Physical AI is moving from an emerging technology topic to a strategic enterprise priority.

HCLTech’s The AI Impact Imperatives, 2026 report found near-universal (90%) agreement among respondents that Physical AI will be critical or important to organizational success over the next three years. But the same research also shows that deployment remains early: organizations not yet exploring Physical AI, researching or evaluating it, piloting solutions or only just deploying initial systems outnumber those with many Physical AI systems in production by more than four to one (81% versus 19%).

For Sukant Acharya, EVP, HCLTech, that gap between ambition and scaled deployment reflects both the size of the opportunity and the complexity of making Physical AI work in real operating environments.

“When we look at the physical dimension, the products and services we consume are intertwined with physical things, physical objects, physical products, facilities and people. The potential is very high,” said Acharya.

That need is becoming more urgent across industries facing labor constraints, rising demand and pressure to improve output. Physical AI has become important because so much of the global economy still depends on physical products, assets, facilities, infrastructure and human operations. A small improvement in how those environments are monitored, optimized and coordinated can have a significant multiplier effect.

The timing also matters. Advances in sensors, actuators, connectivity, edge computing, cloud, in-memory computing, Generative AI and vision-language models are making Physical AI more practical. These technologies allow systems to perceive, analyze, decide and act in physical environments in ways that were previously difficult to scale.

“When there is demand, there is need and there is potential, but also feasibility coming from innovation in the technology arena, it’s a perfect recipe,” said Acharya.

Where progress is happening

Progress is visible, but uneven.

Acharya sees Physical AI developing across three areas: AI-enabled products, AI-assisted assets and AI embedded into operational processes.

AI-enabled products include areas such as autonomous vehicles, where perception, simulation and decision-making capabilities continue to advance. AI-assisted assets include operational environments where AI can improve asset life, asset efficiency and reliability. AI in operational processes includes areas such as warehouse robotics, picking, packing, sorting and specialized quality inspection in manufacturing. The opportunity extends across other industries as well, from reducing turnaround times in oil and gas to reducing inventory shrinkage in retail and improving patient experience in healthcare.

HCLTech’s report supports this shift. It found that 79% of respondents said AI is being used in production control systems, to manage product quality, to automate robotics and to make predictive maintenance more intelligent.

But many examples remain narrow and task-specific. The larger opportunity is to use Physical AI to reshape how value chains operate across functional boundaries. That means insights from one area feeding into another in real time, agents interacting with other agents and operational decisions becoming faster, more connected and more adaptive.

Those examples are still limited because the change burden is high. Physical AI affects how work is designed, how people interact with machines, how operational decisions are governed and how risk is managed.

Why partnerships matter

Physical AI is difficult to scale because it sits at the convergence of multiple disciplines.

“We are talking about a hyperconvergence of three distinct technology principles: engineering technology, operational technology and information technology,” said Acharya.

That convergence is important. Engineering technology brings the physics, design and product understanding. Operational technology brings machines, PLCs, robotic systems and control environments. Information technology brings data, intelligence, cloud, cybersecurity and decision support.

Enterprises also need to make the right compute decisions. Physical AI depends on a compute-smart approach that determines what should happen inside the product, what should happen at the Edge and what should move to the cloud.

“Not everything has to be done in cloud. Not everything has to be done at the Edge. Not everything has to be done in memory. What needs to be processed at the Edge, what will go to the cloud and what will happen in memory inside the product becomes a very critical decision,” said Acharya.

This is one reason partners can become important. HCLTech’s report found that organizations working with third-party experts on AI initiatives are more likely than those that do not to have deployed Physical AI systems to production, by 54% to 26%.

Partners can help enterprises bring together the right engineering, OT, IT, Edge, cloud, data and platform capabilities. They can also help organizations avoid rebuilding every component from the ground up by assembling modular micro-capabilities into more scalable platform-centric models.

Why Physical AI exposes readiness gaps

Physical AI exposes enterprise AI readiness gaps more sharply than many digital AI use cases.

Organizations need to think about three Cs: convergence, change and consequence.

The first is convergence. Physical AI brings AI into a world of people, machines, products, facilities and devices, while also bringing together engineering technology, operational technology and information technology.

The second is change. Physical AI cuts across functions and affects operational processes that may be critical to output, safety and continuity. It can require new operational standards, safety practices and collaboration models between people, machines and robots.

The third is consequence. If a digital AI prediction is wrong, the impact may be contained or reversible. If a Physical AI system fails, it can affect product quality, disrupt downstream operations, shut down facilities or create safety risks.

That makes governance, cybersecurity, safety and human oversight essential from the beginning. Physical AI cannot be scaled through experimentation alone. It needs the right operating model, the right decision rights and the right controls.

The competitive edge is speed

For organizations moving early, Acharya believes the biggest competitive differentiator will be speed. There is no standard playbook or single operating model for Physical AI because each enterprise has different assets, processes, environments and risk profiles.

The operating model needs to be designed around the context of the organization. Industrialization capability can be built with the help of partners, but speed will be decisive: the speed of decision-making, pilots, value proof, learning and scaling.

Organizations need to move with speed.

That does not mean moving recklessly. It means moving with discipline and avoiding the pilot trap, where organizations experiment for months without proving value or designing for scale.

Acharya points to three principles for success: “think expansive, think adaptive and think future-proof.”

Enterprises need to think expansively about where Physical AI can create value across operations. They need to think adaptively because technologies, business conditions and operating environments will keep changing. They also need to think future-proof from the start, so solutions do not become outdated by the time a pilot is complete.

Physical AI will not scale through hype or isolated experimentation. It will scale when organizations combine the right technologies, partners, governance and operating models to create measurable impact in the physical world.

The opportunity is significant, but so is the execution challenge. The enterprises that move fastest with discipline will be best placed to turn Physical AI into a source of competitive advantage.