPU Power' starting with sixth-gen EPYC. The forward roadmap includes MI500 Series GPUs in 2027 and MI600 Series in 2028.
AMD Splits Zen 7 into Three EPYC Families for 2028, Extends Product Cadence to 2030
At its Advancing AI 2026 event in San Francisco, AMD launched sixth-gen EPYC 'Venice' processors, Instinct MI400 Series GPUs, and Helios rack-scale systems. The company confirmed that the upcoming Zen 7 generation, arriving in 2028, will be split into three distinct EPYC families: Florence (fresh Zen 7 cores with AI extensions), Ferrara (AI host node CPU for Helios 600 racks), and Fidenza (agentic sandbox product). AMD extended its annual CPU, GPU, networking, and rack cadence out to 2030, and estimated its total addressable market at roughly $2 trillion. The company introduced a new competitive metric claiming the most AI agents per watt, per dollar, and per rack, though these estimates are based on CPU thread resources as a proxy. AMD also replaced TDP with 'Default CPU Power' starting with sixth-gen EPYC. Helios racks pairing 72 MI455X GPUs with 18 Venice CPUs are now in production.
AMD Splits Zen 7 into Three EPYC Families for 2028, Extends Roadmap to 2030
At its Advancing AI 2026 event, AMD launched sixth-gen EPYC 'Venice' processors, Instinct MI400 Series GPUs, and Helios rack-scale systems. The company confirmed that the upcoming Zen 7 generation in 2028 will be split into three distinct EPYC families: Florence (fresh Zen 7 cores with AI extensions), Ferrara (AI host node), and Fidenza (agentic sandbox). AMD extended its annual CPU, GPU, networking, and rack cadence to 2030, targeting a $2 trillion total addressable market. The company introduced a new competitive metric claiming the most AI agents per watt, per dollar, and per rack, though these estimates are based on CPU thread proxies. AMD also replaced TDP with 'Default CPU Power' starting with sixth-gen EPYC. The Helios racks pair 72 MI455X GPUs with 18 Venice CPUs, claiming up to 30% more inference tokens per dollar than Nvidia's Vera Rubin NVL72.
Show 3 older updatesHide older updates
AMD Unveils CPU Roadmap Through 2028, Teases Zen 7 and Zen 8 Architectures
At its Advancing AI event, AMD CEO Lisa Su unveiled a long-term CPU roadmap extending to Zen 8. The company announced the 256-core EPYC 9996 with Zen 6 architecture and Venice-X for next year. Su confirmed that Zen 7 'Florence' will launch in 2028, featuring next-gen cores, leading-edge process technology, and new AI compute extensions. Florence will be part of a diversified product family including Ferrara AI host node and Faenza Agentic Sandbox. AMD also revealed that Zen 8 'Ravenna' is already deep in development for 2030. On the accelerator side, AMD detailed its MI500-series Instinct products with next-generation HBM and new interconnect technologies, promising over 2,000 times higher inference throughput in four years. The CDNA Next-based Instinct MI600 family is also in development for 2028.
AMD Unveils CPU Roadmap Through 2028, Teases Zen 7 'Florence' and Zen 8 'Ravenna'
At its Advancing AI event, AMD CEO Lisa Su unveiled a long-term CPU roadmap extending to Zen 8. The roadmap includes the upcoming Zen 6 'Venice' and 'Venice-X' EPYC processors, followed by the Zen 7 'Florence' family slated for 2028. Florence will feature next-gen Zen 7 cores, leading-edge process technology, and new AI compute extensions, and will be joined by specialized nodes like Ferrara and Faenza for diverse AI workloads. AMD confirmed it is already deep in development of Zen 8 'Ravenna' for 2030. On the accelerator side, AMD detailed its MI500-series Instinct products with next-generation HBM and new interconnects, aiming for a 2,000x increase in inference throughput in four years. The CDNA Next-based MI600 family is also in development for 2028.
AMD Unveils CPU Roadmap Through 2028, Teases Zen 7 and Zen 8 Architectures
At its Advancing AI event, AMD CEO Lisa Su unveiled a long-term CPU roadmap extending to Zen 8. The company introduced the 256-core EPYC 9996 with Zen 6 architecture and announced Venice-X for next year. Su confirmed that Zen 7, codenamed 'Florence,' will launch in 2028, featuring next-gen cores, leading-edge process technology, and new AI compute extensions. Florence will be part of a diversified product family including the Ferrara AI host node and Faenza Agentic Sandbox. AMD also revealed that Zen 8, codenamed 'Ravenna,' is already deep in development for 2030. On the accelerator side, AMD detailed its MI500-series Instinct products with next-generation HBM and new interconnect technologies, promising a 2,000x increase in inference throughput over four years. The CDNA Next-based Instinct MI600 family is also in development for 2028.