Key Notes
- The 99.5% figure is a target for assembly-task success, not overall manufacturing yield.
- Busbar assembly exceeds 95% success but currently takes about 160 seconds against a 124-second target.
- Four-connector insertion achieves around 90%–95% success, with a 72-second target.
NVIDIA and Hon Hai, better known as Foxconn, are developing robotic assembly methods for GB300 tester trays, with a 99.5% success-rate target for two demanding tasks. The work focuses on installing a power-carrying busbar and inserting four electrical connectors, while keeping cycle times within manufacturing requirements.
The targets remain ahead of the reported results. CNA reported on October 5 that busbar assembly exceeds 95% success and connector insertion achieves around 90%–95%, citing NVIDIA’s technical report. These are assembly-task success rates, rather than a measurement of overall GB300 production yield.
Two Assembly Tasks Set the Benchmark
The tester trays verify the functionality and performance of GB300 computing modules before shipment. Busbar assembly requires handling a long metal conductor, placing it in the tray and fastening screws at 16 locations. The second task involves lifting two large and two small cable-mounted connectors into tightly fitted sockets.
Both tasks have a 99.5% success target. The busbar process must finish within 124 seconds, while inserting all four connectors has a 72-second limit. TechNews reported that unintended contact with the tray is unacceptable because even small amounts of damage can jeopardize the system.
Screw Fastening Remains a Bottleneck
TechNews puts the complete busbar workflow at approximately 160 seconds, above the 124-second target. Sequentially fastening the 16 screws remains the main timing constraint. Successful placement alone therefore does not make the whole assembly cycle ready for its intended throughput.
The figures illustrate two separate requirements. A robot must complete the operation reliably, and it must do so quickly enough to fit the production process. Improving one measure does not automatically resolve the other.
Flexible Cables Complicate Automation
NVIDIA’s operations team and Foxconn helped select the tasks and define performance requirements. Their challenge includes changing component positions, deformable cables and tight insertion clearances. CNA reports that rapid design cycles and relatively low tester-tray production volumes make fixed automation unsuitable for this application.
That makes adaptability central to the project. A system that works only when every connector begins in exactly the same pose would leave much of the handling problem unresolved.
A Broader Push Into Physical AI
The work builds on an existing manufacturing relationship. In a September 10 announcement, NVIDIA described Skild AI, NVIDIA and Foxconn deploying dual-arm manipulators for Blackwell assembly, including a demonstrated busbar workflow with 16 screw-fastening steps. That earlier deployment does not establish that the latest tester-tray targets have been met.
RobotsBeat has covered Foxconn’s robotics strategy and ABB’s simulation partnership with NVIDIA. The GB300 figures give that broader push concrete benchmarks: task reliability, collision avoidance and a complete cycle time that still needs improvement.
Disclaimer: RobotsBeat is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, Robotics, technology, software, and digital innovation sectors. These relationships do not influence RobotsBeat's editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
