Panel 2026

Panel: AI+HW for Reconfigurable Computing:
Shaping the Next Decade

Artificial intelligence and hardware are evolving at unprecedented speeds, but not always in sync. Today’s AI systems are constrained by energy, data movement, and system-level inefficiencies, while hardware struggles to keep pace with rapidly changing AI workloads. This growing gap raises a fundamental question for the FCCM community: how should we rethink the co-evolution of AI and hardware to close this gap and sustain future progress?

This panel brings together leading experts from academia and industry to discuss the future of AI+HW co-design for reconfigurable computing. We will explore emerging directions such as AI-in-the-loop design, multi-agent hardware generation, memory-centric architectures, and AI-driven EDA workflows—approaches that collectively aim to close the gap between rapidly advancing AI workloads and hardware capabilities. While AI-driven hardware design is a key focus, it will be examined as part of a broader co-design paradigm that seeks to significantly improve design productivity, efficiency (e.g., intelligence per joule), and scalability for next-generation systems.

Designed as an interactive and thought-provoking discussion rather than a sequence of talks, the panel will explore forward-looking and potentially controversial questions, including:

  • How should AI and hardware co-evolve to bridge the growing gap between rapidly advancing models and system capabilities?
  • Will AI-driven hardware design fundamentally transform the design process, or primarily serve as a powerful productivity enhancement?
  • As AI evolves, will future hardware architectures continue to center on matrix operations, or shift toward new compute, memory, and dataflow paradigms?
  • What role will reconfigurable computing play in the overall AI+HW co-design landscape?
  • How far are we from semi-autonomous or fully autonomous hardware design systems, and what are the key barriers to getting there? A related question: Will FPGA/RTL design still exist in 10 years, or will GenAI automate it away?

Looking ahead, the panel will offer bold predictions: a future of agentic, self-improving design systems, tightly coupled AI+HW co-evolution, and potentially 100–1000× efficiency gains through cross-layer AI+HW optimization. Yet, it will also confront hard realities, such as verification challenges, scalability limits, and the enduring role of human expertise.

Join us for a lively debate that challenges assumptions, sparks new ideas, and helps define the future of FCCM in the age of generative AI.

Moderator: Deming Chen, Abel Bliss Professor of Engineering, UIUC

Panelists:

  • Yiran Chen, John Cocke Distinguished Professor, Duke University 
  • Jason Cong, Volgenau Chair for Engineering Excellence, UCLA
  • Ruchir Puri, Chief Scientist, IBM Research
  • Nupur Shah, Vice President, Altera