GlobalFoundries turns its Singapore fabs into a test bed for an AI-run control room

GlobalFoundries has transformed its Singapore fabrication facilities into a test bed for an AI-run control room. The company, based in New York, has established an Advanced Facilities Control Centre (AFCC) at its Singapore site, in collaboration with German industrial giant Siemens. This system aims to integrate facilities operations and data across GlobalFoundries's global manufacturing network for the first time.

Facilities in a fabrication plant refer to life-support systems such as power, cooling, ultrapure water, specialty gases, and filtered air for cleanrooms. Any failure in these systems could jeopardize the wafers inside. In the initial phase, five fab-specific control centres within GlobalFoundries's Singapore campus will function as a single command centre.

The system aggregates facilities data and operational metrics into a unified view, employing monitoring and visualization tools and AI features like predictive maintenance. This enables teams to troubleshoot faster, identify root causes sooner, and prioritize problem-solving.

Once fully deployed, the system will be capable of monitoring over one million critical data points from facilities equipment. Round-the-clock equipment health monitoring is expected to reduce unplanned downtime, while metrics can be compared across sites to enhance performance and sustainability. The centre also includes autonomous robots equipped with thermal imaging cameras, gas detectors, and vibration and acoustic sensors, which patrol the fab floor and campus in real-time, streaming data back for analysis.

Singapore was chosen as the starting point for this initiative due to its robust digital foundation, government support, and ecosystem for advanced packaging and silicon photonics. The centre exemplifies Singapore's collaborative approach to manufacturing in these niche areas. GlobalFoundries's senior vice president and general manager for Asia-Pacific manufacturing, Yew Kong Tan, highlighted that the city-state provided the ideal foundation for this project.

While the announcement emphasizes the potential one million data points the system can handle, it lacks specific details on reduced downtime, energy or emissions savings, and the cost of the centre. Moreover, the workforce implications of consolidating control rooms into one are not addressed. Fewer people may end up monitoring the screens, as the tools are intended to free facilities engineers for higher-value diagnostics and problem-solving.

Security and sustainability concerns also arise from this centralized approach. Folding a campus's critical infrastructure data and robot feeds into a single command centre simplifies operations but also presents a high-value target for potential attackers. Chip suppliers in the region face increasing pressure to demonstrate cyber resilience before gaining customer trust.

Additionally, the sustainability benefits of lower energy use and emissions are not quantified, making it difficult for GF to hold itself accountable to these promises.