Asia seen as sweet spot in physical AI

ASIA is emerging as a key investment sweet spot as physical artificial intelligence (AI) moves from the laboratory into factories, logistics networks and other real-world applications.

As the technology scales, the region’s manufacturing depth and hardware expertise could put Asian companies at the centre of the next phase of the AI revolution.

That is the view of Eastspring Investments, with portfolio managers Eric Lin and Jack Hsu of Eastspring Taiwan highlighting Asia’s strengths in the hardware and supply-chain ecosystems needed to make physical AI work.

Physical AI is already making its way into the real world through autonomous systems such as drones, AI-powered smart factories, robots and self-driving cars.

Unlike generative AI, which mainly creates and processes information, physical AI allows machines to perceive, understand, reason and perform complex actions in physical environments.

Its applications are spreading across industries, from manufacturing and logistics to healthcare and transport.

Amazon’s fulfilment centres around the world, for example, are powered by more than one million robots that sort, lift and transport packages to speed up deliveries.

Self-driving taxis are also operating in more than 10 countries, while utility companies are using drones to inspect hazardous environments such as high-voltage lines and open gas pipes.

In healthcare, more than 60% of large hospitals worldwide have integrated surgical robotics into their core platforms, while robotic-assisted procedures account for more than half of complex surgeries at healthcare systems in developed countries.

Lin and Hsu say physical AI remains at an early stage, with limited real-world training data historically constraining development.

Most robotics solutions still require some degree of human supervision, although advances in simulation platforms are providing synthetic training data at scale and opening the door to more autonomous applications.

The potential market is sizeable. An estimated two million humanoid robots – machines designed to look and behave like humans – could be in workplaces by 2035, rising to 300 million by 2050.

This could create a total addressable market of between US$1.4 trillion and US$1.7 trillion, excluding other physical AI categories such as industrial and service robots, autonomous systems and drones.

“Physical AI represents the next phase of the AI revolution where intelligence moves from the cloud into the real world,” Lin and Hsu say.

That shift also creates a different set of requirements from generative AI.

Physical AI needs to compute at the edge, meaning data is processed locally on a device instead of being sent to a centralised cloud or data centre.

This is critical because machines need instant local inference to make real-time autonomous decisions and avoid failures or safety risks. They also cannot rely entirely on network connectivity, which can be intermittent or slow.

These requirements create constraints around power, size and cost.

Devices need to balance battery life with heat management, while computing capabilities must fit inside increasingly compact machines. At the same time, hardware needs to be affordable enough for mass deployment.

“These constraints shift value from software to hardware, where Asia dominates,” Lin and Hsu say.

This is where the region’s investment appeal comes into focus.

Robots, autonomous systems and industrial machines depend heavily on sensors, semiconductor chips, actuators, controls and other precision components.

For robotics, the development of robotic hands that can match the dexterity, precision and adaptability of the human hand is particularly important.

“These are areas in which Asia has a competitive advantage,” Lin and Hsu say.

Asia’s advantage comes from a combination of manufacturing depth, integrated ecosystems and the ability to deploy technologies in real-world settings.

The region has decades of experience producing high-precision components at scale, particularly for consumer electronics.

That expertise gives Asian manufacturers cost efficiencies that become increasingly important as physical AI moves from pilot projects towards mass deployment.

Its deeply integrated supply chains also allow manufacturers to iterate faster, reduce integration costs and shorten lead times.

Such ecosystem density matters because physical AI requires close coordination between sensors, computing modules and mechanical systems.

“Asia’s edge in physical AI hardware is rooted in its manufacturing depth, integrated ecosystems, and real-world deployment advantages,” Lin and Hsu say.

The region’s real-world deployment environment provides another advantage.

As one of the world’s core manufacturing hubs, Asia has established supply chains, supportive government policies and a wide range of applications across factories, logistics networks and automated production systems.

China illustrates how these factors can come together.

Government initiatives have promoted robotics and autonomous driving, while companies continue to invest in product development and commercialisation.

This is gradually creating an ecosystem spanning components, system integration and end applications.

Lin and Hsu point to examples ranging from the large-scale coordinated control capabilities demonstrated by robots at the Spring Festival Gala to humanoid robots beginning to work on production lines at electric vehicle factories.

“Together, these developments highlight how physical AI in Asia is moving from concept to real-world deployment,” they say.

There are, however, risks that investors need to factor in.

The geographical concentration of supply chains leaves production exposed to geopolitical tensions and trade restrictions.

Regulatory constraints, privacy concerns and fragmented standards across markets could also make it harder for physical AI companies to scale.

At the same time, continued investment in research and development in areas such as advanced chip design, packaging and system architecture remains necessary to maintain Asia’s lead in edge AI optimisation.

For investors, the opportunity is therefore extending beyond the companies benefitting from generative AI and agentic AI.

Lin and Hsu see an expanding universe of physical AI enablers, including semiconductor manufacturers, sensor makers, precision component suppliers and electronics manufacturing services providers, many of which are based in Asia.

“In physical AI, constraints around latency, power, size and cost shift value toward hardware and system integration, reinforcing Asia’s strengths,” they say.

The next stage will depend on how companies navigate geopolitical tensions, changing regulations and evolving industry standards.

For investors, this makes company selection increasingly important as the physical AI ecosystem develops and technologies move from experimentation into commercial-scale deployment.

“As these companies navigate geopolitical tensions as well as changes to regulatory frameworks and industry standards, active investing will be needed to identify the alpha opportunities and companies that are most likely to succeed,” Lin and Hsu say.