Robotics is moving from programmed automation towards physical AI: machines that can perceive, decide and act in environments designed for people. The technical progress is real. The commercial challenge is whether robots can perform useful work safely, reliably and economically enough for customers to deploy them repeatedly.
This hub covers humanoid robotics, industrial automation, sensing, touch, dexterity and the systems required to move from impressive demonstrations to dependable operations. It positions robotics as the next major test of technology commercialisation.
The Core Commercialisation Argument
Robotics has spent decades succeeding in structured environments. Industrial robots weld, paint, assemble and move products with extraordinary precision when the task, object and workspace are controlled. The next opportunity is broader: machines that can work across more variable environments, learn new tasks and interact safely with people and unpredictable objects.
Artificial intelligence has accelerated that ambition. Better perception, planning, simulation and language interfaces allow robots to understand tasks that previously required extensive programming. Hardware costs are falling, investment is increasing and companies are moving from prototypes into early deployments.
Yet commercial value is not created by intelligence or movement alone. Customers buy productive output. A robot must complete tasks at the required speed and quality, operate for enough hours, recover from errors, integrate with existing systems and deliver a return after support, supervision, maintenance and downtime are included.
The robotics opportunity is therefore large, but the deployment test is demanding. The companies that win will build complete operating systems around useful work rather than treating the robot body or AI model as the product.
What Is Physical AI?
Physical AI describes intelligent systems that act in the real world. Unlike software operating entirely inside digital environments, a physical-AI system must deal with friction, force, variation, uncertainty, wear, safety and the consequences of mistakes. It needs to understand not only what an object is, but how it moves, feels and responds during contact.
This makes robotics a systems problem. Artificial intelligence may plan an action. Sensors must provide reliable information. Actuators and control systems must execute the movement. The mechanical design must tolerate repeated use. Safety systems must manage failure. The organisation deploying the robot must redesign workflows, train staff and maintain the equipment.
The Deep Tech hub places physical AI within the wider industrial context of semiconductors, sensors, advanced materials, supply chains and commercialisation. Robotics draws on each of those domains, making it a useful test of whether a deep-tech ecosystem can combine capabilities into a dependable product.
From Demonstrations to Early Deployment
The robotics market has crossed an important threshold. Humanoid and general-purpose robots are no longer only research projects or conference demonstrations. Companies are shipping early units, industrial customers are running pilots and investors are funding complete technology stacks around embodied intelligence.
The analysis of the humanoid robotics market in 2026 shows how China, the United States and Europe are developing different advantages. China combines manufacturing scale, component supply chains and deployment volume. The United States combines artificial intelligence, software and venture capital. Europe brings industrial partnerships, safety capability and regulatory trust.
Shipments and funding are evidence that the sector is commercially serious. They are not evidence that it is commercially proven at scale. The more useful measures are productive hours per robot, intervention rates, task-completion reliability, cost per useful hour, integration time and repeat orders across multiple sites.
Robotics commercialisation will accelerate as those measures improve. It will also expose the difference between companies optimised for demonstrations and companies designed for deployment.
Why Humanoid Robotics?
The commercial case for humanoid robots begins with the environment. Factories, warehouses, hospitals, shops and homes have been designed around the human body. Doors, stairs, tools, shelves, vehicles and workstations assume human reach, movement and manipulation.
Traditional automation often requires the environment to be redesigned around the machine. A humanoid form aims to reverse that relationship by creating a machine capable of using spaces and tools that already exist. If successful, that could expand automation into tasks where dedicated machinery is too inflexible or expensive.
The form factor also creates significant engineering and economic challenges. Two-legged movement consumes energy and introduces stability risk. Human-like hands are complex and expensive. A general-purpose machine may perform individual tasks less efficiently than specialised automation. The useful comparison is therefore not humanoid versus human in the abstract. It is humanoid versus the best available combination of people, process change and specialised machinery for a defined task.
Near-term success is likely to come from constrained industrial uses where the environment is sufficiently predictable, labour demand is persistent and the value of flexibility justifies the additional complexity.
Industrial Robotics and Automation
Industrial robotics provides the strongest foundation for the next generation of automation. Manufacturers already understand how to evaluate cycle time, uptime, safety, maintenance and return on investment. They also possess the operational discipline and engineering teams required to integrate machines into production.
The next wave will extend automation beyond fixed cells and highly repetitive tasks. More capable perception and control systems can allow robots to handle variable products, move between work areas and support tasks that change over time. Collaborative systems can work closer to people where complete isolation is impractical.
Commercial adoption will still depend on process design. A robot should not be deployed simply because a task can be automated. The task must be valuable enough, stable enough and measurable enough to support the integration cost and operating risk. In many cases, the strongest result will come from changing the workflow around the combined strengths of people and machines.
Industrial customers buy throughput, quality, safety and resilience. Robotics companies that frame their proposition around those outcomes will have a stronger route to scale than companies selling general capability without a clear operational owner.
Sensing Is the Foundation of Real-World Intelligence
Robots cannot act reliably on information they do not possess. Cameras and machine vision have transformed perception, but real-world work requires multiple forms of sensing. Force, pressure, proximity, temperature, vibration and contact all help a robot understand what is happening during a task.
The analysis of AI sensor technology explains why sensing defines the boundary between intelligence that interprets the world and intelligence that can operate inside it. Better models can improve planning and recognition. They cannot eliminate the need for accurate physical feedback.
Sensor systems must also commercialise. They need suitable cost, form factor, durability, calibration, interfaces and manufacturing consistency. A sensor that performs brilliantly in a laboratory but cannot survive industrial contact or integrate into a control system does not solve the customer’s problem.
Sensing is therefore both an enabling technology and a commercialisation challenge in its own right. Its value appears through improved task reliability, lower damage, safer operation and reduced intervention.
Touch and Dexterity: The Missing Feedback Loop
Dexterous manipulation is one of the most important remaining barriers in robotics. Vision can identify an object and estimate its position. Reliable handling also requires information about grip, force, movement and slip. Humans make continuous adjustments through touch without conscious effort. Robots often lack that feedback loop.
The analysis of why touch is the missing piece in humanoid robotics argues that locomotion is no longer the only defining challenge. Robots must handle damaged packaging, delicate food, irregular components and objects whose behaviour changes during contact. Those conditions are common in commercially valuable environments.
The Kirisense-related analysis, As AI Gives Robots Brains, Touch Will Give Them Dexterity, examines the role of tactile sensing in detecting force and slip. As Chair of Kirisense, Tim Harper is working directly with this commercialisation challenge: translating sensing capability into a subsystem that can improve real robotic tasks.
The commercial test is not whether a tactile sensor can detect contact. It is whether the complete system improves manipulation reliability enough to justify integration, data processing, support and cost. If it does, touch becomes a foundational part of the robotics stack.
The Economics of Robotic Work
Robotics economics are often reduced to the purchase price of a machine compared with the cost of labour. That comparison is incomplete. Customers need to understand cost per useful task or productive hour after integration, supervision, maintenance, energy, software, downtime and process change are included.
A lower-cost robot can be expensive if it requires frequent intervention or fails unpredictably. A high-cost system can be attractive when it operates reliably across multiple shifts, improves safety or solves a labour constraint that prevents production. The value depends on the task and operating model.
Utilisation is critical. Capital equipment creates value while it is productive. Robots that spend long periods charging, waiting for work, recovering from faults or being reconfigured carry a higher effective cost. Fleet-management software, maintenance capability and workflow design can therefore matter as much as the robot’s headline performance.
The strongest commercial cases begin with a measurable customer constraint and build the deployment around it. General capability expands the future market. Specific operating value creates the first repeatable sales.
The Seven Barriers to Robotics Deployment
Robotics exposes all seven barriers in the Technology Commercialisation framework:
- Technical readiness: the robot completes useful tasks reliably across real variation.
- Infrastructure readiness: sites have suitable power, connectivity, safety systems, data and maintenance support.
- Economic competitiveness: cost per useful output supports an acceptable customer return.
- Regulatory alignment: safety, liability, workforce and AI rules provide a workable deployment pathway.
- Supply chain maturity: actuators, sensors, semiconductors, rare-earth materials and manufacturing can scale.
- Capital availability: funding supports hardware development, production and customer deployment.
- Market adoption: organisations can integrate robots into workflows and trust them with valuable work.
The controlling barrier will vary by application. A factory pilot may be limited by technical reliability. A hospital deployment may be limited by safety, trust and workflow integration. A humanoid company may be limited by manufacturing cost or capital. Commercialisation improves when the barrier is stated explicitly rather than hidden inside a general claim that the market is not ready.
Robotics Infrastructure Is More Than Charging
Robots depend on physical and digital infrastructure. Sites need power, charging or battery-swap systems, connectivity, mapping, safety controls, maintenance processes and integration with warehouse, manufacturing or enterprise software. These dependencies affect deployment timing and cost.
Data infrastructure is particularly important. Physical-AI systems improve through operating data, simulation and model updates. Companies that connect deployment, failure analysis, training and redeployment can build a powerful learning loop. Customers will still require governance around data ownership, cybersecurity and operational control.
Infrastructure readiness also includes people. Technicians must maintain machines. Operators need escalation paths. Managers need metrics that distinguish useful output from novelty. A robot placed into an organisation without support infrastructure may demonstrate capability while failing to become an operating asset.
Safety, Regulation and Trust
Robotics moves intelligence into environments where mistakes can damage products, interrupt operations or harm people. Safety cannot be added after the commercial proposition has been designed. It affects hardware, software, testing, deployment scope, insurance, liability and customer trust.
Regulatory alignment can slow deployment, but credible standards can also create markets by reducing uncertainty. Customers are more willing to adopt systems when responsibilities, certification and operating boundaries are clear. Suppliers that treat compliance as evidence of dependable execution may gain an advantage over companies that view it only as friction.
Trust also develops through operational evidence. A customer may accept limited capability from a system with transparent performance and failure modes. It will hesitate to depend on a more capable machine whose behaviour is difficult to predict or explain. Commercial robotics must make reliability legible to the organisations using it.
Semiconductors, Rare Earths and the Robotics Supply Chain
Robotics depends on a global supply chain of semiconductors, sensors, motors, actuators, batteries, precision components and rare-earth materials. Cost and availability across that stack influence whether companies can move from prototypes to dependable volume production.
The semiconductor cluster analysis provides context on the ecosystems behind compute and control hardware. The examination of semiconductor supply-chain disruption shows why qualified inputs can become strategic constraints. The analysis of rare-earth export controls matters directly to motors and actuators.
Robotics companies must decide which capabilities they need to own, which suppliers require redundancy and where design changes can reduce dependency. A compelling prototype built from scarce or expensive components may prove the concept while leaving the commercial product unresolved.
Robotics Deployment Is Organisational Change
A robot changes the system around it. Tasks are reassigned. Workflows and layouts may change. Staff need training. Supervisors need new metrics and escalation processes. Maintenance and safety responsibilities must be clear. The organisation must decide how people and machines work together.
Market adoption will therefore depend on more than technical performance. Customers need confidence that deployment will improve operations without creating unacceptable disruption. Successful robotics companies will help customers manage that transition, not simply deliver hardware.
Workforce concerns also need to be handled honestly. Robotics can address labour shortages, improve safety and remove repetitive work. It can also change roles and create anxiety. Organisations that involve employees, define the operational purpose and invest in new skills are more likely to capture value than organisations deploying automation as a symbolic replacement strategy.
Capital, Manufacturing and the Scale-Up Gap
Robotics companies face a difficult capital path. They must fund hardware development, software, testing, manufacturing, inventory and customer support before the economics of volume production are proven. Growth can increase working-capital pressure even when customer demand is strong.
Capital availability must therefore match the evidence and timescale of the business. Venture funding may support rapid technical learning. Strategic investors can provide manufacturing capability, components or customer access. Customer financing, leasing and robotics-as-a-service models can reduce the adoption barrier while shifting capital requirements back to the supplier.
Governance becomes critical as the company scales. The Scale-Up Governance and Deep-Tech Commercialisation pages examine how boards connect capital allocation to the next commercial proof point. Robotics ventures need to resist expanding across too many use cases before one deployment model is repeatable.
Kirisense and the Commercialisation of Robotic Touch
Kirisense provides a focused example of how an enabling robotics technology moves towards market. Tactile sensing addresses a clear technical barrier: robots need richer information about force, contact and slip to manipulate variable objects reliably. The addressable opportunity spans industrial robotics, logistics, food handling, healthcare and future humanoid systems.
The commercialisation task is narrower and more demanding than the opportunity statement. The sensor must demonstrate relevant performance, integrate into hands and grippers, survive repeated use, support control software and be manufacturable at a cost that fits the customer’s economics.
External validation and collaborative development can reduce technical risk, but customer evidence will determine the commercial path. Which task improves? How much intervention or damage is avoided? What integration work is required? Can the improvement be repeated across customers and robot platforms?
As Chair, Tim Harper’s role connects that technical development to governance, market definition, partnerships and the evidence required for scale. It is a practical continuation of the same commercialisation discipline applied across advanced materials and energy infrastructure.
A Practical Route From Robotics Pilot to Scale
- Select a valuable task. Start where the customer has a measurable constraint and enough repetition to evaluate performance.
- Define productive evidence. Measure completed tasks, reliability, intervention, quality, safety and cost per useful output.
- Design the operating system. Include integration, maintenance, connectivity, charging, supervision and escalation.
- Reduce the controlling barrier. Focus the pilot on the uncertainty that prevents a buying or deployment decision.
- Build customer capability. Help the organisation redesign workflows, roles and metrics around the system.
- Prove repeatability. Convert one successful deployment into a process that can work at another site.
- Align manufacturing and capital. Scale production only when the deployment model creates credible demand and unit economics.
A robotics pilot is commercially useful when it changes a decision. It should show whether the customer can deploy the system, whether the supplier can deliver it repeatedly and what must be improved before scale.
The Future of Robotics Commercialisation
Robotics capability will continue to improve quickly. AI will make systems easier to train and more adaptable. Simulation and fleet data will accelerate learning. Component costs will fall as supply chains scale. Those trends expand what robots can do.
Deployment will remain the source of differentiation. Companies with access to real operating environments will learn which failures matter, which tasks create value and how customers integrate machines. That evidence will improve both technology and commercial design.
The most successful robotics companies may not be the ones with the most dramatic demonstrations. They will be the system builders that combine intelligence, hardware, sensing, service, manufacturing and customer change into dependable productive capacity.
Robotics is therefore not only the next AI market. It is the next major technology-commercialisation challenge.
Robotics and Physical AI: Frequently Asked Questions
What is physical AI?
Physical AI refers to intelligent systems that perceive, decide and act in the real world. It combines AI models with sensors, controls, actuators, mechanical systems and the operational infrastructure required for safe and reliable physical work.
Why are humanoid robots attracting investment?
Humanoids could operate in environments and use tools designed for people, expanding automation without rebuilding every workplace around specialised machinery. The opportunity is large, but the economics and reliability of general-purpose deployment remain unproven at scale.
Why is touch important in robotics?
Touch provides information about force, pressure, contact and slip. It helps robots adjust grip and manipulate variable or delicate objects, addressing tasks where vision alone cannot provide enough feedback for dependable performance.
What determines whether a robot is commercially viable?
Commercial viability depends on productive output after purchase price, integration, supervision, maintenance, energy, support and downtime are included. Customers need a credible return and confidence that the robot can operate safely inside their workflow.
What is the biggest barrier to robotics adoption?
The controlling barrier depends on the use case. Reliability and dexterity may dominate one deployment, while safety, workflow integration, cost or organisational readiness dominate another. The useful approach is to identify and reduce the specific barrier preventing a repeatable customer decision.
Continue the Analysis
- Technology Commercialisation – the parent framework and Seven Barriers to Deployment.
- Deep Tech – advanced materials, semiconductors, sensors and industrial innovation.
- Humanoid Robotics in 2026 – market structure, investment and deployment evidence.
- The Missing Piece Is Touch – why dexterity is a controlling barrier.
- As AI Gives Robots Brains, Touch Will Give Them Dexterity – Kirisense and tactile sensing.
- AI Sensor Technology – the sensing layer behind physical AI.
- Global Semiconductor Clusters – the industrial ecosystem behind compute.
- About Tim Harper – commercialisation experience across changing technologies.