Who makes sensors for AI data centers

Sensors are where the control system gets its picture of the building. Temperature and humidity in the aisles and at the equipment, differential pressure across floors and filters, flow in the water loops, and — in liquid-cooled halls — leak detection under and around the cabinets. Every control decision above them is only as good as this data. Calibration drift is the practical problem and it is chronic. A humidity sensor that has drifted a few percent will have the plant holding the wrong setpoint continuously, at a cost nobody sees because the system reports that everything is fine. The same is true of a pressure sensor that has gone out of range; the control loop keeps working, just toward the wrong target. Sensors are cheap and recalibration is a labour cost, which is exactly the combination that makes it get deferred. Liquid cooling has added a category that did not use to matter much: leak detection at the rack. A small leak found early is a hose replacement, and found late it is hardware. What to look for is measurement accuracy and drift specification rather than headline range, whether the sensor can be calibrated in place, and how the site handles verification — the optimisation software elsewhere in this part is only as trustworthy as the instrumentation under it.