How machines that see, decide, and act are reshaping the factory floor – and why Europe’s reindustrialization moment may hinge on getting this right.

Industry & Technology Part 1 of 2 – March 2026

---

At CES in Las Vegas in January 2026, NVIDIA CEO Jensen Huang made a declaration that resonated far beyond the convention hall:

‘The ChatGPT moment for physical AI is here.'[1]

ChatGPT didn’t just launch a product in 2022. It triggered a capital reallocation, a talent scramble, and a strategic rethink across every industry that touches information.

Europe missed this revolution entirely. No leading large language model lab. No hyperscale cloud. No dominant semiconductor designer. When digital AI crested, Europe was standing on the beach, watching.

I say this not to be harsh, but because it matters for what comes next.

The same tectonic shift is now happening on the factory floor.  Autonomous systems that can perceive, reason, and act in an unstructured world are moving out of research labs and into live production. This time European companies in automotive, precision machinery, pharma, logistics are best positioned. If European industry reads this right, it may not just catch up – it may lead.

Europe has been training for this race for a century. It just didn’t know the race was coming.

For procurement officers, plant managers, and industrial investors in Europe, the questions are urgent: What exactly is physical AI? Where does it stand today? And when will it hit the budgets, the supply chains, and the negotiating tables where industrial decisions are made?

---

What Is Physical AI?

NVIDIA defines the category simply: “Physical AI lets autonomous systems perceive, understand, reason, and perform or orchestrate complex actions in the physical world” [2]. Physical AI shifts automation from rigid execution to adaptive autonomy. The technology opens the long tail of manufacturing tasks that have resisted automation for decades: complex, high variant assembly, dexterous handling of irregular objects, quality inspection. Humanoid robots are a small but prominent subset.

Physical AI is a breakthrough from convergence of several maturing technologies: computer vision, sensor fusion, reinforcement learning, vision-language-action models – VLAs (the physical AI counterparts of LLMs), and sophisticated edge computing. Fundamentally, physical AI is a software-driven paradigm running on a variety of hardware — shifting cost dynamics from linear toward Moore’s Law-like scaling.

---

We Are at the Inflection Point. Right Now.

From Prototype to Pilot

The global robotics market reached nearly USD 50 billion in 2025. The annual industrial robot installations is projected to surpass 700,000 units by 2028. [3] Investment in the AI layer powering these systems has surged.

In October 2025, SoftBank agreed to acquire ABB’s robotics division for approximately USD 5.4 billion. [4] Rivian collected USD 500 million in a series A round spinning out Mind Robotics. [5] Industrial robot fleets are being reframed as assets deploying AI software and harvesting real-world data.

The Deployment Evidence

• Amazon has deployed its one-millionth robot. The AI layer operates at the fleet level – optimizing task allocation and routing across thousands of robots .[6]
• CATL has deployed humanoid robots in battery manufacturing lines that match human speed and precision even with flexible cables.[7]
• In Europe, BMW deploys humanoid robots in Leipzig plant for high-voltage battery assembly and exterior component handling. The AEON unit is developed by the Swiss Hexagon Robotics. [8]
• Siemens and NVIDIA announced a major expansion of their partnership, with the goal of building the world’s first fully AI-driven, adaptive manufacturing site – to start in 2026 at Siemens’ electronics factory in Erlangen. [9]

These examples show how early movers are advancing now and accumulating real-world data and experience.

---

Europe’s Strategic Moment

A Second Chance – in the Right Domain

Europe missed the first wave of digital AI. But growing chorus of European industrial strategists, the World Economic Forum, and the McKinsey Global Institute argue that physical AI is different – and that Europe is structurally better positioned for it. [10, 11]

Physical AI is built on operational data from machines, structured manufacturing environments, and deep engineering expertise. Europe has all three in abundance. Its industrial sectors – automotive, precision machinery, logistics, pharma – ‘this is precisely where Europe’s underlying advantages are hiding in plain sight,‘ writes Fortune, pointing to Europe’s ‘universities, industry, startups, and the public sector’ as favoring the long-horizon, capital-intensive innovation that physical AI demands. [12]

The Policy Infrastructure Is Being Built

The EU is not waiting. Under the AI Continent Action Plan, the European Commission has prioritized the establishment of AI Factories – compute-rich hubs linking supercomputing centers, universities, SMEs, and industry – specifically for industrial AI development [13].

Germany’s €500 billion fiscal stimulus package, announced in early 2026 and allocated to infrastructure, adds a further demand-side catalyst [14]. Europe’s Stoxx 600 Industrials index is projected to deliver 13% earnings-per-share growth in 2026, compared to 6.6% in 2025 – with the strongest tailwinds for companies exposed to AI infrastructure, electrification, and advanced manufacturing [15].

---

What This Means for Industrial Decision-Makers in 2026

I will be direct about what I think this wave means for the people reading this.

Physical AI is not arriving in 2030. It is arriving now – for early movers. The technology is ready for pilot environments. Full-factory deployment without significant integration work is still 3–5 years out for most organizations. But the window for strategic positioning is open right now, and it will not stay open for long.

The companies that start evaluating in 2026 are looking at material production impact in the 2028–2031 window. The companies that wait for “proof of concept from peers” will not catch the early movers’ accumulated operational learning anymore.

The procurement and integration questions – the specifics of how this technology hits capital budgets, supplier negotiations, and factory operating models – are the subject of the next article in this series. I will address those in detail, because the strategic errors being made in early physical AI procurement are avoidable.

For now, the framing question is simpler:

Is physical AI on your strategic agenda for 2026? If not, I would ask why.

---

Key Numbers at a Glance