TAIPEI (Taiwan News) — The next wave of AI will be defined not by larger models but by machines capable of reacting safely in the physical world, NXP Semiconductors President and CEO Rafael Sotomayor said Wednesday.
Speaking at Computex in Taipei, Sotomayor argued that future AI systems will require more than powerful reasoning models. To operate safely in the physical world, he said machines must combine decision-making with fast reflexes and energy-efficient processing.
Sotomayor said many of the most advanced AI systems today excel at language and reasoning but still struggle with physical tasks that humans perform instinctively. He cited examples like movement and balance which are far harder for machines.
The CEO compared future AI systems to the human nervous system, where intelligence is distributed rather than concentrated in a single location, per UDN. Just as the brain, cerebellum, and spinal cord handle different functions, physical AI systems will require separate layers for reasoning and real-time reflexes, he explained.

NXP refers to this approach as its “neural axis” architecture, he said. The framework places intelligence closer to sensors and actuators, allowing robots to react instantly to changing conditions.
Low latency is critical because machines operating in the real world often have only milliseconds to respond, Sotomayer explained. A drone avoiding obstacles or a robot recovering its balance cannot wait for a remote AI model to process every decision, he said.
Trust and safety will be equally important. Sotomayer said, unlike software systems, failures in robots and industrial machines can cause real-world damage, making cybersecurity and fault tolerance essential design requirements.
Sotomayer outlined a framework focused on containing failures and protecting systems from attacks. He said trust must be built into physical AI systems from the start because users cannot afford unexpected behavior in safety-critical applications.
He closed by thanking the company’s Taiwan’s partners, saying NXP’s ability to punch above its weight comes from co-innovation with the country’s industry.






