What will it take for AMD to compete across the full AI infrastructure stack? At AMD Advancing AI 2026 in San Francisco, StackPane’s Sarbjeet Johal and TECHnalysis president Bob O’Donnell join theCUBE’s Dave Vellante to examine AMD’s strategy, the evolving software ecosystem, and the opportunities for enterprise AI.
AMD’s push toward a complete AI platform
The panel discusses AMD’s Helios platform, the importance of rack engineering, and the partnerships shaping its AI ambitions. Johal frames the keynote around accelerators, the CPU opportunity, and relationships with companies including OpenAI, Anthropic, and Cerebras. O’Donnell considers how the combination of Helios and ZT Systems’ engineering strengthens AMD’s competitive position.
Software remains central to the competition
The conversation examines NVIDIA’s CUDA ecosystem and the progress of AMD’s ROCm software. Johal highlights the depth of NVIDIA’s libraries while the panel explores how model porting and tools such as Triton could influence adoption. For enterprises, the discussion connects hardware choice with the software and engineering needed to put it to work.
Hybrid AI and the future of enterprise computing
Johal, O’Donnell, and Vellante also explore the relationship between established x86 workloads and accelerated computing. They consider how on-premises deployments, client devices, and local inference fit into hybrid AI architectures—and how these choices could affect cloud token costs. AMD’s alliance with Cisco adds another dimension to the discussion of managing distributed AI infrastructure.
The recurring theme is the value of competition: a broader set of credible platforms gives customers more options as they combine existing enterprise systems with new AI capabilities.
Explore the conversation
- 00:06 — AMD’s vision and competitive strategy
- 03:21 — Engineering and integration
- 05:39 — Accelerated computing
- 11:15 — Industry collaboration and competition
- 14:03 — Hybrid AI and co-design
- 21:49 — AMD’s engineering and market position
- 25:15 — x86 and the future of computing
- 28:00 — Final thoughts
Video published by SiliconANGLE theCUBE on July 23, 2026. Watch the original panel on YouTube.