AI is changing how data centers are planned, focusing more on where compute can realistically be deployed than on how much can be added. As data center power density grows, cooling, electrical infrastructure, and utility access have become one interconnected constraint. Higher-density AI racks demand a coordinated strategy, balancing available grid power, facility design, deployment timelines, and long-term operational efficiency.
For data center architects, facilities leaders, and AI infrastructure teams, the planning question is “How do we maximize AI capacity within the power we can actually obtain?”
Liquid cooling is increasingly the answer because it helps operators extract more compute from limited electrical capacity.
Data center power density is outpacing facility design
Hyperscale data centers, colocation providers, and neoclouds are moving well beyond the rack densities traditional facilities were designed to support. Average rack density increased from approximately 16 kW in 2025 to 27 kW in 2026, while only one in five operators report being prepared to support the 50–70 kW racks now common in AI deployments. Forecasts put average densities near 40 kW within a few years, and the newest AI systems can reach up to 246 kW per rack.
GPU-based AI systems drive this shift directly. NVIDIA’s Vera Rubin platform can push rack power requirements up to 246 kW. Rather than spreading workloads across hundreds of lower-density servers, cloud and AI infrastructure concentrates power into fewer racks, creating thermal and electrical demands that conventional facilities were not built to handle.
Higher power density means significantly more power must be delivered (and removed as heat) from a much smaller footprint.
Why data center power density is now a grid problem
In practice, data center power and rack density is an infrastructure planning challenge. Global data center electricity demand is projected to reach approximately 132 GW in 2026, climbing toward 290 GW by 2030, driven largely by AI-optimized servers. At the same time, utility interconnection queues have become one of the largest obstacles to new AI deployments. In some major U.S. markets, securing new power capacity can take three to four years, longer than constructing the facility itself.
Time to power is becoming as critical as time to deployment. Organizations increasingly face a familiar bottleneck: utility power can’t be delivered fast enough. Instead of designing around unlimited future power, operators need to maximize the compute they can deploy within existing electrical allocations.
Liquid cooling improves more than thermal performance
Direct-to-chip liquid cooling addresses power constraints in two ways. First, it supports rack densities air cooling can’t reach. Schneider Electric’s end-to-end liquid cooling portfolio is built for the high-density requirements of cloud data centers, colocation providers, and neoclouds.
Second, it reduces facility overhead. Legacy air-cooled facilities commonly operate at PUE values between 1.55 and 1.67, meaning roughly a third of incoming electricity supports infrastructure rather than IT equipment. Direct-to-chip liquid cooling commonly achieves PUE values around 1.10–1.20, freeing electrical capacity for compute instead of cooling. Schneider Electric’s direct-to-chip architectures deliver 30–60% cooling energy reductions in appropriate applications.
For operators constrained by fixed utility allocations, every megawatt no longer required for cooling can support additional AI workloads without waiting years for new grid capacity.
Design around the entire power path
Successful AI infrastructure planning depends on coordinating utility availability, electrical distribution, UPS infrastructure, rack power architecture, cooling distribution, and monitoring and operational visibility together.
Schneider Electric positions liquid cooling within a broader grid-to-chip and chip-to-chiller approach that treats electrical and thermal systems as one system, not two. As buyers increasingly evaluate AI infrastructure as an interconnected whole rather than standalone technologies, conversations have shifted from liquid cooling alone toward broader AI infrastructure planning: power architecture, deployment speed, modular infrastructure, and operational readiness.
Proof point: Planning around existing grid access
TeraWulf’s Lake Mariner campus in Buffalo, New York illustrates this approach at scale. Rather than developing a new greenfield site, the project repurposes a legacy industrial location with existing grid interconnection capacity. Schneider Electric, together with Motivair by Schneider Electric, is delivering integrated power and liquid cooling infrastructure supporting a phased campus expected to reach up to 750 MW. The deployment integrates Galaxy™ VX UPS systems, lithium-ion battery infrastructure, Motivair CDUs, in-rack manifolds, ChilledDoor® rear-door heat exchangers, and EcoStruxure™ IT monitoring software, proving that power and cooling engineered together shorten deployment timelines while maximizing available electrical capacity. Time to power remains one of the defining constraints on AI infrastructure growth today.
Existing facilities have viable retrofit options
Not every cloud or colocation provider is building a new hyperscale AI campus. Many need to introduce higher-density AI workloads into existing facilities designed for conventional compute, and that doesn’t require a full rebuild.
Retrofit pathways capable of supporting higher-density AI environments include direct-to-chip cooling loops, rear-door heat exchangers (RDHx), Coolant Distribution Units (CDUs), and Heat Dissipation Units (HDUs). These approaches let existing facilities accommodate higher rack densities while using much of their existing infrastructure. HDUs can also reject heat to air where chilled water infrastructure is unavailable, adding flexibility for brownfield deployments.
Retrofits still require meaningful capital and operational planning. Whether retrofitting or building new capacity delivers better long-term value depends on available power, facility condition, deployment timelines, and business priorities.
Plan AI infrastructure around power, not just cooling
As data center power density climbs, successful infrastructure planning depends on understanding how power availability, cooling strategy, and deployment timelines influence one another. Cloud data centers, colocation providers, and neoclouds that evaluate these decisions together are better positioned to scale AI efficiently, whether expanding existing facilities or developing new campuses. Explore Schneider Electric’s Liquid Cooling Hub for direct-to-chip cooling, retrofit strategies, and integrated AI-ready cooling architectures.
Add a comment