Physical AI sees rising interest but few firms move beyond pilots

• 
  
  Most companies are still early in physical AI use.
  
  


• 
  
  Only 5% say it is transforming operations, and 3% have fully integrated it.
  
  

A small share of companies have moved beyond testing AI in the physical world, and only a few have turned it into real operational change. New findings from Deloitte suggest that while interest in what it calls “physical AI” is rising, most organisations are still at an early stage of adoption.

The

firm’s latest paper

describes physical AI as the point where software systems move off screens and into machines. Such systems can sense their environment and make decisions, and they can also act in real time. In industrial settings, this includes robots that adjust to changes on a production line. It also includes systems that monitor quality and respond in real time.

Deloitte’s data shows that this change is still in its early phase. Only 5% of firms say physical AI is transforming their organisation today, while just 3% report that it is extensively integrated into operations. Both figures point to limited real-world deployment despite growing attention around the technology.

Expectations are rising. About 41% of organisations expect physical AI to have a transformative impact in the next three years. Deloitte also estimates that the share of companies with extensive integration could rise to 18% in two years. This suggests a sharp increase in adoption, though from a low starting point.

Slow progress beyond pilots

The gap between current use and future expectations highlights a familiar pattern in enterprise technology. Companies are quick to test new tools, but slower to rebuild their operations around them. In the case of physical AI, that gap is tied to integration. Software and data systems need to work as one environment, which adds complexity.

Industrial robotics has emerged as the main testing ground for this shift. According to data cited in the Deloitte paper, more than 500,000 industrial robots were deployed globally in 2024, and annual installations are expected to reach 700,000 by 2028. Collaborative robots are designed to work with humans. They accounted for close to 65,000 installations in 2024.

The systems are becoming a main entry point for physical AI adoption. As more robots gain sensing and decision-making abilities, factories are starting to move away from fixed

automation

. They are changing toward setups that can adjust based on real-time input.

The broader scale of change is also reflected in global robot use. A separate Citi GPS report cited by Deloitte estimates that around 405 million robots are currently in operation worldwide, and that number is projected to reach 1.3 billion by 2035. A growing share of these machines is expected to include some form of AI-driven ability.

Barriers slow adoption

Despite this growth, several barriers continue to slow adoption. Deloitte identifies cost and resource requirements as the most common challenge, cited by 41% of organisations. Another 36% say they struggle to identify clear use cases. A further 33% point to gaps in skills and talent. Data and technology limitations were also flagged by 31% of respondents.

The constraints reflect a broader issue. Physical AI is not a tool that can be added to existing systems. It requires changes to how operations are structured and managed. Machines need access to consistent data. Systems must communicate with each other. Teams also need to understand how the technology works in real operations.

Deloitte’s analysis suggests that progress depends as much on organisational readiness as it does on technical ability. Even advanced systems may deliver limited value if they are deployed in environments that are not designed to support them.

Chris Lewin, Deloitte Asia Pacific AI Lead, described the change in practical terms: “Physical AI is the moment when intelligence moves off the screen and into the real world, transforming factories into learning systems that sense, decide and improve continuously.”

From automation to adaptation

The report also points to differences in sectors. Adoption is expected to be highest in consumer industries and healthcare at 22%. Technology, media and telecommunications follow at 18%. Energy, resources and industrial sectors are at 16%.

The transition is likely to be gradual for manufacturers. Many firms are still working through pilot projects or limited deployments. Moving beyond that stage often requires updates to infrastructure, changes to workflows, and new approaches to managing operations.

Deloitte frames this as a change from installing technology to building ability. Systems need to be tested and refined before they are integrated into daily processes. Organisations need to develop the skills and structures that allow these systems to operate effectively.

The data points to a clear direction. Physical AI is gaining attention and investment, but most companies have not yet reached large-scale deployment. The next phase will depend on how quickly organisations can align their systems and teams to support it.

For now, the technology is moving faster than the changes required to use it fully. That gap may determine which companies can turn early experiments into lasting operational change.

Want to learn more about AI and big data from industry leaders?

Check out

AI & Big Data Expo

taking place in Amsterdam, California, and London. This comprehensive event is part of

TechEx

and is co-located with other leading technology events, click

here

for more information.

Tech Wire Asia is powered by

TechForge Media

. Explore other upcoming enterprise technology events and webinars

here

.