A.I.M.I. Takes Autonomous Control of Cooling at TenHats,
a 2 MW Mission-Critical Data Center
Live Production Deployment
2026
For four months, FLUIX's A.I.M.I. platform autonomously controlled cooling at TenHats, a live, mission-critical 2 MW data center in Knoxville, TN, inside operator-set guardrails. It improved PUE from 1.36 to 1.27 (6.6%), held zero SLA excursions, required zero new sensors, and cut cooling response lag from roughly 15 minutes to five.
Independent analysis by TRC put modeled HVAC savings at 0.4 MWh/day, with a projected path to up to ~130 kW of potential power headroom at full 2 MW load. Following the deployment, TenHats contracted with FLUIX to keep A.I.M.I. in continuous autonomous control. The work was supported by TVA EnergyRight.
Data Center Lab
2025
Data centers have traditionally relied on static, rule-based controls for cooling, power, and IT operations. As AI workloads drive rapid and unpredictable changes in compute demand, these systems struggle to respond in time, leading to inefficiencies, peak events, and operational risk.
A.I.M.I. 1.0 is FLUIX’s reinforcement learning system for autonomous data center cooling. Trained in a live lab environment, the model continuously learns from real-time thermal, power, and operational signals to predict load changes and adjust cooling setpoints before inefficiencies occur. Our results demonstrate how closed-loop Physical AI can reduce energy waste, improve system responsiveness, and enable more resilient, grid-aware data center operations.
Server Room
2024
FLUIX's A.I.M.I. or “AIMI”, an Autonomous AI, significantly reduced the air conditioning energy in Solideon’s server room, a leader in automated aerospace manufacturing. AIMI achieved a remarkable 65% reduction in HVAC energy consumption over the system’s automatic mode. This optimization preserved optimal environmental conditions and server performance, resulting in a site-wide energy consumption reduction of roughly 40%.
Solideon, a Berkeley, CA-based innovator in automated manufacturing, provided FLUIX the opportunity to conduct this validation in a setting reflective of a standard data center's complexity and demands. The experiment's compelling outcomes prompted Solideon to extend a letter of intent to FLUIX for incorporating AIMI’s AI CoPilot into their manufacturing processes. Despite differing requirements from the server room scenario, both Solideon and FLUIX discovered that AIMI’s automated solutions could further enhance Solideon’s cleanroom operations, optimizing cleanliness and resource efficiency.