New Delhi, June 1 -- For years, when we talked about artificial intelligence (AI), we imagined it living in the cloud- giant data centres full of servers analysing information somewhere far away. Your phone sends inquiries, the cloud thinks and sends an answer back. But a quiet shift is happening. AI is moving out of cloud data centres and entering the real physical world, into machines that move, react, and make decisions in real time. It is not just cloud AI anymore. New terms to signify this move into the real world have emerged, e.g. Edge AI and Physical AI. These encompass autos, humanoids, agricultural combines and even drones.
One of those important areas where this shift is manifesting is inside modern cars - the Software Defined Vehicles (SDV) model. SDVs are defined as vehicle platforms that can be transformed with the help of new software uploaded into them. As an illustration, a Mercedes "C" class model can be transformed into "E" class. All transformation done remotely simply by updating a SW definition into the vehicle components by reprogramming and repurposing.
SDV: A car that behaves more like a smartphone
Traditionally, cars were mechanical machines with electronics added later. Every function, braking, steering, infotainment had its own small computer. These systems were isolated, difficult to update, and rarely talked to each other. This traditional model is completely flipped by an SDV version. Instead of dozens of disconnected control units, an SDV uses centralized computing platforms that act more like the brain of a smartphone. The vehicle's behaviour is increasingly controlled by software that can be updated over time.
Think of it like this: older cars were like calculators, fixed in function while SDVs are like smartphones, they can evolve - this evolution requires intelligence. And that's where AI comes in.
Cloud AI vs Edge AI vs Physical AI
Most people are familiar with Cloud AI. When you ask a voice assistant a question, your request travels to a remote server, gets processed, and comes back as an answer. This works well when speed isn't critical. But a car cannot wait for the cloud to decide whether to brake.
This is where Edge AI enters. Edge AI runs directly on the device, such as, inside the brake controller or Lidar or Radar or any critical ECU. It could process camera feeds, radar signals, and sensor data instantly, without needing to send information across the internet. Edge AI is about speed and reliability and auto correction, assuming a trained model governs the actions the vehicle is taking to a stimuli or obstruction.
Now imagine Edge AI combined with physical action.
A vehicle sees a pedestrian. It predicts movement. It applies brakes in milliseconds. That decision is not theoretical; it changes the physical world. This tight loop between perception, decision, and action is what defines Physical AI. It doesn't just analyse data; it interacts with reality. Why the excitement around Physical AI?
The excitement exists for three main reasons. 1. AI is leaving the screen: For decades, software lived mostly in digital spaces - emails, spreadsheets, social media. Physical AI brings intelligence into machines that move - cars, robots, drones, factories. We are entering an era where software directly shapes the physical world.
2. Real-time intelligence is finally possible: Advances in chips, sensors, and networking mean that vehicles can now process enormous amounts of data locally. A modern SDV can run AI models that were once only possible in large data centres. This allows vehicles to make decisions instantly, a requirement for safety-critical systems.
3. Continuous improvement: Because SDVs are software-driven, they can be updated long after they leave the factory. New features, performance improvements, and safety enhancements can be delivered through over-the-air (OTA) updates. That means, the car you buy today can become smarter tomorrow.
This combination, local intelligence, physical action, and ongoing evolution, is what makes Physical AI transformative.
Why SDVs are the perfect platform for Physical AI
Vehicles are one of the complex consumer machines being built. They operate in unpredictable environments, interact with humans, and must meet strict safety standards. This makes them an ideal testing ground for Physical AI.
An SDV acts as a distributed computing platform on wheels. Cameras, lidar, radar, and ultrasonic sensors generate massive data streams. AI models interpret this information to assist drivers, optimize energy use, improve navigation, and enable advanced driver assistance.
But the vehicle doesn't operate alone. Cloud AI still plays an important role.
While vehicles act as intelligent edge devices, making instant decisions, the cloud aggregates data from thousands of vehicles, trains better AI models and then deploys them to the edge, inside cars, where they run in real time, and sends improvements back to the fleet are. Together, they form a feedback loop.
Interestingly, this collaboration is especially relevant in India, where large-scale mobility data can accelerate innovation in traffic management, smart cities, and transportation planning - who would not like to have a more managed traffic in Bengaluru city!
The bigger picture
Physical AI in vehicles is not just about autonomy. It is about creating machines that can sense, learn, and adapt in dynamic environments. The same technologies powering SDVs are influencing robotics, logistics, smart cities, and industrial automation. For India, this shift presents both opportunity and urgency. As vehicles become software platforms, the value chain moves from mechanical manufacturing toward computing, data, and intelligent systems. Companies that master SDV architectures and Physical AI will define the next generation of mobility.
A future that learns while moving
The reason everyone is talking about Physical AI is simple: we are witnessing AI step out of the virtual world and into motion. Cars are becoming intelligent agents, capable of perception and decision-making in the real world. A decade ago, vehicles were machines we controlled. Today, they assist us. Tomorrow, they will collaborate with us.
Published by HT Digital Content Services with permission from TechCircle.