Trane, Eaton partner on integrated AI data center reference design
Trane Technologies and Eaton have partnered to create an integrated reference design for AI data centers based on NVIDIA's DSX platform. The collaboration aims to address rising power density demands by improving energy efficiency by up to 15%, lowering installation costs by up to 30%, and reducing copper usage by up to 80%. This unified approach replaces siloed design processes, aligning with projections that AI will drive 70% of global data center capacity growth by 2030.

*this image is generated using AI for illustrative purposes only.
Trane Technologies (NYSE: TT) and Eaton (NYSE: ETN) announced a strategic collaboration to launch an industry-first reference design for next-generation AI data centers. The partnership integrates advanced thermal management and intelligent power system architectures into a unified framework, designed to accelerate deployment and enhance operational efficiency for high-performance computing environments.
The new reference design addresses the increasing power density demands of AI factories by replacing traditional siloed design processes with coordinated systems. According to the companies, this approach can deliver combined energy efficiency gains of up to 15% and reduce installation costs by up to 30% compared to conventional low-voltage designs. Additionally, the integrated architecture aims to cut copper usage by as much as 80%.
Technical Integration and Market Context
The collaboration aligns with the NVIDIA DSX platforms, specifically incorporating the Trane Continuum Rubin DSX and Eaton Beam Rubin DSX solutions. The integrated design is built for the widely adopted NVIDIA DSX AI Factory Reference Design. Eaton’s technology provides power distribution for the Trane platform, enabling dynamic exchange of leading indicators between power and cooling systems. This coordination allows the infrastructure to respond more effectively to real-time operational needs.
The announcement comes as global data center capacity is projected to almost triple by 2030, with AI driving approximately 70% of that growth. The reference design is built to work with the NVIDIA Omniverse DSX Blueprint, offering a predictable method for planning electrical, thermal, and digital control infrastructure.
Key Efficiency Metrics
| Metric | Improvement vs Conventional Designs |
|---|---|
| Energy Efficiency | Up to 15% gain |
| Installation Costs | Up to 30% reduction |
| Copper Usage | Up to 80% reduction |
Mauro J. Atalla, Senior Vice President and Chief Technology and Sustainability Officer at Trane Technologies, stated that the combined design helps customers accelerate deployment and plan for future scaling. Michael Regelski, Senior Vice President and Chief Technology Officer of Eaton’s Electrical Sector, noted that advancing reference designs into unified systems helps progress the industry standard for deployment speed.
What the Numbers Show
The disclosed efficiency metrics highlight a significant shift in capital expenditure structure for data center operators. While energy efficiency improves by 15%, the reduction in copper usage (80%) and installation costs (30%) suggests that the primary value proposition lies in upfront capital optimization rather than just operational savings. This divergence indicates that the integrated medium-voltage design primarily targets the high material and labor costs associated with traditional low-voltage copper-intensive infrastructure.
How might the 80% reduction in copper usage impact global copper supply chains and pricing volatility for data center construction?
Will other major infrastructure providers like Schneider Electric or Vertiv develop competing integrated reference designs to challenge the Trane-Eaton-NVIDIA standard?
What are the potential cybersecurity risks associated with the dynamic exchange of leading indicators between power and cooling systems in real-time?






























