Most AI data halls will operate with a combination of air and liquid cooling, but simply deploying both technologies in the same facility does not create an effective hybrid system. The challenge is understanding what the infrastructure actually needs to support, how much heat each system is expected to manage, and how that balance changes across different workloads, rack densities and generations of compute. Growing adoption of liquid cooling is making the interface between air and liquid increasingly important to maintaining resilience, avoiding unnecessary capacity and ensuring cooling infrastructure is aligned with the demands placed on the facility. This session will explore:
- Why hybrid cooling should start with the workload, not the technology
- The relationship between liquid heat capture and remaining air-cooling requirements
- What changing rack densities and compute generations mean for the air–liquid balance
- Where coordinated design and controls can prevent overbuilding and stranded capacity
