Swiss start-ups compete to turn physical AI into a global robotics industry

Switzerland’s robotics sector now counts 182 small and medium-sized enterprises. Their machines are moving beyond controlled factory demonstrations into construction, logistics, inspection, cleaning and maintenance, bringing artificial intelligence into physical environments where conditions change by the minute.

The shift has acquired a name: physical AI. These systems combine sensors, mechanical hardware and AI models so robots can interpret surroundings, make decisions and adapt to unfamiliar tasks. That capability matters in production lines that handle varied goods, construction sites with changing terrain and dangerous facilities where people face avoidable risks.

Swiss companies are developing humanoid systems, autonomous drones, self-driving vehicles and machines designed for hazardous environments. The country’s research base gives them a strong starting point. Many ventures grew out of ETH Zurich and EPFL, whose laboratories have helped make Switzerland a recognised centre for robotics and AI.

The commercial race is global. US technology companies and Chinese manufacturers can draw on larger domestic markets, deeper pools of capital and high-volume production. Swiss founders therefore need to convert laboratory expertise into reliable products, recurring revenue and international distribution. The next phase will test whether Switzerland can build enduring robotics companies, rather than serving only as an acquisition pipeline for foreign groups.

Gravis Robotics has attracted $200 million from SoftBank for automated construction machinery. The Zurich-based company develops earth excavators and loaders that can operate with greater autonomy, targeting a sector where terrain, materials and working conditions change constantly.

The investment shows how Swiss robotics firms can attract global capital when they address a clear industrial problem. Construction equipment operates outdoors, often in dusty, uneven and hazardous settings. Automation could help contractors carry out repetitive excavation and loading work while allowing skilled workers to supervise machines, plan projects and handle exceptions.

Gravis is part of a broader Swiss effort to apply robotics to specific commercial environments. Other companies are working on aerial reconnaissance, oil rig inspection, office cleaning and data centre maintenance. Each application places different demands on navigation, safety, endurance and interaction with people.

The funding also brings a strategic complication. SoftBank is becoming a major force in Swiss robotics after agreeing to buy ABB’s robotics division for $5.4 billion in 2025. Foreign investment can accelerate development and open international markets, while ownership decisions can influence where intellectual property, manufacturing and strategic control ultimately sit. Swiss start-ups must use the capital available to scale without losing the ability to shape their own long-term direction.

Mimic Robotics wants robots to learn new production skills from human demonstrations. The Zurich start-up combines human-like robotic hands with foundation models, allowing its systems to perform repetitive manufacturing and logistics tasks without the re-coding required by many traditional machines.

That approach targets one of the persistent barriers to industrial automation. Conventional equipment can perform a narrowly defined operation with speed and precision, but changing the product or workflow often requires engineering time and new programming. A robot that can observe a worker, interpret the task and transfer the skill to its own movements could make automation more practical for smaller production runs and mixed inventories.

Mimic chief technology officer Elvis Nava says the company intends to remain independent. “We have a real opportunity to become a hyper-scaler frontier AI company from Europe,” he told SWI swissinfo.ch. “We are well positioned to play a much bigger role than just being absorbed by a bigger US company.”

The ambition places Mimic in a demanding contest. General-purpose robotics requires dependable hands, robust perception, safe movement and models that work outside the training environment. Customers will judge the system by uptime, accuracy and the cost of deployment, not by demonstrations alone. Mimic’s challenge is to prove that flexible learning can deliver measurable gains on real factory and warehouse floors.

Foreign buyers have already begun to reshape Switzerland’s robotics landscape. SoftBank’s $5.4 billion purchase of ABB’s robotics division in 2025 placed a major Swiss industrial asset under Japanese ownership. Earlier in 2026, Amazon acquired Rivr, a Swiss start-up developing autonomous four-legged robots for logistics.

Those deals offer founders access to capital, distribution networks and large industrial customers. They also underline the pressure on young companies to scale before better-funded competitors take the lead. A start-up can develop a strong prototype in Switzerland, yet struggle to finance certification, manufacturing, sales and aftercare across several continents.

The country’s ecosystem offers advantages. ETH Zurich and EPFL continue to generate research and spin-offs, while Switzerland’s manufacturing, pharmaceutical and logistics sectors provide demanding early customers. Its reputation for precision can help companies win trust in markets where a mechanical failure carries a high cost.

Robert MacKenzie, a venture partner at Zurich-based Ellipsis Venture, argues that Swiss founders should concentrate on targeted supply-chain problems rather than trying to match rivals that mass-produce complete robotic platforms. That strategy points toward specialised machines, software and components that companies can sell globally. The ownership question will remain active as investors weigh the value of Swiss independence against the speed and reach that international partnerships can provide.

Switzerland’s route to a global robotics industry runs through specialised machines that solve expensive, practical problems. The country has the research talent and industrial expertise to compete, yet its companies face rivals from the United States and China with greater scale and access to manufacturing capacity.

The most promising Swiss ventures are targeting environments where precision and adaptability matter: construction sites, production lines, logistics networks, oil installations and data centres. These markets reward systems that can work safely around people, cope with changing conditions and integrate with existing equipment. They also offer clearer commercial pathways than a race to build a universal humanoid for every task.

For Swiss workers and companies, the effects will depend on how quickly deployment moves from trials to routine operations. Robots could take on repetitive or dangerous duties while increasing demand for technicians, supervisors, software specialists and maintenance teams. Companies will need to demonstrate that the productivity gains justify new equipment, training and operational changes.

The next milestones are commercial rather than purely scientific: repeat orders, dependable performance and production at a sustainable cost. Switzerland’s 182 robotics SMEs give the country a broad base. Whether that base becomes a durable European industry will depend on founders’ ability to retain strategic control, attract patient capital and sell solutions beyond the Swiss market.