THE ESSENTIALS IN 30 SECONDS
- Gallop is a two-armed, fixed-base humanoid running on one Nvidia Jetson AGX Thor computer; ASUS designs the body and a touch-sensing hand in-house.
- Its first job is packing 15 motherboard accessories at an ASUS group electronics factory, with a target of 99.5% success within 20 seconds per board.
- ASUS plans to show Gallop to customers at Computex 2027, run customer pilots in 2027 and aims for market availability in the second half of 2028.
ASUS, the Taiwanese computer maker, has presented Gallop, its first humanoid robot. In a post on the ASUS Pressroom dated 13 November, Albert Chang of the company's Physical AI Solutions Business Group explains the project, which co-CEO S.Y. Hsu introduced in an appearance on CNBC. Rather than a general-purpose machine, Gallop is being trained for one specific, unglamorous job in an electronics factory.
A single station, a human benchmark
Gallop's first assignment is a motherboard accessory packing station at an electronics manufacturing services (EMS) factory belonging to the ASUS group. The task involves placing 15 small items with each board, many of them thin parts sealed in soft plastic bags that simple grippers struggle with. ASUS says the station is a real workstation built for people, not a staged demonstration.
The company has set the robot the same bar it applies to human workers: a 99.5% success rate, with each board completed within about 20 seconds. Unite.AI stresses that these figures are targets, and ASUS has not reported results achieved so far.
In its current form, Gallop has two arms mounted on a fixed base. All computing runs on a single Nvidia Jetson AGX Thor module inside the robot. ASUS designs the body and a dexterous hand with touch sensing itself, while motors and other standard components come from partners it does not name. According to Unite.AI, the in-house hand has not yet been fitted to the robot.
How it learns
ASUS starts from openly available AI models rather than building a large model of its own. Experienced workers wear head-mounted cameras, with their consent, to record how they do the job; simulation then multiplies those examples, and an operator using a VR headset and gloves guides the robot through practice. Data recorded once the robot is working feeds back into training.
Timeline and open questions
The programme began in early 2026. Over the next year ASUS plans to add its own hand and a wheeled version, show Gallop to customers at Computex 2027 and start customer pilots the same year, with market availability targeted for the second half of 2028. Electronics manufacturing comes first, with hospitals and service sites mentioned as later options.
ASUS presents the robot as a way to staff repetitive, physically demanding stations, saying workers who train it could move into setup and supervision roles. Price, specifications such as payload and speed, and evidence that the robot meets its targets have not been published.
