Korea Funds the Sensors, Motors and Batteries Robots Need

South Korea announced on November 11 that it will fund three physical AI components: next-generation sensors, robot actuators and humanoid batteries. The sensor work centers on on-sensor AI, chips that analyze data where it is captured. It is a sign that governments now see robot parts, not just robots, as a strategic industry to back.

• Finance Minister Lee Hyoung-il announced the plan in Seoul on November 11, 2026.

• Sensors, actuators and humanoid batteries join 15 existing projects in Korea's growth plan.

• The sensor task targets on-sensor AI for robotics, defense and mobility.

• Korea aims to raise the local share of core robot components to 80 percent by 2030.

• Sony and TSMC plan a joint image sensor venture in Japan aimed partly at robotics.

What Korea announced

South Korea has added three physical AI components to its list of state-backed growth projects. Finance Minister Lee Hyoung-il presented them at a meeting of economy ministers in Seoul on November 11. They join 15 existing projects, including next-generation chips, solar energy and power grids.

The finance ministry explained its reasoning in a press release:

Each component will have its own track. For sensors, the focus is on-sensor AI for robotics, defense and mobility. For actuators, the goal is lighter, stronger and more durable parts built on domestic technology, with Korean actuators in mass production by 2029.

The battery track covers cells with higher output and energy density, plus next-generation all-solid-state batteries for humanoid robots. Korean reports say the government will fund the sensor task with 49.8 billion won between 2027 and 2030.

Actuators and batteries get the bigger checks

The sensor track has the smallest budget of the three, but it is the most novel. According to Korean reports, next year's budget includes about 9 billion won for on-device AI that handles both sensing and judgment inside the sensor. Another 1.8 billion won is set aside for training specialists, and a public and private task force will steer the work.

Actuators draw more. Reports put next year's actuator research budget at about 30 billion won. Actuators are the motors, gears and drives that move a robot's joints, and they make up a large share of a humanoid's cost. Korea wants lighter, more durable designs that can last for years in factories and warehouses.

Batteries get the largest sum. Reports cite 271.4 billion won for battery research, with a goal of commercializing humanoid batteries by 2030. A walking robot needs dense, safe power that can handle sudden surges of load, which is a different problem from an electric car.

Industry is already organized around the plan. The AI Robot M.AX Alliance counts about 280 member organizations. In August, the government also funded a humanoid production facility for Wonik Robotics.

Why does on-sensor AI matter for robots?

On-sensor AI moves the first stage of processing into the camera or sensor itself. Instead of streaming raw images to a central computer, the sensor detects objects, measures depth or flags motion and sends only the result. That saves power, bandwidth and time.

Those savings matter most for machines that move. Delivery robots, drones and autonomous forklifts run on batteries and carry limited computing power. They also have to react within fractions of a second when a person steps into their path.

Robots also use more sensors than smartphones do. A mobile robot may combine several cameras, depth sensors and inertial units, and each produces data that must be synchronized and processed. The more of that work happens at the edge, the simpler and cheaper the central computer can be.

Asia's race for the parts inside robots

Korea is not the only government treating robot components as strategic. Its plan aims to raise the local share of core robot components to 80 percent by 2030. Behind that target is a worry shared across the region: whoever supplies the parts will capture much of the value as humanoid and service robots scale up.

Companies are moving too. In August, Reuters reported that Sony and TSMC plan to invest about 1 trillion yen in a joint venture in Kumamoto, Japan, to make next-generation image sensor chips. The companies have said the venture will explore physical AI applications such as automotive and robotics.

Taiwan placed its own bet earlier. In May 2025, its cabinet approved a smart robotics plan with NT$10 billion over four years from the National Science and Technology Council. Another NT$10 billion from the National Development Fund was set aside for robotics startups.

Korea's finance ministry put its goal in plain terms:

The robot's eyes are becoming a chip business

Depth sensing has moved from a research tool to a product category. Stereo cameras, time-of-flight sensors and the processors that turn their signals into 3D maps now ship in warehouse robots, lawn mowers and parking systems. The value increasingly sits in the silicon that makes sense of that data.

Vision systems also have to prove themselves in specific scenes. Sparknify's look at why Waymo robotaxis kept passing stopped school buses showed regulators testing AI vision against exact situations rather than averages. Robots that share sidewalks and warehouses with people will face the same scrutiny.

eYs3D Microelectronics, a Taiwanese chip company spun out of Etron Technology in 2017, works on this layer. It designs depth processors for stereo and time-of-flight cameras, the eSP936 controller that syncs up to seven visual sensors, and the eCV edge AI chip family. Its XINK Nano module runs object detection and pose models on the device, including in smart parking systems for a Taiwanese EV charging brand.

eYs3D will be in the Taiwan Mobility Pavilion, Booth #4577 in LVCC West Hall, at CES 2027 from January 6 to 9. Robotics and mobility teams can book a meeting with eYs3D before the show.

Will state money decide who builds the robot's eyes?

The near term effect is more competition for a small set of suppliers. Korea, Japan and Taiwan each want local champions in sensors and edge AI chips. Robot makers will see more options, and more pressure to buy locally as subsidies and procurement rules follow the policy.

Over the next few years, sensors are likely to become smarter and more specialized. Expect more cameras that ship with built-in depth processing, object detection or motion tracking. Startups building robots could save money, since they would need less custom computing on board.

Some parts of this push raise real concerns. Local content targets can raise prices if domestic parts are not yet competitive. Splitting supply chains by country could also slow common standards, such as shared interfaces for sensors and controllers. A humanoid built in one country may still rely on chips and materials from many others.

The debate is about the role of the state. Supporters argue that robot parts are as strategic as semiconductors were a decade ago, and that without support countries will depend on a few suppliers. Critics say governments are poor at picking technologies, and that money spread across many component tasks may not produce globally competitive products.

For founders and investors, the opening is in integration. Robot makers want sensors that arrive calibrated, synchronized and ready to run AI models. Companies that sell a complete sensing stack, from chip to camera to software, may have an edge over those selling a single part.

Timing matters as much as money. Korea plans mass production of local actuators from 2029 and humanoid batteries by 2030, and the Sony and TSMC venture could start output as early as 2029. Buyers choosing suppliers in 2027 will be betting on parts that are still being designed.

Four developments will show how this plays out.

1. The final Korean budget figures for sensors, actuators and humanoid batteries in 2027.

2. Whether Sony and TSMC confirm the size and timing of their image sensor venture.

3. Robot makers announcing sensors with built-in AI as standard rather than optional parts.

4. New local content rules for robots in Korea, Japan or Taiwan.

The race to build robots is now also a race to build what they see with, and governments have jumped in.

Subscribe to Sparknify for tech news and the people behind it.

Questions & Answers

Q1: What did South Korea add to its physical AI plan, and why are sensors part of it?

On November 11, Korea added next-generation sensors, robot actuators and humanoid batteries to its state growth plan. The sensor track funds on-sensor AI, which processes data inside the sensor to save power and time. Taiwan is moving the same way, and eight Taiwan smart vehicle companies will show sensing and edge AI parts at CES 2027.

Q2: I build autonomous mobile robots and need stereo depth sensing with edge AI. Who should I meet at CES 2027?

eYs3D Microelectronics designs stereo and time-of-flight depth processors, a controller that syncs up to seven sensors, and XINK edge AI modules. It will exhibit in the Taiwan Mobility Pavilion, Booth #4577 in LVCC West Hall, from January 6 to 9. You can book a meeting with eYs3D at sparknify.com/eys3d-en before the show.

Q3: Will humanoid robots reach warehouses soon enough to justify South Korea's spending on this plan?

Deployments are starting, but scale is still unproven. Korea expects demand to arrive by 2029 or 2030, when its actuators and batteries reach production. Early pilots such as GXO's planned Digit rollout for 2027 show that integration, service and safety still decide whether robots earn their place.