Opinion
When AI meets reality, the real revolution begins
"The next major opportunity in the evolution of AI is emerging: the use of Physical AI," writes Yair Snir, Managing Director at Dell Technologies Capital. "Applying AI at the physical layer of reality can create profound disruption, improve processes, and address complex challenges that until now have remained unsolved."
The conversation around AI currently centers on industry giants such as OpenAI and Anthropic, which operate primarily in the software market and the enterprise sector. These environments already enjoyed a high level of technological maturity and substantial budgets before the current AI wave. Yet the next major opportunity lies beyond the screen. The industries that power our everyday reality, from logistics and manufacturing to infrastructure, energy, transportation, services, and defense, still face technological gaps and a structural labor shortage. It is precisely where physical infrastructure has been left behind that the next major opportunity in the evolution of AI is emerging: the use of Physical AI. Applying AI at the physical layer of reality can create profound disruption, improve processes, and address complex challenges that until now have remained unsolved.
The economic potential is significant. According to data from the International Federation of Robotics, the Physical AI market is expected to grow at an annual rate of more than 30% over the next decade. At the same time, forecasts estimate that by 2035, more than 1.3 billion AI-powered robots and autonomous systems will be operating worldwide, helping address labor shortages and industrial productivity gaps. These figures underscore the need to deploy artificial intelligence at the edge, at the physical layer of reality, and provide solutions to problems that until now simply had no technological answer. There is also a structural advantage. While in the software world a major player can roll out a single update and render an entire product category obsolete, solutions at the physical layer require deep specialization, a complex technology stack, and an understanding of the operating environment itself. This complexity makes them more resilient to changes introduced by the large models. This is the key difference between the current AI wave and its next phase.
We can already see several areas in which this shift is taking place. The first is the transition from language and visual models to world models that combine language, vision, and spatial understanding. In the long term, they may become the brains and eyes of robots and autonomous systems. In the short term, they enable monitoring and action in complex environments such as oil, gas, and water infrastructure, as well as defense applications, where the system must understand the physical context rather than merely process data.
Another area is the autonomous control of swarms of drones and robots. As drones themselves become inexpensive, accessible off-the-shelf products, value is shifting to the software and algorithms that make it possible to manage entire fleets autonomously. Applications can range from urban traffic management and border control to delivery networks and large-scale logistics operations.
Related articles:
- AI sparked a global arms race among tech giants for electric reserves—can we maintain energy independence?
- Startup Nation’s response to October 7 will “improve the prestige of the Israeli high-tech industry,” says DTC
- How should startups and VC funds navigate their operations during times of conflict?
New possibilities are also emerging in engineering and advanced manufacturing. Unlike the software world, where open-source code and information sharing are common, in industries such as CNC, component design, and industrial manufacturing, data and scripts are strategic assets. Assets that major companies and organizations such as Boeing, Apple, Cartier, and Formula1 teams will never expose. AI solutions are therefore needed to automate engineering and manufacturing processes while using sensitive information without exposing it.
At the same time, service and maintenance sectors, such as hospitality and building management, devote substantial resources to manual, repetitive work. AI-powered robotics combined with visual models can perform some of these tasks autonomously and help address labor shortages.
The transition of AI into the physical world is not without challenges. These industries tend to be more conservative, and their pace of adoption is limited by organizations’ ability to implement new technologies. Nevertheless, the urgent need to solve complex real-world problems will ultimately tip the balance. The key to success lies in combining advanced models with a deep understanding of the reality in which they operate. The ability to identify where algorithms meet a genuine industrial need will create the next significant source of value and transform Physical AI from a technology trend into a force that reshapes entire industries.
Yair Snir is Managing Director at Dell Technologies Capital.
