# The Physical AI Opportunity: Investing Beyond Software

Author: Norbert Biedrzycki
Published: 2026-11-15
Language: en
URL: https://norbertbiedrzycki.pl/en/the-physical-ai-opportunity-investing-beyond-software/
Summary: For six years, the AI investment story has been about screens: text, images, code and software. The next chapter is about atoms: robots, factories, vehicles, laboratories and the energy that…

**For six years, the AI investment story has been about screens: text, images, code and software. The next chapter is about atoms: robots, factories, vehicles, laboratories and the energy that powers them.**

When I wrote about autonomous driving in 2020, I described it as an extraordinary bet with uncertain timing. Six years later, Waymo is completing hundreds of thousands of paid driverless rides every week across several American cities. The timing was longer and the cost higher than the optimists predicted. But the technology arrived.

That pattern is worth keeping in mind, because the next phase of AI investment is increasingly connected to the physical world.

**What is changing**

Several developments are converging. Foundation models are being trained not only on text and images but on video, sensor data and robot interactions, giving machines a much more general ability to perceive and act. Simulation tools allow robots to be trained in virtual environments before they ever touch a real object. Hardware costs for sensors, actuators and compute are falling.

Capital has noticed. Humanoid and general purpose robotics companies such as Figure AI and Physical Intelligence raised large rounds in 2025 at multibillion dollar valuations. Nvidia has made "physical AI" a central theme of its strategy. China, which according to the International Federation of Robotics already has a higher density of industrial robots in manufacturing than Germany or Japan, is investing heavily in robotics as an industrial priority.

Beyond robots, AI is accelerating scientific discovery in materials, chemistry and biology, and is becoming a key tool in energy systems, logistics and advanced manufacturing.

**Different economics**

For investors used to software, physical AI has a very different profile. It requires hardware, manufacturing capability, supply chains, integration with existing equipment, safety certification and often significant upfront capital. Gross margins are lower. Sales cycles are longer. Failure modes are physical, and sometimes dangerous.

But those same characteristics create defensibility. A software interface can be replicated by a competitor in weeks. A robot fleet integrated into a customer's operations, with years of proprietary operating data and certified safety processes, is much harder to displace. The moats in physical AI are built from things that cannot be downloaded.

**Where I see the opportunity**

From an investment perspective, I see three areas of particular interest.

The first is applied automation in industries facing structural labour shortages: logistics, warehousing, agriculture, construction, healthcare support and manufacturing. Here, the business case does not depend on futuristic humanoids. It depends on reliable machines doing specific tasks at a lower total cost than the alternative.

The second is the enabling infrastructure: components, sensors, simulation software, edge computing and the energy systems required to power and charge automated operations.

The third, and in my view the most underrated, is industrial companies themselves. Many mid sized manufacturers and industrial service businesses in Europe, including in Central and Eastern Europe, have the domain knowledge, customer access and operational data to deploy physical AI successfully. What they often lack is capital and transformation capability. That is precisely what private equity can provide.

**A return to the question I asked in 2020**

In 2020, the question was whether AI was a sector or a technology that would reprice every sector. Six years later, I think the answer is clear. AI is a general purpose technology, in the same family as electricity and the internal combustion engine. Its first phase reshaped digital work. Its next phase will reshape physical work.

The investors who understand both, and who can bridge the world of models with the world of factories, fleets and power plants, will be best placed for the decade ahead.

*Investment question: Is the next major AI investment opportunity in the digital world, or in the automation of the physical one?*
