AMD (AMD) AMD Advancing AI 2026 Investor Roundtable summary

Market opportunity and growth outlook

• AI market opportunity is rapidly expanding, with CPU TAM projected to exceed $200 billion by 2030 and expectations to capture over 50% market share in the next 3–4 years.

• Agentic AI workloads and orchestration are driving higher CPU-to-GPU ratios, with future configurations potentially reaching two CPUs per GPU.

• Accelerator TAM growth is significant, requiring adjustments in supply chain and manufacturing capacity to meet demand through 2027–2028.

• Large-scale deployments with foundational model companies like Anthropic, OpenAI, and Meta are now the norm, with no small or pilot projects.

• Demand projections are based on holistic analysis of customer workloads, adoption rates, and capital availability, with actual market growth often outpacing forecasts.

Product launches, partnerships, and technology roadmap

• Helios systems will begin shipping in September, ramping through Q4 and into the first half of next year, aligning with major data center build-outs.

• The Anthropic MI450 engagement targets up to 2 GW, with the first gigawatt ramping aggressively in 2027 and discussions already underway for future MI500 and MI600 generations.

• Collaboration with Cerebras is progressing, with initial deployments expected by year-end and strong integration between platforms.

• ROCm.ai software is positioned as a major leap, enhancing developer productivity and platform accessibility, with significant differentiation expected over the coming months.

• MI500 generation will initiate a transition from electrical to optical interconnects for scale-up networking, with phased adoption and ongoing ecosystem partnerships.

Competitive positioning and market share

• Venice CPUs are optimized for a broad range of AI workloads, delivering top performance and efficiency, and are expected to drive further market share gains.

• The product portfolio is designed for flexibility, supporting both x86 and overall market share growth, with strong adoption across major cloud providers.

• Chiplet architecture enables flexible memory bandwidth and capacity, supporting performance and TCO advantages in both current and future products.

• ROCm.ai and internal AI adoption are accelerating development cycles and enhancing both internal and external productivity.

• Deep relationships with leading AI companies are accelerating innovation and expanding deployment across the portfolio.