Summary
- InfraPartners and Phaidra are integrating automated facility controls with prefabricated AI data-centre infrastructure.
- The system is intended to optimise cooling and available electrical capacity in real time.
- Automation and prefabrication are being used to address commissioning, labour, and operating complexity as facilities become larger and denser.
InfraPartners and Phaidra are integrating automated facility-control systems into prefabricated AI data-centre infrastructure, combining offsite construction with real-time optimisation of cooling and power use.
InfraPartners develops prefabricated data-centre systems for high-density AI workloads, while Phaidra develops AI agents intended to monitor and optimise critical facility infrastructure.
The partnership is intended to embed Phaidra’s operating technology into InfraPartners’ prefabricated deployments rather than adding controls only after a facility has been commissioned.
Phaidra describes its systems as managing mission-critical infrastructure in real time, including cooling optimisation, thermal monitoring, and identification of operational issues. Its stated objective is to increase the amount of electrical capacity available for productive compute while reducing thermal excursions.
That objective is becoming more difficult as rack density rises. Conventional data-centre controls were designed around facilities where thermal loads were comparatively predictable and operators could respond to alarms through established building-management systems.
High-density GPU environments change that operating profile. Loads can vary quickly, cooling systems may incorporate direct liquid cooling alongside conventional plant, and a large campus can contain thousands of individually monitored components.
Prefabrication addresses a different part of the same scaling problem. Moving repeatable mechanical and electrical assemblies into controlled manufacturing environments can reduce the amount of work carried out on site, standardise quality, and shorten construction and commissioning programmes.
InfraPartners has previously described its model as using standardised designs and offsite manufacturing to reduce schedule risk. The company has also worked with JLL on an end-to-end deployment model covering site selection, project delivery, commissioning, and facilities management.
Integrating operating intelligence earlier in that process could make controls part of the repeatable factory-built design rather than a bespoke layer commissioned separately at every project.
The relevance for Europe is the pressure on skilled engineering labour. Large data-centre pipelines are increasing demand for electricians, controls engineers, commissioning specialists, cooling technicians, and facility operators at the same time that project schedules are being compressed.
Automation cannot remove the need for those roles. Physical equipment still requires installation, testing, maintenance, isolation, and repair by trained personnel. It can, however, reduce the volume of routine monitoring and help operators identify deviations across larger estates.
The more difficult question is how much operational authority should be delegated to automated systems in mission-critical environments. Cooling and power optimisation can improve capacity utilisation, but control systems need defined boundaries, fail-safe behaviour, cybersecurity protection, and a clear route back to manual operation.
That becomes particularly important where software can change plant operation automatically. An optimisation error is no longer confined to an inaccurate dashboard; it could affect pumps, valves, fans, temperatures, or the amount of reserve available across electrical systems.
The InfraPartners-Phaidra model therefore combines two attempts to industrialise AI data-centre delivery: manufacturing more of the facility away from the construction site and automating more of its behaviour once operating. The effectiveness of the approach will depend on whether those efficiencies can be achieved without reducing the transparency and resilience expected from critical infrastructure.
