FREMONT, CA
DeepCoolAI (DCAI), a leading provider of advanced liquid cooling and AI data center infrastructure solutions, today announced the development of its Modular 2.5MW InRow Cooling Distribution Unit (CDU) platform, engineered to scale seamlessly to 10MW and beyond.
As AI compute density accelerates, hyperscalers are demanding cooling architectures that scale as rapidly as GPU clusters. DeepCoolAI’s new 2.5MW modular CDU platform is purpose-built to serve as the thermal backbone of AI factories, delivering scalable, serviceable, and deployment-ready liquid cooling infrastructure.
Kris Holla, Founder & CEO of DeepCoolAI, commented:
“AI factories demand cooling infrastructure that scales in megawatts, not kilowatts. Our modular 2.5MW CDU platform provides a future-proof thermal backbone designed specifically for AI. This is cooling infrastructure built for the next decade of compute.”
DeepCoolAI Introduces Modular 2.5MW InRow CDU Platform Scalable to 10MW+ for AI Factories
Built for AI — Not Retrofitted for It
Unlike legacy plant-based cooling systems originally designed for air-cooled environments, the DeepCoolAI 2.5MW platform is engineered specifically for:
• High-density AI and accelerated computing
• ORV3 and next-generation rack architectures
• Direct-to-chip and immersion environments
• Phased AI cluster expansion
Each 2.5MW module functions as a self-contained cooling block, capable of operating independently or as part of a multi-module deployment.
True Modular Scalability: 2.5 → 5 → 7.5 → 10MW+
DeepCoolAI’s platform allows customers to:
• Deploy 2.5MW at initial cluster build
• Expand linearly to 5MW, 7.5MW, or 10MW
• Operate in distributed InRow architecture or centralized CDU plant
• Add capacity without redesigning mechanical infrastructure
Cooling should scale with compute — not require new engineering every time capacity doubles.
Engineered for Redundancy and Maintainability
The 2.5MW CDU platform incorporates:
• Parallel pump architecture with VFD control
• Redundant heat exchanger design
• Integrated filtration and monitoring systems
• Configurable N+1 or distributed redundancy
• Factory-tested modular skids for rapid commissioning
Each module is designed for high serviceability and operational continuity critical in AI environments where downtime is unacceptable.
Capital Efficiency Through Phased Deployment
By aligning cooling capacity directly with AI workload ramp-up, DeepCoolAI enables operators to:
• Avoid stranded cooling capacity
• Reduce initial capital expenditure
• Deploy infrastructure in revenue-aligned phases
• Accelerate time-to-compute
This modular approach provides hyperscalers with both financial flexibility and infrastructure agility.
Integrated AI Infrastructure Platform
The 2.5MW modular CDU integrates seamlessly with DeepCoolAI’s broader ecosystem, including:
• Hybrid Load Banks for thermal validation
• Liquid flushing and fill systems
• RDHx platforms
• High-density rack-level CDUs
Together, these form a vertically integrated AI cooling platform — from commissioning to full production.
About DeepCoolAI
DeepCoolAI is a One Stop Factory Direct Liquid Cooling for Next Gen AI Factories.
DeepCoolAI, offers a comprehensive range of solutions for liquid cooling, including CDUs, Load Banks, Refill Carts, and supporting products such as RDHx and Fanwalls. We specialize in custom-tailored CDUs, providing bespoke solutions for data centers ranging from 1 megawatt to 6 megawatts. Our plug-and-play Load Banks are designed to streamline the commissioning and startup of AI liquid-cooled data centers. Additionally, our state-of-the-art Refill Carts ensure that your liquid-cooled servers and CDUs remain operational, helping you maximize uptime. Recently, Sanmina and DeepCoolAI entered into a strategic partnership for global manufacturing and supply chain at scale.
4/2/2026