The Emerging Consensus on AI Power Architecture at Datacloud 2026

During the Datacloud Global Congress 2026, held in Paris, leading industry representatives from Ampace, Eaton, Siemens, Riello UPS, and Ramboll gathered to discuss the critical evolution of power architectures to support the next generation of high-density AI computing.

The panel, titled “What is Better for Future High-Density Computing Needs: AC or DC?”, served as a platform for experts across the digital and energy infrastructure ecosystem. Discussions revolved around how AI workloads challenge conventional power systems, accelerating the need for adaptable, scalable, and interoperable infrastructures.

As rack densities continue to rise and workload volatility increases, panelists emphasized the importance of closer collaboration among large-scale cloud service providers, technology vendors, and the broader data center ecosystem. While high-voltage direct current (HVDC) architectures are gaining popularity, participants agreed that practical implementation, operational stability, and compatibility with existing infrastructure remain critical considerations.

A recurring theme throughout the discussion was the evolving role of batteries in AI power systems. Energy storage is being increasingly recognized not just as a backup power source but as an active infrastructure layer that enables load stabilization, grid coordination, and dynamic energy management. During the session, James Li, General Manager of UPS, Data Centers, and Telecommunications at Ampace, pointed out that AI workloads introduce energy consumption profiles that fundamentally differ from traditional data center environments. He stressed the need for battery systems capable of responding to high-frequency load fluctuations while maintaining safety, reliability, and backup capabilities.

This perspective was further explored during Datacloud by Aaron Schott, UPS Sales Manager at Ampace. In his presentation titled “Smoothing the AI Power Curve,” Schott highlighted how advanced lithium battery systems can help absorb power fluctuations driven by AI at the UPS layer, thereby supporting both infrastructure stability and future-ready AI implementation.

These discussions reflect broader conversations within the sector, set to continue at major infrastructure events in 2026, including topics covered by Ampace and Eaton at Data Center World Washington on load optimization via AI, battery intelligence, and the evolving role of energy storage in UPS architectures.

The relevance of these discussions is particularly pronounced in Europe, where limitations in the power grid, rising energy costs, and the requirements of the Energy Efficiency Directive (EED) are increasing demand for secure, efficient energy solutions that are compatible with existing infrastructure ecosystems. As these sessions at Datacloud 2026 illustrated, the industry is actively navigating these challenges, fostering innovation and collaboration to better equip power architectures for the next generation of AI-driven applications.

In summary, Datacloud 2026 has highlighted the urgency of redefining power architecture to suit AI's unique demands. As the industry converges on a shared vision, the future of power systems for AI computing looks to be dynamic, collaborative, and forward-thinking, indicative of the broader technological shifts occurring in the global landscape today.