A real breakthrough could happen when AI is not just the target payload and a tool to help to speed up the classic design process, but used to overcome the classical Von Neumann approach.
Interview Thomas Dombek

Thomas Dombek is Head of the IC and Digital Systems Division at CEA-LIST, Saclay, France. He holds double engineering degrees from École Centrale de Lyon, France, and Technische Universität Darmstadt, Germany, obtained in 1999. He has more than 15 years of experience in the semiconductor industry, where he held research and management positions at Philips, NXP, and ST-Ericsson, focusing on advanced image and video processing architectures and integrated circuits for embedded systems.
Since joining CEA-LIST in 2011, he has been involved in research and innovation in embedded computing architectures and tools. His expertise covers tools for hardware/software co-design, digital system architecture and solutions for embedded signal processing and AI. As head of the division, he oversees research activities spanning cyber-security, efficient AI solutions, advanced design tools, innovative circuit and heterogeneous system architectures and quantum computing.
Why must this challenge exist, and why now?
On existing, stable challenges, incremental approvement can result in competitive solutions. The current pace of change in AI requires solutions to adapt much faster, which is a particular challenge on the hardware side. The required breakthrough innovations need a framework allowing for more risk and speed than classic research programs.
What would a real breakthrough look like?
A real breakthrough could happen when AI is not just the target payload and a tool to help to speed up the classic design process, but used to overcome the classical Von Neumann approach.
What is the biggest challenge for chip design in Europe right now?
The barrier to master design and achieve adoption of a new chip solution is high: investment in competences and tools, turn-around-time for validation, software environment and compatibility.
Knowledge sharing and initial investment facility are key factors for startups to reach an initial technology level required for the required capital raise.