On September 1, Samsung Electronics quietly stood up a new team inside the division that runs its chip business. Industry sources named it Thursday: AX/PI, short for “AI transformation” and “process innovation,” sitting under the Global Manufacturing & Infra Technology unit of the Device Solutions division. Its job is to build and deploy robots inside Samsung’s semiconductor fabs.
The team is not new people so much as reassembled ones. Many members came out of Samsung’s equipment technology research institute, which did automation and localization research on fab equipment until it was downsized and reorganized in 2024. Researchers with master’s and doctoral degrees scattered across process-specific technology groups after that. Samsung pulled a group of them back together late last year as a task force, then upgraded it to a permanent team this month.
The target is maintenance, not assembly
Equipment maintenance is the first area AX/PI is aiming at — work currently done by contractors and on-site engineers. Industry officials point to Dextro, a collaborative robot Lam Research launched in 2024 for semiconductor equipment maintenance, as the kind of machine Samsung could build: installing, securing and cleaning consumable parts with enough precision to improve process stability and yield.
The project is still early. Samsung is mapping robot travel routes, defining work zones and specifying parts storage while it collects the operational data those decisions require. Maintenance procedures and part locations differ by tool, so the work has to be standardized before any robot goes on shift.
One industry source framed the stakes plainly: fabs have fixed routes, fixed equipment and fixed working environments, which makes them easier to automate quickly than the unstructured spaces general-purpose humanoids have to operate in. The long-term question, in the same source’s words, is how much of the work currently done by on-site engineers can be transferred to robots.
The third robotics bet Samsung has made this year
AX/PI is not Samsung’s first move into robots in 2026. It is the third, and each one targets a different job category. On July 21, Samsung folded its scattered robotics teams into the Robotics eXperience business office, a ₩19 trillion (~$13B) unit built around general-purpose humanoids, run under the Device eXperience division by Dongkun Lee, who joined from Hyundai’s Boston Dynamics-linked robotics strategy team. On September 8, Samsung SDS supplied the software layer: an orchestration platform launching in 2027 that computes collision-free paths for mixed robot fleets and reassigns tasks around breakdowns, already sold into 430 manufacturing clients.
RX chases general factory labor and commercialization. SDS sells scheduling software. AX/PI, sitting inside the chip division rather than the consumer-electronics side, is scoped narrowly at fab maintenance, and its first named target isn’t line labor: it’s the specialized technical work of equipment engineers and contractors. Three different Samsung units are now building robots for three different categories of worker, in parallel, inside the same twelve months.
A higher-skill job just got a deployment timeline
Most humanoid-replaces-labor coverage this year has been about sorting, assembly and material handling: high-turnover, low-barrier work. Fab maintenance sits higher up the skill ladder: reading equipment behavior, diagnosing failures, doing precision physical work inside tool chambers. That job currently belongs to Samsung’s own on-site engineers and to outside maintenance contractors.
The industry source’s comment about fabs being easier to automate than open environments is the load-bearing point here. It means the constraint on this rollout isn’t robot capability (cobots like Dextro already do version one of this job); it’s procedure standardization, which is a project-management problem, not a research problem. Once standardization is done, deployment can move faster than most observers expect. For fab maintenance contractors and on-site engineers, AX/PI’s move from task force to permanent team is itself the signal: this isn’t a side experiment, it’s a funded production line with a named target.
Sources:
