Silicon & Signals: Architecting Next-Gen AI Systems Beyond the Chip

Spanning the full AI compute stack, this session connects silicon innovation, data center infrastructure, software execution, and market strategy. Key topics range from chiplets, memory, and high-speed networking to inference serving and agentic AI runtimes, anchored by a core debate on who will control AI hardware by 2030: custom hyperscaler silicon or merchant vendors.

TOPICS:

• Rack-Scale AI System Architecture & Heterogeneous Compute
• AI Interconnect & Networking
• AI Memory, Packaging & Chip Integration
• AI Inference Architecture & Serving Systems
• Hyperscale AI Infrastructure & Data Centers
• Agentic AI Software Stack & Execution
• Hyperscaler Silicon vs. Merchant Hardware: Who Owns 2030?

SPEAKERS

• Dr. Pooya Jannaty, Director, Generative AI & World Models, NVIDIA, [https://www.linkedin.com/in/pooyaj/ ](https://www.linkedin.com/in/pooyaj/)
• Vamsi Kamanuru, Principle Machine Learning Engineer, Samba Nova Systems, https://www.linkedin.com/in/vamsidharkamanuru/
• Saurabh Shrivastava, Global Head of Solutions Architecture Amazon Web Services (AWS) Enterprise Core Modernization, https://www.linkedin.com/in/saurabhtechleader/
• Jin Kim, CEO, XCENA, https://www.linkedin.com/in/jin-kim-21a5a584/
• Ahmed Shariq, Distinguished Engineer, Ampere Computing, https://www.linkedin.com/in/ahmed-shariq-6bb96367/

Learn More and Request an invite to join: https://community.accel.ai/event/silicon-and-signals