LiquidStack secured a 300-megawatt order for Coolant Distribution Unit (CDU) capacity from a major U.S.-based data center operator, signaling growing commitment to liquid cooling as core infrastructure for AI workloads. The multi-site deployment will support large-scale, AI-ready data center expansions across the United States as operators push toward higher rack densities and accelerated build-out schedules.
The order centers on LiquidStack’s CDU-1MW platform, designed for rapid deployment and scalability in next-generation facilities. The customer, described as a long-established operator with a growing national footprint of AI-ready sites, selected LiquidStack to support high-density environments using direct-to-chip and hybrid liquid cooling architectures.
LiquidStack said its manufacturing and delivery capabilities will enable accelerated fulfillment of the 300-MW order across multiple sites. The announcement builds on recent momentum for the company, including expanded manufacturing capacity in Carrollton, Texas, and its inclusion on NVIDIA’s Recommended Vendor List for CDUs supporting accelerated computing platforms.
- Order size: 300 MW of CDU capacity
- Product: CDU-1MW
- Deployment: Multi-site, U.S.-based AI-ready data centers
- Use case: High-density AI and accelerated computing workloads
- Cooling architectures: Direct-to-chip and hybrid liquid cooling
“Orders of this size signal a clear inflection point for liquid cooling,” said Joe Capes, CEO of LiquidStack. “Operators are committing to liquid cooling as core infrastructure for AI, and LiquidStack is uniquely positioned to support that transition at scale.”
🌐 Analysis
The 300-MW order reflects how liquid cooling is shifting from pilot deployments to standardized infrastructure in AI-focused data centers, particularly as rack power densities climb well beyond air-cooling limits. As hyperscale, colocation, and neocloud operators accelerate AI capacity build-outs, suppliers that can deliver CDU platforms at scale—and integrate cleanly with direct-to-chip designs tied to NVIDIA-class accelerators—are increasingly positioned as strategic partners rather than niche vendors.
