The Data Center Frontier Show From Air to Liquid at Scale: What Hyperscalers Really Need from the Cooling Supply Chain
Jan 7, 2026
Axel Bokiba, General Manager of Data Center Cooling at Moog, shares his expertise in liquid cooling solutions for hyperscale data centers. He highlights the critical need for liquid cooling due to high heat loads and outlines how it revolutionizes data center architecture with direct-to-chip methods. Axel discusses the reliability challenges of pumps, emphasizing the expectation of 99.999% reliability for hyperscalers. He also stresses the importance of rigorous component testing and Moog's partnerships for domestic sourcing, ensuring resilience in the supply chain.
AI Snips
Chapters
Transcript
Episode notes
Liquid Cooling Rewrites Data Center Architecture
- Processors now dissipate heat levels air cooling can't handle at rack density required by hyperscalers.
- Liquid cooling moves heat removal to the rack/server level and changes data center architecture from centralized to distributed.
Reliability Becomes The Top Design Constraint
- Reliability demands rise sharply with distributed liquid systems because pump failures can cascade quickly.
- Hyperscalers require far higher reliability than typical industrial expectations to avoid rapid, damaging outages.
Design To 5.9 Reliability Standards
- Design pumps and CDUs to meet hyperscaler uptime needs and manage coolant velocity precisely.
- Aim for 99.999% (5.9) reliability, not typical industrial 'acceptable' rates like 95% or 99.9%.