We're used to talking about horsepower, torque, and battery range in the automotive industry. Right now, the vocabulary is transforming and evolving—we talk about digital twins, agentic AI, physical AI, and "eyes-off" driving. And above all, about figures that have nothing to do with mechanics anymore.
Recent news stories all tell the same tale—that of a sector whose value is migrating from metal to software. Taken together, they sketch out a fundamental shift. And when we bring them back to Swiss reality, they become far less abstract than they might appear.
The factory becomes a system piloted by AI
Stellantis—the world's fourth-largest automaker, born from the PSA-FCA merger, with more than a hundred production sites worldwide—has partnered with Accenture and Nvidia to transform its factories into systems piloted by artificial intelligence.
The principle: create digital twins, that is, virtual replicas of production lines capable of simulating real-world operations in real time. Slowdowns and defects are detected before they affect a single physical car. Nvidia brings computing power and its Omniverse libraries; Accenture serves as the architect industrializing deployment from one factory to another. The first pilots are scheduled for 2026.
What's at stake here goes beyond buying a technology. It's a change in nature: the factory is no longer just a place where parts are manufactured and assembled, but a software system piloted by data. Value no longer resides solely in the vehicle produced, but in the capacity to orchestrate production itself.
The car becomes a driver
In February 2026, British startup Wayve closed a record funding round: 1.2 billion dollars for a valuation of 8.6 billion. The syndicate included Microsoft, Nvidia, Mercedes-Benz, Nissan, Stellantis, Uber, led by Eclipse, Balderton, and SoftBank.
Wayve's proposition is clear and radical: the company wants neither to build cars nor operate fleets. It sells an embedded AI layer—its "AI Driver"—to automakers and mobility platforms. Two levels: an "eyes-on" system to enhance driving assistance (ADAS), and an "eyes-off" system targeting level 4 autonomy in certain environments. Nissan plans a launch in Japan in 2027; Mercedes and Stellantis are working with Wayve on consumer applications and robotaxis.
In other words: driving capability becomes a software that you buy from a third party, just as you would buy a DMS (Dealer Management System) or an operating system. An automaker that doesn't control this layer risks becoming, ultimately, merely a chassis manufacturer on which another's intelligence runs.
The entire ecosystem is getting financed
The last signal, the most discreet but perhaps the most telling: BMW, through its venture capital subsidiary BMW i Ventures, launched a third 300-million-dollar fund in late April 2026, bringing its assets under management to approximately 1.1 billion.
The target: European and North American startups, from seed to Series B, specializing in agentic AI (software that executes complete end-to-end tasks), physical AI (which enables robots and autonomous machines to perceive, plan, and act in the real world), industrial software, advanced materials, and supply chain.
This is no longer marginal experimentation. BMW is already testing a humanoid robot at its Leipzig factory for battery assembly, and ran a pilot project with startup Figure AI at its Spartanburg site in the U.S. An automaker investing in humanoid robotics and agentic AI is no longer thinking solely about selling cars: it's repositioning itself across the entire industrial value chain.
The common thread: value leaves the vehicle
These developments reveal the same dynamic. The factory, driving, the ecosystem: at each level, value is shifting from the physical product to the intelligence layer that pilots it.
The same names appear as connecting threads. Nvidia is everywhere—supplier to Stellantis, investor in Wayve, reference technology for physical AI. It's the shovel and pickaxe seller in this gold rush, while Stellantis transforms its factories and invests in the brains of its future cars. The message is clear: a major automaker can no longer afford to be absent from one or the other of these layers.
For those following the automotive industry, this is a shift comparable to when the telephone became a smartphone: hardware remains necessary, but that's no longer where the bulk of value is captured.
And Switzerland, in all this?
This is where the subject becomes concrete—and surprisingly little discussed here.
Switzerland doesn't produce cars in series. One might therefore think this revolution is happening elsewhere, between Munich, Detroit, and London... that would be a misreading, for at least three reasons.
1. Swiss deep tech is already inside this shift. The best example is from Fribourg: Bcomp, founded in 2011, develops high-performance composites based on linen fibers (ampliTex, powerRibs). The company is in the portfolio of… BMW i Ventures, which participated in its funding rounds in 2022 and 2024. And since 2025, these Swiss materials are in series production at BMW, on exterior and interior elements (the roof of the next M3, for example), replacing carbon fiber. Add to this the robotics laboratories at EPFL and ETH Zurich, among the world's best on precisely what automakers are seeking today: physical AI. Switzerland isn't a spectator to this wave; it's a leading supplier, on the materials and robotics layer.
2. Switzerland is ahead on regulation. Since March 1, 2025, the automated driving ordinance is in force. It authorizes three uses: automated driving on highways (the driver can let go of the wheel while remaining ready to take control), operation of driverless vehicles on segments approved by cantons, and automated parking in designated lots. The first Swiss autonomous delivery vehicle, Loxo, is already operating. Concretely, this means that technologies like Wayve's "AI Driver" have here a legal landing strip—something not yet the case everywhere in Europe.
3. Switzerland is a market that will absorb these cars very quickly. With no factories but a dense and structured distribution network, Switzerland is a market where "software-defined" vehicles will arrive without delay. Cars sold here tomorrow will be rolling data platforms, updated remotely, with an increasing share of their value residing in software layers designed abroad.
The blind spot: what about downstream?
What strikes you when you add up these billions is that they irrigate almost entirely the upstream of the supply chain—manufacturing, driving, components. The part that nonetheless touches every Swiss household, that of buying, selling, owning, and reselling a car, remains largely outside these spotlights and capital.
It's paradoxical, because the same logic applies word for word. An automotive marketplace is exactly that: value captured on the software layer and data, placed over physical transactions. As long as this layer cost a fortune to build and operate, it remained the preserve of a few dominant players. Today, technology changes that equation: it drastically lowers the cost of designing and operating a quality platform, mechanically making the "toll" contestable.
And then there's a question that this shift makes increasingly pressing: as the car becomes a data platform, who controls Swiss automotive data? At the moment when upstream is being built around a few American tech giants, keeping in Switzerland an independent infrastructure, hosted locally, for the part that directly concerns automobilists here, isn't a romantic detail. It's a question of sovereignty.
The human counterpoint
One last point remains, and it's not the least. The more the industry automates, the more the factory pilots itself, the more the car drives without us—the more the human and passionate relationship with automobiles becomes, paradoxically, precious. Technology can optimize everything; it doesn't replace the desire to pass on a car you've loved to someone who will love it in turn.
The AI wave is real; it's already here. Switzerland's role won't be to build these cars. It can be to bring its deep tech to them, to regulate intelligently, and—for those who buy, sell, and love cars—to keep an independent actor, Swiss and human-scaled to do it.
