Bright Machines Hybrid Cell Aims to Speed Up AI Server Production

San Francisco-based manufacturer Bright Machines announced its Hybrid BRC (Bright Robotic Cell) today, an expansion of its Bright Factory platform. The system is designed to address a bottleneck in AI server production by allowing human operators to perform specific assembly steps within a sensor-monitored robotic cell, while maintaining the digital traceability of each server.

According to Sviat Dulianinov, CEO of Bright Machines, the initial yield for assembling modern AI servers can be as low as 20% when starting with manual operations. Given that a single AI server can cost hundreds of thousands of dollars, improving this figure is critical for hyperscalers awaiting infrastructure. The Hybrid BRC aims to keep human hands in the loop for necessary tasks without reintroducing human error through continuous monitoring.

The new robotic cell integrates safety features directly into the production line. When an operator opens guarded doors, the robotic arm deactivates, and on-screen instructions guide the operator. The cell's sensor array continues to monitor for incorrect installs or missed steps, ensuring the same quality checks used in full automation and preserving the serial-number level traceability.

Dulianinov stated that robotic operations typically achieve a yield of over 98% per station. This significant improvement over manual assembly is crucial for the profitability of manufacturing high-value hardware. Bright Machines believes its technology can reduce the time required for hardware deployment by at least one-third.

Bright Machines is already operating several hybrid lines in the U.S., having produced over 10,000 compute nodes with the new system. The company plans to manufacture over half a gigawatt of compute capacity this year. Customer names are being withheld due to confidentiality agreements.