Opinion
Artificial intelligence enters the battlefield – Physical AI is enhancing combat capabilities
Physical AI systems control tangible entities like vehicles, drones, or industrial robots using sensors such as cameras, microphones, and radars. They collect data from their surroundings, process it in real time, and operate accordingly.
The war with Iran, along with the escalating threat across the Middle East, highlights the critical importance of Israel’s technological superiority - not only in cyber or intelligence domains, but also in physical, on-the-ground operations. This is where one of the most profound revolutions of the coming years emerges: Physical AI - artificial intelligence that operates in the real world, beyond the borders of our screens.
This technology holds significant potential to transform the battlefield. AI-powered systems that identify targets, control drones and other platforms with surgical precision, make decisions in real time, and function in extreme conditions are no longer theory. Drones for surveillance and destruction, robots for mine clearance, and defense systems leveraging computer vision and signal analysis are either in current use or under rapid development - and they will reshape combat operations in real time.
These capabilities not only grant Israel a strategic operational edge but also open doors to leveraging military achievements for broader technological and industrial growth. While civilian industries like logistics, manufacturing, and healthcare are leading much of the development in their respective fields, defense needs are pushing technological boundaries further. The link between security and innovation is getting stronger, positioning Israel as a significant player on the global stage.
At the same time, these developments generate a positive feedback loop. When military technologies go commercial, just as GPS and electro-optical systems did in the past, they spread through civilian markets and spark cross-sector innovation. Physical AI could follow a similar trajectory in the coming years. We’re already seeing military-originated solutions spilling into civilian industries, and the line between the two is becoming increasingly blurred.
Until now, we’ve grown accustomed to AI that operates behind screens, such as chatbots, content recommendations, or spreadsheet automations. The next wave is already acting in our physical reality. Physical AI systems control tangible entities like vehicles, drones, or industrial robots using sensors such as cameras, microphones, and radars. They collect data from their surroundings, process it in real time, and operate accordingly. From autonomous delivery machines and robots in operation rooms to unmanned aerial vehicles executing missions independently - AI is no longer just thinking, it’s doing.
Related articles:
According to Precedence Research, the Physical AI market is expected to reach $34 billion by the end of 2025 and skyrocket to $210 billion by 2034. NVIDIA’s CEO Jensen Huang recently remarked that this domain will revolutionize industries like manufacturing and logistics, on a scale of tens of trillions of dollars. This is a profound shift, in which intelligent systems are not just optimizing or replacing manual processes or pre-programmed robotic operations; they are redefining them.
This has been made possible by a convergence of breakthroughs in AI models and algorithmic integration, alongside significant progress in hardware: sensors, cameras, and computing components that have become much more accurate, compact, and affordable. This synergy enables the deployment of Physical AI in traditional sectors that were once hesitant to rely on intelligent technology.
So, what’s the “secret sauce” for Physical AI entrepreneurs? In two words: the data layer. In the realms of text, images, and code, data is abundant, labeled, and readily available. But in the physical world? There’s no Stack Overflow for milling machines, no structured texts that describe friction, heat, deviation, or wear and tear. There’s no Wikipedia for anti-drone systems or open datasets that teach a robot how to handle real-world tools under changing conditions. Every AI-enabled system must build its own data layer almost from scratch - from sensor readings, engineering files, or human error logs, and the learning process is slower, noisier, and highly contextual. Therefore, one must understand not only what’s right, but understanding why it’s right - for a specific machine, in specific conditions, using a particular material. It is precisely this complexity that creates a rare opportunity to build truly differentiated technology.
In such systems, it’s nearly impossible to replicate a competitive moat. They rely on proprietary organizational data, unique machine configurations, and accumulated expertise embedded over time within the system itself. When done right, the direct impact is substantial: improved industrial efficiency, shorter production cycles, enhanced product quality, and sometimes even a global advantage that affects the entire economic value chain. Israel is already home to several companies leading this domain and some that are still in the development stages. ASCOS is building smart industrial applications, Sidis Labs is focusing on AI-driven medical technologies, and LimitlessCNC is bringing AI to the world of semiconductor processing.
The thread connecting systems on the battlefield and those on the factory floor is clear: both require artificial intelligence that not only thinks - but also acts. This is the future of AI, and it can be the future of Israel, if we choose to invest in it today.
Lior Handelsman is a General Partner at Grove Ventures, a venture capital fund managing over half a billion dollars, focused on early-stage investments in AI, deep tech, health-tech, and enterprise SaaS.
