Meta Tests Robots for Data Center Maintenance in Altoona and New Albany, Targeting AI Infrastructure Support

Meta has initiated testing of robotic systems for automated data center maintenance, a strategic move aimed at supporting its expanding AI infrastructure. These tests are underway at Meta's data centers in Altoona, Iowa, and New Albany, Ohio, involving robots from Watney Robotics, Kinova, and ABB. The initiative seeks to automate routine tasks such as cable swapping, server restarts, and hardware reseating, which are critical for maintaining the operational efficiency of large-scale data centers. For broader context, explore our Top 100 AI Tools.

Robots Deployed for Core Maintenance Tasks

Since June 2025, two dual-armed robots from Watney Robotics have been actively engaged in cabling tasks at Meta's Altoona facility. This deployment is part of a broader effort to assess the feasibility and impact of robotic automation on data center operations. A Meta worker has indicated that a successful cable-swapping system could potentially reduce the workload in certain roles by up to 80%, highlighting the significant efficiency gains Meta aims to achieve.

In addition to cabling, Meta is evaluating other specialized robotic applications. A Kinova Gen3 robotic arm is being tested for server power-cycling, a task that involves remotely restarting servers. Concurrently, ABB machines are being utilized at the New Albany data center for reseating components, ensuring hardware is correctly installed and connected.

Long-Term Vision for Data Center Robotics

Eric Xu, a senior robotics manager at Meta, has outlined the company's long-term objectives for data center robotics. These goals extend beyond current testing to include incident response, environmental monitoring, and preventive maintenance. This comprehensive vision suggests a future where robots play a more integrated role in the autonomous management of data center environments.

Meta has previously demonstrated autonomous machines designed for moving server racks and tracking inventory. These earlier initiatives underscore a consistent commitment to exploring automation within its infrastructure. The current tests with Watney, Kinova, and ABB robots represent the next phase in this ongoing strategy to enhance operational efficiency and scalability.

Current Limitations and Challenges

Despite the advancements, Meta acknowledges several limitations with its current robotic deployments. Robots are presently slower than human technicians and require close supervision to ensure tasks are completed accurately and safely. Furthermore, Meta's inventory robots have encountered difficulties navigating corners and managing cables, indicating areas for further development.

A notable challenge highlighted by Meta is the limitation of its inventory robots' grayscale cameras, which are unable to differentiate between green and red equipment indicators. This visual constraint impacts their ability to interpret status signals from hardware. Additionally, Meta has confirmed that the current generation of robots is not equipped to handle the complex and intensive cabling requirements of advanced systems like the Nvidia GB300.

Conclusion

Meta's ongoing tests with robots from Watney Robotics, Kinova, and ABB at its Altoona and New Albany data centers mark a significant step towards automating critical maintenance tasks. While the potential for increased efficiency, such as an 80% reduction in workload for some roles, is substantial, current robotic systems still face challenges in speed, supervision requirements, navigation, and visual recognition. As Meta continues to refine these technologies, the integration of robotics into data center operations is expected to evolve, moving closer to the long-term goals of comprehensive autonomous management for its growing AI infrastructure.

Sources

• Inside Meta's Push to Put Robots to Work in Data Centers
• Inside Facebook's New Robotics Lab, Where AI and...
• Mark Zuckerberg says Meta is building a 5GW AI data center | TechCrunch
• These Industrial Robots Get More Adept With Every Task
• Inside Meta’s push to put robots to work in data centers