Physical AI and Robotics Summit · 13 Oct 2026 · London, United Kingdom
Robots need more computing power to carry out more complex tasks. In line with Moore’s Law, established in the 1960s, this has meant cramming more transistors onto each microchip. But dense microchips threaten to overheat or weigh down industrial machines, drones and autonomous vehicles. Sending decisions to the cloud could be an option, but it isn’t always practical or reliable, especially for systems that need to react safely and immediately. How are engineers working to deliver more intelligence within the constraints of energy, weight and latency? What trade-offs must they consider between power, performance and accuracy? And if adopting a new architecture means that businesses have to rethink their workflows and supply chains, how should they weigh the cost of change against the potential efficiency gains?
Mythic develops analog AI chips built on memory-cell technology, with energy-efficient processing units for running AI models and sensing.
US developer of analog AI chips that compute inside the memory.
Says its chips deliver AI output at up to 80 times lower cost than the latest graphics chips. Honda is a joint development partner on an automotive AI chip, with Lockheed Martin a strategic relationship.
They also get the Frontier Tech Digest every morning and can ask their own questions. What Frontier Tech Intelligence is →