Edge AI, Not Cloud AI, to Define the Next Generation of Autonomous Vehicles: NXP

The next generation of autonomous vehicles will increasingly rely on edge AI rather than cloud computing, with real-time intelligence enabling vehicles to make safety-critical decisions within milliseconds, according to Hitesh Garg, who leads NXP Semiconductors' engineering organisation in India.

Speaking at Siemens Realize LIVE APAC 2026, Garg said the industry's challenge is no longer developing more powerful AI models, but bringing AI safely into the physical world, where vehicles, robots and industrial machines must react instantly to changing conditions.

"The next step is AI coming to the physical world," he said, arguing that autonomous systems require a fundamentally different computing architecture from traditional cloud AI.

Garg compared future autonomous vehicles to the human nervous system. Just as the brain handles reasoning while the spinal cord executes instant reflexes, he said future vehicle architectures will combine central AI computing with distributed edge processors capable of making immediate safety decisions while reducing dependence on cloud connectivity for time-critical functions.

For motorists, this means critical functions such as emergency braking, collision avoidance, steering corrections and other advanced safety features can continue to operate with ultra-low latency, even when cloud connectivity is unavailable or too slow to respond.

According to Garg, three principles will define successful edge AI systems: ultra-low latency, low power consumption, and high levels of trust. As vehicles become increasingly software-defined, he said computing architectures must deliver not only intelligence but also functional safety, cybersecurity, and resilience against failures.

"The real world has no undo button," he said, stressing that autonomous systems should be designed to recover safely even when faults occur, rather than assuming failures can be completely eliminated.

Garg said NXP is applying the same architectural approach across multiple industries, including automotive, robotics, drones, and industrial automation, with distributed intelligence allowing systems to make local decisions while remaining coordinated with central computing platforms.

The comments come as automakers increasingly adopt software-defined vehicle architectures that consolidate multiple electronic control units into central and zonal computing platforms, creating the foundation for AI-powered driving, over-the-air software updates, and future autonomous functions. Garg said this shift will make edge AI a critical enabler for advanced driver assistance systems, autonomous driving, and the next generation of intelligent mobility.