Test report DSG-1461 · Rev D · tested November 16, 2026
Foundries & ManufacturingDevice under test
Daim Technology Ran 1,000 Wafer Robots on One AI System, Gained 20% Output
Daim Technology CEO Jang Young-jae told Semicon Europa 2026 that one AI system ran 1,000 wafer carriers and lifted robot productivity over 20% — software only, no hardware changes.
- Read
- 4 min
- Words
- 708
- Node
- 3nm
- Operator
- Grace Kim
Spec summary
- A proof-of-concept with a global chipmaker ran 1,000 overhead hoist transports on a single AI system, raising robot productivity over 20% without hardware changes.
- CEO Jang Young-jae presented the fab-as-one-robot concept at Semicon Europa 2026 in Munich on the 12nd; Daim Technology announced it on the 16th.
- Daim unveiled Kairobus, comprising xSIM simulation, xMS AI control and xDT digital twin, commercializing KAIST's KAIROS37 dark factory platform.
- Daim plans to develop robot charging, communication and layout standards with SEMI and industry partners.

A single AI system took control of 1,000 overhead hoist transports (ceiling-rail wafer carriers) at a global chipmaker's fab and lifted robot productivity by more than 20% — through software alone, with no hardware changes. That proof-of-concept result anchors the case Daim Technology CEO Jang Young-jae laid out at Semicon Europa 2026, the chip industry exhibition held in Munich, Germany, on the 12nd of this month, the company said on the 16th.
His core argument: fabs need a redesign around AI and robots, not incremental automation of individual tools.
Where today's fabs fall short
AI already works inside chip plants. Jang noted that manufacturers deploy separate AI systems for quality control, equipment maintenance and production operations. Each runs in its own silo. That limits any attempt to optimize the plant as a whole.
The transport layer has the same problem. Overhead hoist transports and autonomous mobile robots (self-navigating factory robots) come from different manufacturers with different control and data systems. Integrating them is difficult. Robots can deadlock — block one another and halt — because nothing above the individual machine coordinates them.
What does 'the fab as one giant robot' mean?
Jang's proposed answer is to run an entire semiconductor fab as a single, giant robot. The plant connects into one intelligent system that goes beyond individual robots and equipment. AI makes comprehensive judgments covering production, logistics and equipment operations.
Before AI decisions touch the real line, a digital twin — a virtual replica of the actual plant — verifies them in simulation. The combination de-risks autonomy: nothing reaches the floor untested.
Jang cited KAIROS37 as a working example. The dark factory platform (a fully automated, human-free production facility), developed by the Korea Advanced Institute of Science and Technology (KAIST), links three layers:
- AI decision-making
- digital twin simulation
- integrated control of robots and equipment from different makers
Kairobus: the commercial package
At the Munich event, Jang unveiled Kairobus, Daim Technology's commercial platform that extends the KAIST research into industrial settings. It has three components:
- xSIM — an ultra-high-speed simulation tool
- xMS — an integrated AI control system
- xDT — a digital twin
The platform connects plant design with real-time operations and continuous optimization, so the fab model built at design stage keeps learning from live production.
The proof-of-concept project with the unnamed global chipmaker tested the claim at scale. Daim Technology controlled 1,000 overhead hoist transports through a single AI system and resolved the deadlocks that plague multi-vendor robot fleets. The productivity gain — more than 20% — came from software alone.
Standards must change too
Software won't be enough for next-generation fabs built around AI and robots, Jang argued. Design criteria at manufacturing sites and industry standards have to change as well. His list of requirements:
- a common charging interface for robots
- communication and data standards
- robot-friendly plant layouts
Daim Technology plans to develop these standards together with SEMI, the global semiconductor equipment and materials industry association, and with the wider industry. That work targets the physical and protocol groundwork — charging, connectivity, floor plans — that current fabs simply don't provide.
What does this change for fab operators?
The near-term takeaway is measurable: a 20%-plus robot productivity gain without capital spending on new hardware. The longer-term bet is architectural. Fabs designed from the ground up for integrated AI control — rather than retrofitted silos — would treat transport, equipment and production planning as one optimization problem.
"The competitiveness of future chip fabs depends on how intelligently the entire plant is connected and operated, not just on the level of automation of individual equipment or robots," Jang said.
He added: "Based on the research accumulated at KAIST and our proof-of-concept experience at industrial sites, Daim Technology will expand the field application of autonomous operation technology based on physical AI."
Physical AI here means AI systems that sense and act on the real world — robots, carriers, tools — rather than only analyzing data. For Daim Technology, the Munich lecture doubles as a commercial debut: KAIROS37 proved the concept in the lab, Kairobus takes it to paying fabs, and the SEMI standards push aims to make sure the next plants are built to receive it.
via en.sedaily.com (Source)
More from Grace Kim
Show full bio
Market editor covering regulation, policy and their impact on hardware layer of artificial intelligence at Die Signal.
665 articles
Same lot · LOT-362C
- DSG-163110nmSamsung, SK hynix and Micron back robot charging standard for fabs
- DSG-250620nmHygon pushes into physical AI with chips for robots and factories
- DSG-192014nmGenerative Bionics to build AI data factory for humanoid robots in Italy
- DSG-618414nmKorea Puts 49.8 Billion Won Behind Robot Chips, Targets 2030 Batteries
- DSG-628520nmHyundai to Field-Test Atlas Humanoid Robot at US Chip Plant