Skip to content
Artificial intelligence. Clearly. RSS feed
AI NEWS AI

News. Evidence. Perspective.

BusinessAnalysis

Renesas sees its next AI opportunity inside robot joints

The chipmaker is expanding from individual components into complete robotics designs. Its ambition to serve more of a humanoid's parts list reveals where the business of physical AI extends beyond the main processor.

The wrist of a humanoid robot is crowded. Motors, wires and control electronics must fit into a small space while producing precise movement. Renesas is shrinking semiconductor packages for joints such as wrists and ankles, according to The Elec's account of a company presentation in Seoul on 16 October. Those cramped spaces are becoming part of the Japanese chipmaker's growth plan.[1]

Director Yang Tae-young told the conference that Renesas had more than 108 people in its physical AI team, The Elec reported. The company's role is expanding into engineering support and system architecture: helping customers assemble the sensing, processing, power and movement capabilities a humanoid needs.[1]

The strategy has been taking shape for months. Renesas established its Physical AI Division on 1 July and opened a robotics laboratory in Beijing on 27 August. At the lab's opening, executive Ivo Marocco said its portfolio could address roughly 30% of a humanoid robot's bill of materials—the parts required to build it—with an opportunity to reach 70%. Those figures describe the company's proposed breadth of supply; they are not shares of the worldwide robotics market.[2]

Many chips share the movement

A robot's most conspicuous intelligence may be recognising an object or interpreting a request. Carrying out the movement also requires it to measure joint position, regulate motor power and respond to contact. Renesas' humanoid portfolio spans those jobs, from position sensors and motor-control processors to battery management and vision AI. The commercial opportunity grows with the number of things that must work together.[5]

The division of labour is visible inside an existing Renesas processor. Its RZ/V2H, introduced in 2024, combines Linux application processors and an AI accelerator with dedicated real-time processors. Renesas' engineering explanation assigns image AI to one side and time-sensitive tasks such as motor control to the other, connecting them through communication between processors.[4]

That arrangement addresses a physical problem: deciding what an object is and keeping a motor on schedule have different computing demands. A delay that is tolerable while processing a camera frame may be too long for a control loop adjusting movement. The design lets the perception system inform the movement without asking one general-purpose software environment to handle every timing requirement.[4]

In a demonstration described in its September engineering article, Renesas combined vision, tactile sensing and force feedback in a robotic hand. The hand adjusted its grip across a can, a paper cup and an egg. The interesting engineering task is the connection between those capabilities: recognising an egg helps only if the fingers can apply an appropriate force as they touch it.[3]

Engineering support becomes part of the sale

Renesas' Beijing lab is intended for joint development and validation of whole systems. Its Renesas 365 platform connects design information, firmware and hardware, while allowing engineers to deploy changes remotely and watch measurements such as processor utilisation, power consumption and frame rate. These tools give the chip supplier a role throughout the process of making a robot work.[2][3]

There is already a business to build on. Yang put humanoids at 8% of revenue from the joint-motor-control and industrial-networking products discussed in his presentation, according to The Elec. The denominator matters: this was a slice of particular product lines, rather than Renesas' total sales.[1]

The strategic attraction is that a supplier can serve several robot makers at once. It can win more business when a customer adds a joint, changes a sensor or redesigns the power system, even if that customer buys its main AI processor elsewhere. Engineering support can make those later decisions easier to keep within the same supplier's portfolio.

Renesas' 70% ambition therefore points to a broad contest for the contents of a robot. A useful hand, a stable ankle and a battery that lasts through a shift each create a design decision—and potentially another component order.

Follow the evidence

Sources & further reading

  1. Beyond Silicon: Humanoid Robotics Design Is a Team Sport Renesas · 8 Sep 2026 · Accessed 18 Oct 2026
  2. Humanoid Robots Renesas · Accessed 18 Oct 2026