Trane Technologies and Eaton have jointly released an integrated reference design for AI data center thermal and electrical architecture, claiming 15% energy efficiency gains and 80% copper reduction versus traditional low-voltage designs. The Trane Continuum Rubin DSX and Eaton Beam Rubin DSX platforms are built to align with NVIDIA's DSX blueprint for AI data centers, targeting compatibility across cooling and power subsystems. Both vendors committed to evolving the spec as liquid cooling and direct current (DC) power architectures become standard.
The move reflects a structural shift in 2026: power and cooling are now harder constraints than compute. AI data centers require 100–750 MW per site, with individual GPU racks (like NVIDIA Blackwell GB200 NVL72) pulling 120–130 kW, approaching or exceeding 370 kW per rack in next-generation designs. Hyperscalers have collectively shifted $450 billion of their $600B+ H1 2026 capex into AI infrastructure, and grid interconnection wait times stretch past 5 years in many US regions. Liquid-cooled AI servers jumped from 15% of deployments in 2024 to 54% in 2025, expected to reach 76% by 2026.
For infrastructure architects, this signals that integrated power–cooling design is now a prerequisite for competitive deployments, not an afterthought. The constraint set has moved from 'can we get H100s?' to 'do we have energized capacity and the thermal architecture to use them?' Copper availability and grid capacity, not GPU allocation, are the real bottlenecks. The Trane–Eaton partnership and NVIDIA alignment point toward vendor-standardized modular designs to accelerate deployment velocity in energy-constrained sites.