Arm Holdings has restructured its operations to create a Physical AI business unit. The move expands its footprint in robotics, as automation systems dominated this year's Consumer Electronics Show.
Exhibitors at the Las Vegas trade show demonstrated robots designed to assemble vehicles, clean industrial facilities, and handle repetitive factory tasks. Reuters reported the creation of the Physical AI unit and the accompanying corporate reorganization.
The UK-based chip designer will now operate through three main business lines. These include Cloud and AI, an Edge unit covering mobile devices and personal computers, and Physical AI, which houses the company's automotive activities.
Arm licenses processor designs rather than manufacturing chips. Its architecture powers most global smartphones and an increasing share of laptops and data centers. The company generates revenue through licensing fees and royalties paid by chipmakers.
Robotics and vehicles at the core
Robotics and vehicles form the core of Physical AI. These systems must sense environments, make decisions, and act in the physical world with strict limits on power, speed, and safety. Ami Badani, Arm's chief marketing officer, said the requirements for robots and cars overlap significantly. This led the company to combine the two segments into one unit.
Automotive and robotics converge
The move reflects a broader industry trend. Car companies and robotics firms now share hardware and software. Tesla is developing humanoid robots to automate warehouse work. Meanwhile, robotics companies are adopting sensors and real-time control software originally built for the automotive sector.
Arm designs are already used by dozens of automakers and robotics firms, such as Boston Dynamics. Boston Dynamics and its owner, Hyundai, recently debuted a production-ready version of the Atlas humanoid robot. Hyundai plans to deploy the machines in US factories by 2028.
The expansion is part of a growth strategy led by Chief Executive Rene Haas. Since taking over four years ago, Haas has increased prices for some technologies and explored offering full chip designs. Drew Henry, who leads the Physical AI unit, told Reuters these systems could improve labor productivity and increase economic output over the long term. Badani said the unit plans to hire more staff specifically for robotics development.
Edge computing takes center stage
Dozens of exhibitors at CES showcased humanoid robots performing tasks like sorting objects or playing games. Many use on-device artificial intelligence rather than relying on distant data centers. C.J. Finn, a US automotive industry leader at PwC, said investment is shifting toward applications that integrate mechanical systems with AI to enhance precision.
Robert Playter, chief executive of Boston Dynamics, said humanoid robots remain in a hype phase. Yet his company sells thousands of smaller, four-legged robots for a profit.
Arm executives said the reorganization mirrors a shift where AI moves from the cloud into physical devices at the "edge" of the network. The company noted that many robots and vehicles at CES utilize its architecture because the processors are optimized for low-power, high-speed computing.
Partners at the show highlighted this ecosystem. Nvidia demonstrated robotics hardware built on Arm designs. Qualcomm introduced new processors for industrial robots that also use Arm technology. Both companies showcased vehicle software stacks that Arm executives said demonstrate how automotive technology can transition into the robotics market.
The surge in activity reflects how tech and car companies view human-form robots as a primary frontier for automation. Tesla CEO Elon Musk has said the Optimus robot could eventually become more economically important than the company's car business. Separately, Mobileye announced plans to acquire robotics firm Mentee for US$900 million. Nvidia also introduced new software tools designed to power autonomous machines.
Betting on the future of intelligence
Arm executives said the common requirement across these deals is the need for efficient, safe computing. CES highlighted the closing gap between AI, robotics, and cars. The company is betting its architecture will serve as the foundation for intelligence moving into factories and onto the streets.
Some analysts note that as artificial intelligence moves further into physical devices, longer-term questions are emerging around CPU architectures beyond x86 and Arm. RISC-V, an open instruction set architecture, has gained traction in embedded and control functions within AI systems. But it remains constrained by limited commercial software support. Industry observers say its ability to challenge established platforms will likely depend on whether future AI hardware cycles create opportunities for new software ecosystems to take hold. This is particularly true in edge and consumer devices.
Article edited by Jerry Chen
