Written by: Andrew Whitmore, Vice President of Sales at Motivair by Schneider Electric
AI leadership won’t be determined by who builds the biggest AI factory first—it will be defined by who deploys compute the fastest. The organizations gaining ground today aren’t waiting years for new, purpose-built campuses. They’re standing up lower-density GPU clusters inside existing facilities, extracting new AI capacity from infrastructure already connected to power, cooling, and network ecosystems.
Key takeaways
- Speed wins. Brownfield data center modernization enables faster AI deployment.
- Liquid cooling is critical. It unlocks higher rack densities within existing infrastructure constraints.
- Not a fit for every facility. Brownfield data center modernization depends on available infrastructure.
- Phased upgrades reduce risk. Modular, rack-by-rack execution supports hybrid air/liquid transitions.
- Assess deployment models for liquid cooling. Power, cooling, and monitoring systems must work together.
Why brownfield deployments accelerate AI growth
While purpose-built AI factories—newly built, AI-optimized data centers—dominate headlines, brownfield deployments offer speed to revenue, lower capital exposure, and the ability to scale incrementally—all critical advantages in a market where silicon cycles move faster than construction timelines.
Most data center providers already have infrastructure located near data, users, and networks, making AI data center retrofits more desirable than starting from scratch. This allows them to meet fast-growing AI demand more quickly.
By accelerating speed to market, brownfield projects provide a competitive edge. This approach eliminates challenges associated with new data center builds such as grid interconnection delays, limited land availability, and asset monetization. AI deployment in existing data centers provides an easier growth path, not merely a workaround, giving operators a faster time to revenue.
Explore AI-ready liquid cooling resources
Infrastructure constraints in existing data centers
Modernizing data centers requires overcoming constraints that can be addressed with liquid cooling for AI workloads. AI accelerates data center densification and concentrates heat and power at the rack and chip level. Still, existing facilities were designed for lower rack densities and air cooling, not high‑density racks for AI. Existing spaces have fixed-rack layouts and floor-loading limits.
Other constraints include limited chilled-water capacity or availability, power distribution systems sized for lower rack loads, low ceilings, and airflow constraints. Operators also want to avoid the risks associated with live data center upgrades, which can disrupt operations.
How liquid cooling enables AI retrofits
However, implementing liquid cooling for AI infrastructure modernization addresses these issues. Liquid cooling removes heat at the source, reducing the need for airflow redesign. It makes higher rack densities possible with some facility adjustments, as opposed to a full retrofit. Operators can deploy AI incrementally as demand grows.
This approach simplifies integration with existing mechanical systems and enables localized, rack-level heat rejection strategies. As a result, retrofitting smaller or constrained sites becomes more feasible.
Deployment models for liquid cooling in brownfield sites
Deploying AI in brownfield data centers is often possible but requires a shift from air to liquid cooling. Strategies you can consider in deploying liquid cooling in an existing build include the following:
- Localized Scaling and Heat Rejection – A switch to full liquid cooling can occur in stages throughout a facility, avoiding disruptions. Localized upgrades with direct‑to‑chip liquid cooling at the rack prevent major central plant retrofits, which is especially attractive to operators of colocation and service provider environments.
- Rear-Door Heat Exchangers – Motivair’s ChilledDoor® Rack Cooling System improves GPU thermal management by stabilizing rack temperature even under peak AI workloads. Deployable in air and air-assisted liquid cooling, it delivers an ideal AI server cooling option for data centers utilizing NVIDIA’s H100 and H200 GPUs.
- Liquid-to-Air Exchanger – For facilities without access to water, an available option is to deploy a heat dissipation unit (HDU) that removes heat from AI servers and sends the heat into the air. This approach doesn’t require routing facility water to the rack.
- Modular Execution Model – Rack-by-rack deployment enables phased scaling and allows hybrid air and liquid environments to coexist during transition periods. To reduce engineering risk, operators can leverage reference designs validated for retrofit conditions. Also worth considering are prefabricated components such as data halls and power blocks. This enables off-site construction and commissioning for accelerated deployment.
- Power, Controls, and Visibility Integration – Successful brownfield modernization requires syncing all relevant components – power distribution (cables, busways), monitoring and visibility tools such as DCIM monitoring for high‑density racks, as well as cooling system controls. Power and cooling must evolve together to support higher densities safely and predictably.
- Operational Continuity – Phased upgrades minimize downtime and protect existing workloads, enabling operators to execute service strategies support reliability during the transition.
When partial upgrades make sense
Not every legacy facility is suited for high-density liquid cooling retrofits. Brownfield data center modernization depends on available infrastructure capacity and economic viability. Where power, cooling, or facility constraints are too limiting, partial upgrades—or new builds—may be required. The key is evaluating the existing performance envelope before determining how much AI density it can support.
Strategic decision framework
Liquid cooling retrofit for AI workloads in brownfield facilities allows data center operators and service providers to deploy AI capabilities faster than through greenfield projects. This translates to a boost in revenue per square foot in existing data halls, lower capital investment, and reduced operational risk compared to wholesale infrastructure changes. This allows them to maintain a competitive edge. Brownfield data center modernization also provides greater adaptability for future platforms and workloads. It’s an intelligent approach to modernization, enabling operators to turn existing data centers from underleveraged assets into growth drivers. Organizations that modernize their brownfield spaces fastest will capture AI growth sooner and more efficiently. For additional guidance, access the report AI Workload Deployment in Data Centers: Retrofit, Outsource or Build New? A strategic framework for decision-making.
About the author
Andrew Whitmore, Vice President of Sales at Motivair by Schneider Electric
Sales executive focused on helping customers address complex data center cooling and infrastructure challenges through a consultative, solutions-driven approach. Known for building trusted client relationships and expanding business by aligning advanced cooling solutions with customer performance, efficiency, and scalability goals.
Add a comment