AI BIG Idea Series | Helios: AMD’s Next-Generation Rack-Scale Architecture | Forrest Norrod & Doug Huang

Join us for the final session of last year's Generative AI Big Idea Series with Forrest Norrod, EVP & GM of Datacenter Solutions at AMD, and Doug Huang, President of ZT Systems (now AMD). Forrest has led AMD's datacenter business for over 11 years, from EPYC CPUs to Instinct GPUs, and now Helios. Doug has led ZT's Engineering Platform for over 13 years, and brings unique systems expertise to AMD.
​​​​​​
Topics to be discussed:

• Helios: how did AMD achieve such a tight co-design across CPUs, GPUs, and networking. Rationale for building on Meta’s OCP Open Rack Wide.
• What ZT adds: rack integration, thermal & power engineering, validation, etc.
• Complexity beyond silicon: yield, reliability across thousands of components, and how AMD anticipates volume production challenges.
• Supply chain: how could bottlenecks across HBM, packaging, power, cooling, optics, PCBs, module, rack-level assembly affect deployment velocity.
• OpenAI partnership: what OpenAI contributes in terms of next-gen workload visibility and system-level requirements.
• Next-gen architectures: rack standardization vs. customization, rising power density, etc. How will AMD  differentiate as clusters scale to multi-GW?