Europe’s Emerging Physical AI and Robotics Stack

Table of contents

• Robot Intelligence & Control
• Construction Robotics
• Factory & Manufacturing Robotics
• Warehouse & Logistics Robotics

Robotics is moving beyond machines that simply repeat the same movement in a controlled environment. Across Europe, a new generation of companies is developing systems that can understand what is happening around them, adjust when conditions change and operate in workplaces that were previously too unpredictable to automate. This article looks at how this shift is taking shape across the emerging European robotics market.

<a id="robot-intelligence"></a>

Category: Robot Intelligence & Control

What is this category about?

• This category is the intelligence layer that determines how robots perceive their environment, learn tasks and decide how to move. Traditional industrial robots follow sequences of instructions created for a specific machine and production environment. The companies in this category are developing software that allows robots to remain useful when objects, tasks or physical conditions change.

• The main transition is from programmed behaviour toward learned behaviour. Rather than describing every movement in advance, these systems can learn a skill from a human demonstration or through simulation. Some also improve through reinforcement learning or data collected during real deployments. Once a skill has been learned, the robot can adapt it to situations that were not included in the original programming.

• This intelligence layer also makes existing robots easier to operate. Industrial robot programming usually requires specialist engineers, which makes automation expensive whenever production changes frequently. New platforms allow an operator to describe the desired outcome in natural language or demonstrate the task physically. The system then translates this input into reusable robot behaviour.

• An important technical challenge is connecting high-level reasoning with physical control. A robot may understand that it needs to retrieve a parcel or insert a component, but it must still convert this objective into precise and safe movements. The products in this category connect perception and planning with the control of motion and force.

• The boundary of the category does not depend on whether a company sells hardware. Sereact and Trener Robotics primarily position themselves as software platforms, while Mimic combines its models with a dexterous robotic hand. HIVE installs sensors and computing hardware onto existing industrial machines. In each case, the main source of differentiation is the intelligence controlling the machine.

• The economic objective is to expand the number of tasks that can be automated. Traditional robots work well when products and processes remain highly predictable. Robot intelligence becomes more valuable when objects vary or production changes regularly. It is also useful in physical environments where unexpected events make rigid automation unreliable.

What do products in this category do?

• Most of these products sit between the robot hardware and the operational workflow. For example, a customer can keep using an existing industrial robot from a manufacturer such as ABB or FANUC while adding software that gives it perception and task planning capabilities. This intelligence layer can also introduce skill learning and adaptive control without requiring the entire robotic system to be replaced.

• How does it work in practice? The software is usually installed on an external industrial computer connected to the robot’s existing controller. Depending on the application, additional cameras, force sensors or other hardware may be added so the software can perceive its environment. It then communicates with the robot through the manufacturer’s existing interfaces and sends task or movement instructions, while the robot’s original controller continues to manage low level motion.

• Some companies are building what can be described as a general purpose robot brain. Sereact’s Cortex combines vision with reasoning and control so that industrial robots can manipulate objects they have not previously encountered. The company initially focused on autonomous picking, but its product has expanded into returns handling and kitting. It can now support longer warehouse processes that involve inspection and operational decision making. Data from production deployments is used to improve the system over time.

• Trener Robotics is developing a similar software defined approach for industrial robot cells. Its Acteris platform allows an operator to describe a task conversationally. The system uses perception and simulation to translate this request into executable behaviour. The platform supports several established robot brands and is initially focused on applications such as machine tending and high-mix production.

• Flexion Robotics focuses on giving robots broader, more general autonomy. Its Reflect software lets users give instructions in natural language, then combines visual understanding, action planning and full body movement control to carry them out.

• Dexterous manipulation is another important product direction. Mimic Robotics combines anthropomorphic robotic hands with models trained from human demonstrations. This allows robots to reproduce complex manual techniques and adjust their behaviour when the object moves. 

• Other companies operate deeper inside the control stack. Luffy AI develops adaptive controllers that can run directly on motors and robots. The same technology can also be applied to drones or other industrial equipment. Rather than only deciding what a machine should do, the software continuously adjusts how it behaves as physical conditions change.

• HIVE applies this intelligence layer to existing heavy machinery. Its system retrofits standard industrial vehicles with sensing and supervised autonomy. This allows an operator to control several machines from a safer location while preserving the existing equipment fleet.

• MicroAGI combines robot deployment with the data infrastructure required to improve models. Atlas is positioned as its industrial deployment product, while Shift captures how humans perform physical work. These demonstrations can then be transformed into training data. The company illustrates how closely deployment and data collection are becoming connected in embodied AI.

• Deployment is a substantial part of all these products. The software must be integrated with real machines and safety systems before it can operate inside a customer environment. It also needs to work with the company’s existing operational software. The number of forward deployed and field engineering hires across the category shows that robot intelligence is not a plug-and-play type of product.

What are the major trends shaping this category?

• Europe enters this transition with a large industrial base and a substantial installed robot fleet. Western Europe has one of the highest levels of robot density in the world. However, European companies represent a much smaller share of new global robot installations than Asian manufacturers. This creates pressure to increase the capabilities of machines that are already deployed rather than relying only on the sale of new hardware.

• The technology is moving from models designed for a single task toward broader foundation models for robotics. “Vision language action” systems connect visual understanding with language instructions and physical execution. This makes it possible for the same model to reason across several tasks rather than requiring a completely separate program for each one.

• Europe may be better positioned to apply these models than to build the largest general purpose foundation model. The region has strong robotics research and deep expertise in industrial engineering. It also offers access to manufacturing environments where useful physical data can be collected. Companies such as Sereact and HIVE are therefore building their learning loops around real industrial deployments rather than relying entirely on laboratory demonstrations.

• Training data is becoming one of the most important competitive assets in the category. Physical data is expensive to produce because it requires real people and machines. Mimic records skilled human movements, while MicroAGI captures first person demonstrations of physical work. Sereact follows a different model by learning from production activity across its deployed robot fleet.

• Simulation remains necessary because collecting every possible situation in the real world would be too expensive. A simulated environment can expose a robot to changes in objects or lighting before the system reaches the customer site. It can also generate unusual situations that would be difficult or unsafe to reproduce physically. Real operational data is still needed, however, because simulation cannot fully represent the complexity of the physical world.

• Hardware agnostic intelligence is becoming an important European positioning. Many factories already contain machines from several vendors and generations. Replacing this installed equipment would take considerable time and capital. Trener therefore supports several established robot brands, while HIVE retrofits existing heavy machinery. Sereact also presents its intelligence layer as something that can eventually operate across several types of robotic systems.

• System integrators remain central to distribution. European manufacturers commonly purchase a complete automation cell rather than selecting each software component independently. Partnerships with companies such as Kardex or AWL therefore give startups access to customers and implementation resources. They also provide the procurement credibility required to sell into conservative industrial environments.

• Smaller manufacturers will be especially important for the category. Europe has many industrial SMEs operating high-mix production environments where traditional automation is often too rigid. Robot intelligence can improve the economics by reducing the time needed to configure a new task. It can therefore bring automation into factories that could not previously justify a large custom integration project.

• Labour shortages also strengthen the demand for these products. The main value is not necessarily the complete replacement of workers. In many use cases, the robot removes a difficult or hazardous part of the job while a human remains responsible for supervision and exception management.

What types of customers are buying these products?

• Warehouse and logistics operators are among the most advanced customers. Sereact has deployed its technology with companies such as Active Ants and Austrian Post. Its robots are used in environments where the assortment changes continuously, making traditional object specific programming difficult. DeltiLog expanded its deployment after the initial systems proved successful, while Sereact’s partnership with Kardex is intended to bring the technology into a much larger number of warehouse installations.

• Manufacturers and machine shops are using robot intelligence to automate variable production. Sereact lists several major automotive manufacturers among its customers, while Mimic is running pilots with industrial companies that need to automate complex manual work. Trener reaches machining customers through automation partners and existing robot ecosystems. These buyers are generally interested in processes that change too frequently for traditional programming to remain economical.

• Heavy industry and infrastructure operators are testing supervised autonomy. HIVE is working with Yara to retrofit a Volvo wheel loader at a fertilizer production facility. The machine can be supervised from a remote control room rather than operated continuously from inside the vehicle. HIVE has also tested autonomous machinery with Veidekke and Telia in difficult industrial environments.

• Robot manufacturers and automation integrators are becoming customers as well as distribution partners. Flexion works with robot makers that need an autonomy layer but do not want to build the entire intelligence stack internally. Trener sells through integrators, while Sereact works with companies such as Kardex and Hörmann. At this stage of the market, these partners often help shape the product while also introducing it to end customers.

How does the funding environment look?

• The category has attracted unusually large financing rounds (for Europe) even though many of the products remain relatively early. Capital is flowing toward companies with substantial live deployments, but investors are also backing teams that are still building more general autonomy and data platforms.

• Sereact is currently the clearest commercial funding leader. It raised a $110 million Series B in April 2026 after securing a €25 million Series A the previous year. The company was able to support the funding story with evidence from more than 200 deployed systems and over one billion production picks. 

• Investors are also making large bets much earlier in company development. MicroAGI raised a $55 million seed round in July 2026. The financing is intended to expand its deployment platform while increasing its computing capacity and field operations. Flexion raised a $50 million Series A shortly after its seed round, with the capital supporting model development and access to larger robot fleets.

• The use of capital reflects how operationally demanding the category is. These companies need model researchers, but they also require robotics engineers and physical machines. Their expansion depends on simulation infrastructure and proprietary training data. Field technicians and systems integrators are equally important because the product must work inside a real industrial environment.

• Public support remains important in Europe, particularly during the technical development phase. Luffy has received several UK research grants, while Mimic received Venture Kick support before raising larger institutional rounds. Strategic investors from the industrial and computing ecosystems are also participating because robot intelligence depends closely on simulation and specialized hardware.

• The category has not yet entered a visible consolidation phase. The companies are still trying to prove which technical approach can produce reliable commercial results. The next stage will depend on whether these platforms can move beyond pilots and operate across several customer sites without requiring a highly customized engineering project each time.

<a id="construction-robotics"></a>

Category: Construction Robotics

What is this category about?

• This category covers robots that perform or directly support physical construction work. Unlike the “Robot Intelligence & Control” category, the companies are not primarily selling a general intelligence layer that could be used across many industries. They package robotics around a specific construction process such as excavation, bricklaying, concrete renovation or material transport.

• Construction creates a particularly difficult environment for automation. A factory can be designed around a robot, while a construction site changes every day. The ground is uneven, the weather varies and other workers continuously modify the physical environment. Robots therefore need to tolerate imperfect plans and frequent differences between the digital model and the real site.

• The category also extends beyond robots physically moving around a building site. Hyperion Robotics uses robotized microfactories to manufacture foundations and infrastructure components off-site, while Sodex turns construction machines into automated surveying systems. Both automate part of the construction process even though they do not resemble a conventional construction robot.

• The shared objective is to turn a difficult manual process into something more predictable. Customers are generally buying faster execution or reduced labour exposure rather than robotics itself.

What do products in this category do?

• Gravis Robotics installs a retrofit kit on an existing excavator and connects it to the machine’s controls. The system uses onboard sensors and the digital site plan to determine where and how deep to dig, then controls the excavator’s movements automatically. An operator can supervise the work remotely or use the system as an assistance tool while remaining in the cab.

• Monumental has developed a coordinated fleet of smaller robots for bricklaying. Its Atrium software interprets the building design and manages execution on the physical site. Rather than selling the equipment to contractors, Monumental operates as an autonomous masonry subcontractor. The customer outsources the work in much the same way it would hire a traditional bricklaying company.

• Other companies begin with narrower site tasks. KEWAZO’s LIFTBOT transports materials vertically through scaffolding and other constrained structures. Sitegeist is developing modular robots for concrete renovation, where processes such as hydrodemolition expose workers to water pressure, dust and difficult working positions. These tasks offer a clearer starting point than trying to automate an entire construction sequence.

• Sodex follows another approach by mounting scanning equipment onto working machines. The terrain is captured as the operator performs normal earthmoving work, which reduces the need for separate surveying visits and keeps project documentation current.

• The product is rarely limited to the robot itself. Construction customers also need planning tools, documentation and support during deployment. This explains why Monumental built Atrium, why KEWAZO added its ONSITE analytics platform and why Sodex developed SDX-Cloud around its capture hardware. The software connects the physical work with the project record that contractors and asset owners must maintain.

What are the major trends shaping this category?

• The market is developing around narrow tasks rather than general purpose construction robots. Each successful product begins with a process where the work is repetitive enough to automate but difficult enough to justify the equipment. Excavation and bricklaying fit this model, as do concrete removal and vertical material transport. This specialization reduces technical risk and gives customers a specific manual workflow against which they can measure the result.

• The commercial model is becoming as important as the machine. Construction is dominated by smaller companies, and the European Commission reports that micro firms account for most businesses in the sector. Many customers cannot purchase expensive robots or create an internal robotics team. Monumental therefore sells completed masonry work, while Gravis is entering the market through rental companies and equipment dealers. These models allow contractors to use automation without making a large permanent investment.

• Retrofitting existing equipment is likely to remain important in Europe. Contractors already own excavators and other machines that can operate for many years. Gravis adds autonomy to this installed fleet, while Sodex adds automated data capture. This can be easier to adopt than introducing a completely unfamiliar machine that must be transported and supported separately.

• Off-site automation is gaining momentum because a microfactory offers more control than a changing construction site. Hyperion can manufacture standardized components in batches before delivering them to the project. This model fits the broader European push toward prefabrication and industrialized construction while still allowing components to be adapted digitally for each location.

• Infrastructure renovation may become a stronger opportunity than new building construction. Europe has an ageing stock of bridges, utilities and industrial facilities that must remain operational during repair. Sitegeist is targeting renovation work directly, while KEWAZO increasingly sells into industrial shutdowns where delays are extremely expensive. Hyperion is finding repeatable applications around electricity networks and EV charging infrastructure.

• Labour pressure remains an important driver. Between 25% and 30% of European construction firms report that worker shortages restrict their output. However, the strongest products do not simply promise to replace an entire trade. They remove the most repetitive or physically demanding part of the process while leaving workers responsible for preparation and exception handling.

What types of customers are buying these products?

• Large contractors and infrastructure projects. Gravis deployed autonomous excavation with Taylor Woodrow at Manchester Airport, while Monumental works as a masonry subcontractor for builders across residential and commercial projects. These customers need the robot to fit an existing schedule rather than forcing the entire project to be reorganized around it.

• Industrial asset owners and maintenance contractors. KEWAZO has operated at BASF sites and worked with contractors including Altrad and Bilfinger. The product is particularly relevant during maintenance shutdowns, when material must move through scaffolding quickly and delays can affect the wider industrial operation.

• Utilities and energy infrastructure operators. Hyperion has delivered or tested robotically manufactured components with National Grid, Costain and Fastned. These customers often need the same type of foundation across many locations, which creates a stronger opportunity for standardized off-site production.

• Equipment manufacturers, rental companies and dealers. Gravis works with heavy equipment manufacturers and the rental company Flannery, while Sodex has used Hitachi and SITECH channels to demonstrate and distribute its surveying systems. These partners give startups access to contractors that already buy machines and services through established equipment networks.

How does the funding environment look?

• Funding is concentrated around companies that have moved from prototypes into repeatable deployment. Monumental raised a $32 million Series B in July 2026 after its earlier $25 million round. The company now operates more than 100 robots and is using the new financing to expand its fleet and enter additional markets.

• Gravis raised a $23 million Series A in November 2025. The round followed active site deployments and partnerships with contractors, equipment manufacturers and rental channels. This suggests investors are rewarding the ability to distribute the technology through the existing construction-equipment ecosystem rather than relying only on direct robot sales.

• The rest of the category remains earlier. Sitegeist raised a €4 million pre-seed round to develop its renovation robots, while Sodex raised a €4 million seed round around its machine-integrated surveying platform. Hyperion secured €6.4 million in growth funding in July 2026 as it moved from project delivery toward a network of robotic microfactories.

<a id="factory-robotics"></a>

Category: Factory & Manufacturing Robotics

What is this category about?

• This category covers companies that sell complete robotic systems for work performed inside factories. The products combine the robot, the intelligence layer and the deployment workflow around a specific manufacturing process. Customers are not expected to assemble the system themselves from separate hardware and software components.

• The main opportunity lies outside the highly standardized automotive production lines where industrial robots are already common. Many European manufacturers operate high mix production, where products change frequently and volumes are too low to justify a long custom automation project. The companies in this category are trying to make robotics viable in these less predictable environments.

• The category includes both general manufacturing platforms and highly specialized systems. RobCo and Sunrise Robotics are building configurable robot cells that can be adapted to several production tasks. R3 Robotics focuses on the automated disassembly of electric vehicle systems, while Flink Robotics applies flexible material handling to the movement of components and finished products around production lines.

• The economic objective is to reduce the cost and time required to automate a new process. A traditional robotic cell may take months to design and integrate. These companies aim to standardize more of the hardware and software so that the same system can be reused across several customers or reconfigured when production changes.

What do products in this category do?

• Most products in this category are built around robotic arms. Because factory work usually involves picking up a part and then moving, assembling, processing or dismantling it. The main difference is how these arms are packaged. Some companies build fixed workstations for one production process, while others use two arms for more complex manual work. Specialized systems combine several arms and tools into a complete production line, while mobile robotic arms can move between different work areas.

• RobCo installs complete robotic workstations around existing factory processes. For machine tending, for example, the robot picks a metal part from a tray, places it inside a CNC machine, waits for machining to finish and removes the completed part. RobCo configures the robot, gripper and safety equipment, while its software lets the customer adjust the workflow and monitor several systems. The robots can be purchased or leased as a service.

• Sunrise Robotics builds dual arm robotic cells for manual assembly work. A manufacturer places the cell next to an existing production line, where its cameras locate the components and its two arms perform tasks such as fastening screws, cutting material or joining parts. Sunrise first creates a digital copy of the workstation and trains the robot in simulation, reducing the amount of programming required on the factory floor.

• R3 Robotics builds robotic lines that dismantle electric vehicle batteries. Cameras identify the battery model and locate screws, covers and internal components. Robotic arms then open the pack and remove the parts in a defined sequence so that they can be reused or recycled safely. When a new battery model arrives, R3 scans it and creates a new disassembly workflow. Customers can install the line at their own facility or send batteries to an R3-operated site.

• Flink Robotics builds mobile robotic arms for moving mixed goods around factories and warehouses. The robot is positioned beside a conveyor, container or pile of components. Its cameras identify each item, after which the arm picks it up and places it into the next production, packaging or logistics step. Flinkbot Mini handles smaller items, while Flinkbot Duo uses two coordinated robots for heavier loads or faster processing.

What are the major trends shaping this category?

• The central market shift is from automation designed for stable mass production toward automation that can support product variation. European manufacturing contains many SMEs and specialist suppliers whose production volumes do not justify a unique robotic cell for every process. Modular hardware, simulation and easier programming reduce the amount of engineering that must be repeated for each deployment.

• This matters because Europe already has a large installed base of factory robots, but adoption remains uneven. Western Europe reached 267 industrial robots per 10,000 manufacturing employees in 2024. European installations nevertheless declined by 8% that year, even though the total remained the second highest on record. The next growth opportunity therefore depends partly on bringing robotics beyond the large automotive plants that drove earlier adoption.

• Brownfield compatibility is becoming more important than building a perfectly automated factory from scratch. RobCo emphasizes compact systems that fit existing production lines, while Flink positions its robots as equipment that can be placed directly into a current workflow. Sunrise is designing cells for workspaces originally created for people rather than robots. This makes installation speed and physical footprint central parts of the product.

• The integration model is also becoming more standardized. Instead of developing every deployment as a separate engineering project, vendors are creating reusable robot configurations and cloud-based workflow tools. RobCo’s partnership with Koenig & Bauer goes further by embedding its robotics platform into industrial machines themselves. This could turn robotics from an optional factory project into a capability distributed through equipment manufacturers.

• Commercial models are adapting to the economics of mid-sized manufacturers. Leasing and Robotics-as-a-Service reduce the initial capital requirement and allow a factory to compare the monthly cost of the robot with the cost of the existing manual process. RobCo customers have highlighted this model as an important reason they could introduce automation without waiting for a large capital expenditure approval.

• Another trend is the expansion of robotics into disassembly. Traditional automation has concentrated on assembling standardized new products, while end-of-life products arrive in different conditions and configurations. AI vision and adaptive tooling are making it possible to automate parts of this reverse workflow. Europe’s battery regulation reinforces the opportunity by introducing stronger traceability requirements and a digital battery passport from February 2027. The passport must include information relevant to safe dismantling and disassembly.

• Labour constraints remain part of the demand, but the factory specific problem is often the availability of operators for repetitive shift work. European SMEs continue to report difficulty finding workers with the required skills. The products in this category generally automate machine loading or handling tasks so that existing employees can supervise more production equipment rather than remain attached to one repetitive process.

What types of customers are buying these products?

• Industrial SMEs introducing their first significant automation. RobCo has deployed systems with manufacturers such as KABZA, where a robot loads metal parts for further processing, and Dr. Willi Knoll, where compact palletizing robots were integrated into existing food production lines. These customers value a right-sized system that their own employees can operate without an internal robotics department.

• Larger manufacturers automating complex or variable processes. RobCo automated vehicle scanning inside a BMW crash testing facility and works with Rosenberger on flexible CNC production. Sunrise is targeting manufacturers in areas such as specialist automotive production, advanced batteries and consumer electronics. Its early commercial activity has consisted largely of pilot discussions and initial deployments rather than publicly named large scale rollouts.

• Recyclers, automotive OEMs and remanufacturing operators. R3 Robotics targets organizations processing mixed electric vehicle batteries and other high-voltage systems. These customers need higher throughput, but they also require safe handling and a traceable record of each component. The company’s early proof is based on onboarding battery packs from major OEM platforms rather than disclosing a long list of named customers.

How does the funding environment look?

• Funding is heavily concentrated in RobCo. The company raised a $100 million Series C in January 2026 after a $42.5 million Series B and a $14 million Series A. The latest round is being used to expand enterprise deployments and strengthen its presence in the United States. This places RobCo in a different financing category from the other companies on the landscape.

• R3 Robotics raised a €14 million Series A alongside €6 million in European grants in February 2026. The financing supports its expansion from batteries into complete electric vehicle systems. The funding profile illustrates how circular-economy robotics can combine venture capital with European public funding tied to industrial resilience and resource recovery.

• The broader funding pattern rewards companies that can reduce deployment risk. Investors are not only financing more capable robots. They are backing standardized systems that can be installed without restarting the integration process for every customer. RobCo has already demonstrated this at commercial scale, while Sunrise and Flink are still proving that their general-purpose cells can become repeatable products.

• Vertical specialization offers another route to defensibility. R3 does not need to automate every factory process if it becomes the standard system for dismantling electric vehicle components. In this part of the market, detailed process knowledge and access to many product variants may create a stronger advantage than a broadly capable robot.

• The category contains one interesting acquisition. RobCo acquired the assets and local team of Rapid Robotics when entering the US market in 2025. This suggests that well-funded European robotics companies may use acquisitions to obtain installed systems, customer relationships and field-deployment capacity in new markets.

<a id="warehouse-robotics"></a>

Category: Warehouse & Logistics Robotics

What is this category about?

• This category covers robots that move, identify or manipulate goods inside warehouses and logistics facilities. The products address processes such as inventory counting, order picking, palletizing and container unloading. 

• Warehouses are a particularly attractive environment for robotics because goods already move through defined processes. The challenge is that the goods themselves are highly variable. A warehouse robot may encounter thousands of product shapes, damaged packaging or pallets that do not match the expected digital record.

• The category therefore includes several distinct types of systems. Dexory automates the collection of warehouse data, while Nomagic and Progressive Robotics automate the physical handling of goods. NEOintralogistics moves inventory between shelves and picking stations, and Deus Robotics coordinates robots from different manufacturers.

• This has become the largest professional service robot market. More than half of the professional service robots sold globally in 2024 were designed for transportation and logistics, with sales in the segment increasing by 14% to approximately 102,900 units.

What do products in this category do?

• Warehouse robotics covers several different types of machines. Some companies build robotic arms that pick products or stack boxes onto pallets. Others build mobile robots that carry shelves or bins through the warehouse. A third group uses autonomous scanning robots to collect inventory data, while software platforms coordinate robots from different manufacturers. Ports and terminals create another use case, with specialized robots handling container equipment outdoors.

• Dexory builds mobile scanning robots for inventory control. A tall robot drives through warehouse aisles and scans pallet labels and rack locations at every height. Its software compares what the robot sees with the warehouse management system and shows missing stock, misplaced pallets and empty locations on a live map.

• Nomagic builds fixed robotic arm stations for picking and packing. A camera identifies the requested product inside a bin, after which the arm grips it and places it into an order box or sorting system. The robot can scan the item during the movement to confirm that it picked the correct product.

• NEOintralogistics builds mobile robots that bring goods to workers. A robot enters the existing shelving, lifts the required storage bin and carries it to a picking station. Once the item has been removed, the robot returns the bin to storage. NEO’s software decides which robot should complete each movement.

• Progressive Robotics builds fixed palletizing stations. Boxes arrive on a conveyor, where a 3D camera measures their size and position. A robotic arm then places each box onto the pallet, while the software continuously calculates a stable arrangement. The system can handle mixed boxes arriving in an unpredictable order.

• Servo7 is developing mobile robotic arms for heavy warehouse work. Its main use case is unloading loose boxes from shipping containers. The robot identifies the boxes inside the changing pile, removes them one by one and places them onto a conveyor or pallet.

• Deus Robotics provides the control layer for mixed robot fleets. It connects robots from different manufacturers with the warehouse management system, assigns tasks and manages traffic through one interface. This lets operators coordinate several robot types without using a separate control system for each supplier.

• SEAL Robotics builds specialized robots for ports and rail terminals. Cameras locate the twistlocks that secure a shipping container, and a robotic arm removes or installs them automatically. The same platform is being developed for related tasks such as pin handling and damage inspection.

What are the major trends shaping this category?

• Warehouse robotics is expanding from moving goods toward understanding and manipulating them. Autonomous mobile robots are already relatively mature, but picking a changing assortment of products remains difficult. Nomagic, Progressive Robotics and Servo7 are using vision and learned manipulation to automate the point where a robot must physically interact with irregular goods rather than simply transport a standardized shelf or pallet.

• Brownfield deployment is especially important in Europe. Many operators cannot close a warehouse or replace its racking to install a new automation system. NEO has built its product specifically around existing shelf infrastructure, while Dexory scans the warehouse without changing the operating layout. Nomagic also integrates its arms with established systems such as AutoStore.

• The commercial model is moving away from large upfront investments. NEO charges per completed pick, while Nomagic and Progressive Robotics offer leasing or service-based structures. This links the cost of automation more closely with order volumes and makes the system easier to justify in warehouses affected by seasonal demand. ASOS adopted Nomagic through a Robotics-as-a-Service structure that avoided an initial hardware investment.

• Warehouse automation is also becoming more modular. Operators increasingly combine storage systems, mobile robots and robotic arms from different suppliers. This creates demand for software that connects the equipment with the WMS and coordinates the overall workflow. Deus is building this control layer directly, while Dexory and Nomagic are expanding through partnerships with established warehouse-software and automation providers.

• The products are moving beyond individual tasks into broader operational platforms. Dexory began with inventory scanning but now uses the same physical data for space optimization, hygiene monitoring and operational recommendations. Nomagic has expanded from picking into packing and sorting, while its robots learn from production activity across customer sites. The long-term competition may therefore be over who owns the intelligence layer across several warehouse processes rather than who provides the best isolated robot.

• Labour availability remains an important driver, particularly for repetitive work performed during nights and peaks. The European Commission continues to identify transport-related occupations among areas affected by persistent shortages. Warehouse robots are therefore often sold around extending operating hours or reducing dependence on temporary peak labour rather than completely removing people from the facility.

What types of customers are buying these products?

• Large e-commerce and retail fulfilment operators. Zalando expanded its Nomagic pilot after the robots reached around 10,000 daily picks, while Brack.Alltron uses the system during autonomous Sunday and early morning shifts. Komplett has integrated Nomagic picking and packing stations with an existing AutoStore installation.

• Third-party logistics providers managing several customer inventories. ID Logistics uses Dexory robots to perform regular full site inventory scans and has expanded the system across multiple warehouses. Fiege and Arvato use Nomagic for item handling, while Deus deployed a 12 robot system with Nova Post. These operators value automation that can accommodate changing products and customer contracts.

• Operators of existing shelf based warehouses. NEOintralogistics works with Conrad Electronic and Versandmanufaktur, part of the GLS Group. Versandmanufaktur moved from one system to three while continuing normal operations, showing that automation can be expanded gradually rather than installed as one large project.

• Ports and terminal operators testing task specific robots. SEAL Robotics is expanding pilot projects in Northern Europe and Southeast Asia around container securing work. At this stage, these operators are primarily evaluating whether the robot can operate reliably outdoors and fit between existing cranes, vehicles and terminal procedures.

How does the funding environment look?

• Funding is highly concentrated in Dexory and Nomagic. Dexory raised a $100 million Series C in October 2025 as part of a $165 million financing package, following an $80 million Series B the previous year. The company has used the capital to expand internationally and develop DexoryView from an inventory product into a wider warehouse-intelligence platform.

• Nomagic raised $44 million in February 2025 and added another $10 million in early 2026. The funding is supporting larger deployments across Europe and further development of its AI models. Nomagic reported that customers were increasingly moving toward sites with more than ten robots, which suggests that financing is now being used to scale fleets rather than only fund initial pilots.

• The remaining companies are much earlier. NEOintralogistics raised a €3 million seed round, while Progressive Robotics has raised approximately €1.7 million across its early rounds. Deus Robotics has raised around $4.5 million, and SEAL Robotics secured a €1.8 million pre-seed round. Servo7 has so far recorded a $500,000 Y Combinator investment.

• European public and institutional capital has played a meaningful role. The European Investment Bank has financed Nomagic’s research, while the EBRD led its later equity round. Dexory has received backing from the British Business Bank. This support reflects the strategic importance of logistics automation but also the amount of capital required to manufacture robots and finance deployments before customer revenue is fully established.

• The category also shows why Robotics-as-a-Service requires significant financing. Vendors must build and install hardware before recovering the cost through monthly fees or usage payments. Better funded companies can therefore offer lower friction commercial models while supporting maintenance and expansion across several customer sites.