Summary
- Schneider Electric and Hon Hai Technology Group will co-develop reference architectures for next generation AI data centres.
- The collaboration combines Foxconn’s AI rack integration and manufacturing scale with Schneider’s power, cooling, and energy management systems.
- The work reflects the move towards more repeatable AI data centre delivery, although grid access and planning remain separate constraints.
Schneider Electric and Hon Hai Technology Group, better known as Foxconn, are collaborating on infrastructure systems for next generation AI data centres.
The companies will co-develop reference architectures that combine Foxconn’s compute platforms, AI rack integration, and manufacturing capacity with Schneider’s power systems, cooling, and energy management portfolio.
Production under the collaboration is expected to begin later in 2026, with the work also covering closed loop energy optimisation, modular power and cooling skids, and standardised design frameworks.
AI infrastructure has pulled IT manufacturing and facility design closer together, because GPU racks, high current distribution, liquid cooling, controls, energy metering, and commissioning have to be planned as connected systems.
Foxconn gives the partnership a manufacturing and systems integration base, while Schneider brings the electrical and thermal infrastructure that surrounds dense IT loads.
Rather than leaving servers, racks, power trains, coolant loops, and plant rooms to be reconciled late in a project, the collaboration is intended to produce repeatable designs that can be adapted across sites.
Repeatable design still meets local limits
Standardised architectures can reduce engineering uncertainty, shorten design cycles, and make procurement easier for customers trying to deploy similar AI systems across multiple locations.
Even with a more standard technical package, AI capacity still needs a viable site, planning consent, a grid connection, substation works, service access, and local compliance approval.
Those physical constraints are often the hardest part of European data centre development, especially in markets where grid capacity is limited and planning scrutiny is rising.
The likely value of the Schneider and Foxconn work sits in reducing interface risk once a site can be built, because rack level design, electrical distribution, cooling, and controls can be specified together.
Thermal design will be central to that process, as AI clusters can concentrate heat in ways that conventional air cooled data halls were not built to manage.
Direct to chip cooling, rear door heat exchangers, coolant distribution units, and heat rejection systems need to be planned as a chain, rather than separate packages assembled around the IT load.
Electrical design faces similar pressure, because high density racks demand more robust distribution, protection, monitoring, and resilience planning across both the white space and the plant room.
Schneider’s European base gives the collaboration a direct regional link, even though Foxconn’s manufacturing reach and customer base are global.
If the partners can turn reference designs into available systems, the effect will be felt in the procurement and engineering packages behind future AI data centre campuses.
