The end of pilot projects: robots are deployed in fleets
Just two years ago, most companies were still just testing robotics. One robot in one hall, evaluation after a year, cautious decision-making. Data from mid-2026 paints a different picture. According to the Association for Advancing Automation (A3), North American companies ordered a total of 36,766 robots worth $2.25 billion in 2025 — and this year is characterized by the transition from pilots to full deployment across multiple operations at once.
A summary overview by MarketScale identifies four technologies that define the factory of 2026: edge AI (processing data directly on the machine instead of in the cloud), machine vision, collaborative robots, and digital twins. The key point is that companies are no longer deploying them in isolation — they are assembling them into an integrated whole, where camera systems inspect thousands of parts per minute, robots work alongside people without safety cages, and digital copies of the production line predict failures before they happen.
ABB and Roche: physical AI enters laboratories
The most interesting announcement in recent weeks came from Zurich. ABB Robotics has entered into an agreement with pharmaceutical and diagnostics company Roche to deploy so-called physical AI in a laboratory environment. By physical AI, imagine artificial intelligence that controls not just software, but actual machines in the physical world — robots that see, grasp, and manipulate objects.
The first deployment will focus on handling pathology samples on slides and autonomous logistics within large diagnostic laboratories. Mobile manipulation robots will be combined with fixed robotic arms. This is a significant step for healthcare: laboratories worldwide, including Czech ones, are struggling with a shortage of lab technicians, and automating routine sample handling can speed up diagnostics for patients.
ABB has also completed its portfolio of autonomous mobile robots (AMRs) with the new autonomous forklift Flexley Stack F712. It navigates warehouses using Visual SLAM technology — camera-based navigation enhanced with artificial intelligence that doesn't need magnetic tapes or wall reflectors. All ABB AMR machines now run on the unified AMR Studio platform, which according to the manufacturer shortens commissioning by up to 20 percent.
Autonomous pallet and drilling robot for data centers
Alongside established players, capital is flowing into the sector through startups as well. Logistics company Logic introduced the Logic Pallet — according to its claims, the first autonomous mobile pallet designed for operation across multiple facilities. While the first generation of warehouse robots was optimized for a single building, the autonomous pallet is designed to transport goods within the entire distribution network. It solves one of the biggest limitations of existing warehouse automation.
In construction, a fleet-capable drilling robot for data center construction has caught attention, now commercially available. According to Robotics Tomorrow, it drills up to 10× faster than conventional methods and has shortened construction schedules by a total of 190 weeks across 26 major projects. At a time when demand for AI data centers exceeds construction capacity, this is a figure investors won't overlook.
Canadian company InDro Robotics has meanwhile started delivering the first units of the Axiom platform — a modular, affordable humanoid-type robot designed for research teams. It's a clever strategy: instead of competing with giants for factory deployments, it offers universities and companies an inexpensive tool to get hands-on experience with physical AI development before investing millions.
Software and batteries: the invisible layer maturing
Two less conspicuous announcements suggest that robotics is maturing beneath the surface as well. Company Peridio released Avocado OS 1.0, a production operating system for physical AI devices. Creating a software image for an entire fleet of robots — previously a months-long task — can now, according to the manufacturer, be done from a laptop using a single configuration file and three commands. Battery startup Addionics simultaneously launched the Autonomous Architecture platform, targeted at robots, drones, and autonomous vehicles in continuous 24/7 operation.
The message is clear: hardware is no longer the only bottleneck. Investments are flowing into software, firmware, and power — that is, the infrastructure without which individual robots will never become reliable fleets.
A shadow over the boom: global manufacturing slows down
This entire expansion is taking place in an uncertain macroeconomic environment. The analytics firm Interact Analysis lowered its global industrial production growth estimate for 2026 from 2.9 to 2.6 percent. The reasons are higher input costs following the oil shock associated with the Middle East conflict and ongoing US tariffs. The average annual growth for the 2025–2030 period was revised from 3.1 to 2.9 percent. America is expected to grow by only 1.9 percent, Asia by 2.9 percent.
What's interesting is how companies are responding to the uncertainty. According to analysts, they are no longer perceiving global shocks as exceptions and are taking them as the new normal — and this, paradoxically, works in automation's favor. The shift away from just-in-time production, the return of manufacturing closer to domestic markets (reshoring), and the shortage of workers are pushing companies toward robots that help keep local production competitive. The only unequivocally growing area remains semiconductors: the US chip sector is expected to grow by 10.7 percent this year, South Korea's by 10.3 percent — largely due to demand for AI chips.
What this means for the Czech Republic and Europe
The Czech Republic is one of the most industrialized countries in the EU and has long ranked among the states with the highest robot density in Central Europe. The trends described above directly affect Czech companies: ABB has a strong presence in the Czech Republic including development capacities, and its AMR platform is available for domestic warehouses and manufacturing as well. The reshoring of European production, mentioned by analysts, could bring orders to Central European subcontractors — but only to those that can compete on productivity, not on labor costs.
The message for Czech manufacturers is therefore twofold. Robots are getting cheaper and their deployment is accelerating — collaborative robots and pre-packaged automation bundles can now be deployed even by smaller engineering firms without their own engineering team. At the same time, the window of opportunity won't stay open forever: whoever postpones automation will be competing in five years with rivals already running fully integrated, software-driven operations.
What is physical AI and how does it differ from regular artificial intelligence?
Physical AI refers to artificial intelligence that controls machines in the real world — robots, carts, or manipulation arms. While language models like ChatGPT work with text, physical AI must process data from cameras and sensors in real time and use it to safely move physical objects. That's why it is often combined with edge AI, meaning computations done directly on the device without reliance on the cloud.
Will autonomous mobile robots in warehouses take people's jobs?
Experience from existing deployments points more to a shift in job roles rather than their elimination. AMR robots take over the physically demanding transport of pallets and goods, while people move to operations, quality control, and fleet management. Moreover, in Europe, logistics has long reported a shortage of workers, so robots more often fill vacant positions than replace existing employees.
Is robotic automation accessible for smaller Czech companies?
Yes, and accessibility is rapidly improving. Collaborative robots (cobots) don't require safety cages or extensive hall modifications, and their purchase price starts at roughly 600,000 CZK. The growing supply of pre-packaged automation bundles and "robot as a service" models (leasing instead of purchasing) further reduces the initial investment — so a smaller company can start by automating a single operation and gradually expand.
