HyperCool: The Data Center Cooling System for Today’s AI Factories

Modern data centers are being pushed to their limits. AI, high-performance computing (HPC), and advanced workloads have made traditional cooling systems obsolete. With individual processors now able to operate at 2,800 watts (and beyond), data center heat generation has skyrocketed. The result? Power consumption has, too. New and immense power draws for cooling have laid bare the shortcomings of legacy air and water cooling infrastructure. Air can’t keep up. Water is simply too expensive, risky, and demanding on space and resources.

The waterless, direct-to-chip HyperCool® liquid cooling system from ZutaCore® is the solution. Purpose-built for AI factories and high-density computing, HyperCool eliminates the risks and inefficiencies of water-based systems while unlocking unmatched performance and sustainability.

Let’s dive into how and why HyperCool has quickly set the bar for data center cooling that meets the demands of today’s technology.

Table of Contents:

• 5 Challenges AI Factories & Data Centers Must Overcome
• The Solution: Waterless, Closed-Loop, Two-Phase, Direct-to-Chip Cooling
• Get to Know the HyperCool System’s Components
• How HyperCool Delivers ROI & Sustainability at Scale
• Where’s the Proof? Real-World HyperCool Scenarios
• The Next Step: Transform Your Data Center With HyperCool

5 Challenges AI Factories & Data Centers Must Overcome

As AI and HPC workloads grow, data center operators face at least five critical challenges:

1. Rising Power Consumption

The International Energy Agency predicts that global data center power usage could double by 2026, driven largely by AI. Hotter processors, like NVIDIA’s Grace Blackwell chips, demand unprecedented cooling capabilities. Air cooling simply cannot keep up.

2. Heat Management at Hyperscale

Hyperscale data centers are experiencing unprecedented heat loads. Without effective heat rejection and dissipation, facilities risk hotspots, inefficiencies, and performance degradation. HyperCool addresses this with configurable heat rejection units that support 100% heat reuse. Facilities can repurpose waste heat for district heating or industrial processes, as needed. As a result, net zero data centers are still on the menu, even in the era of AI.

3. Densification of Infrastructure

Modern data centers must maximize compute density within limited footprints. HyperCool triples processing capacity while reducing energy use by 50%. Its low-profile design fits within 1U dimensions, enabling seamless retrofits and high-density configurations.

4. Sustainability Pressures

With public scrutiny on carbon footprints and water scarcity, data centers must adopt greener practices. HyperCool eliminates water usage entirely while reducing energy consumption by up to 80%. Its closed-loop system supports zero-emission, net-zero energy data center strategies.

5. Limitations of Air Cooling

Air-cooled systems consume vast amounts of energy and require significant physical space. As processors grow hotter, air cooling becomes inefficient, leading to hotspots and performance degradation. The direct-to-chip HyperCool solution eliminates these shortcomings while using half the space and 80% less energy than air cooling.

The Solution: Waterless, Closed-Loop, Two-Phase, Direct-to-Chip Cooling

A new approach to cooling technology can solve all five of the main challenges faced by modern data centers and AI factories. HyperCool flips the script on those struggles by integrating four core features:

Waterless Heat Transfer Fluid

Unlike water-based systems, HyperCool uses a non-toxic, non-conductive heat transfer fluid. In the unlikely event of a leak, the fluid evaporates harmlessly into the air without damaging IT hardware or harming the environment. Operators eliminate the risks of corrosion, mold growth, or catastrophic downtime normally tied to (water-based) liquid cooling.

Closed-Loop Design

A closed-loop liquid cooling system keeps the fluid contained within a sealed environment. This dramatically reduces the need for frequent maintenance, since the fluid retains its integrity indefinitely without contamination and doesn’t exit the system via evaporation.

Two-Phase Cooling

Two-phase systems are a leap forward from traditional single-phase cooling. Instead of relying solely on liquid to absorb heat, the fluid evaporates into vapor upon contact with the processor. The vapor then condenses back into liquid within a heat rejection unit (HRU), completing the cycle. This phase change significantly improves cooling efficiency.

Direct-to-Chip Cooling

Direct-to-chip cooling applies the heat transfer fluid directly to the processor surface. By targeting the hottest part of the server—the CPU or GPU—this method extracts and dissipates heat quickly and evenly. Direct-to-chip cooling maximizes performance while eliminating the need for bulky fans, heat sinks, and inefficient air systems.

Compared to immersion cooling, which requires hundreds of gallons of fluid, HyperCool operates with less than 4 gallons of heat transfer fluid. Its compact footprint and low-pressure design make it ideal for high-density, modern data centers.

Get to Know the HyperCool System’s Components

The HyperCool system delivers exceptional cooling performance through its modular and scalable design. Its key components work seamlessly to manage extreme heat loads, ensuring maximum efficiency, safety, and reliability for AI, HPC, and densified workloads.

Cold Plates

HyperCool cold plates deliver direct-to-chip cooling to the hottest parts of the server—the CPUs and GPUs—from leading manufacturers, including Intel, AMD, and NVIDIA. With a non-toxic, waterless heat transfer fluid, the cold plates efficiently extract heat without the risks associated with water-based systems.

• Advanced Pool-Boiling Technology: Unlike traditional systems, HyperCool employs pool-boiling instead of flow-boiling. This innovation ensures even heat distribution, eliminating thermal hotspots that can degrade performance.
• High Wattage Support: HyperCool cold plates handle processors exceeding 2,800W, making them ideal for AI, HPC, and high-density computing environments.
• Compact Design: With a low-profile form factor, the cold plates fit within standard 1U servers, replacing bulky heat sinks and fans, enabling increased compute density.
• IT Safe: In the unlikely event of a leak, the non-conductive fluid evaporates harmlessly, preventing downtime or damage to IT equipment.

Heat Rejection Units (HRUs)

Our HRUs are responsible for condensing the vaporized fluid back into liquid form in a closed-loop cooling cycle. Available in air-based and water-based options, HRUs offer flexibility to meet the needs of any data center environment.

• Air-Based HRUs: These units reject heat into the ambient air, which can help with overall facility temperature control. Save energy with less need for traditional HVAC systems — ideal for facilities that want to prioritize energy savings.
• Water-Based HRUs: For facilities with existing water infrastructure, these HRUs enable efficient heat reuse. Heat captured from the servers can be repurposed for district heating, industrial processes, or internal applications, reducing overall energy consumption.
• Scalable Design: HRUs are modular and support deployments from 60kW (3U configuration) up to 120kW (6U configuration) and beyond, ensuring compatibility with both new builds and retrofitted facilities.
• Redundant Features: Independent sensors, pumps, and power supplies provide multi-level leak detection for uninterrupted operation and complete system reliability.

HyperCool Manifold

The manifold acts as the central hub for fluid delivery and heat transport. It seamlessly integrates with server racks for uniform thermal management.

• Dual-Chamber Design: The manifold separates vapor and liquid channels. The liquid heat transfer fluid heads to the cold plates in a blue line while vaporized fluid flows back to the HRU for condensation in a red line.
• Hot-Swappable Functionality: Designed for operational flexibility, the manifold allows servers to be serviced without disrupting the system. Less downtime means more productivity during server services.
• Compact Installation: Flexible mounting options allow the manifold to fit seamlessly on standard rack configurations. The system supports up to 42 servers per rack.
• Customizable Integration: Adjustable length and modular connections make the manifold adaptable to both standard and custom rack setups.

Software Defined Cooling (SDC)

The Software Defined Cooling platform provides real-time monitoring and automated optimization to ensure peak system performance.

• Advanced Analytics: Our optional SDC system tracks CPU and GPU temperatures, power consumption, fan speeds, and overall system utilization. Operators can get actionable data-based insights for more responsive and efficient operations.
• Leak Detection and Alerts: Automated monitoring in the SDC software identifies anomalies and potential leaks for proactive maintenance.
• Efficiency Optimization: By adjusting cooling performance based on real-time load demands, SDC minimizes energy usage while maintaining stability.
• Seamless Integration: SDC integrates smoothly with existing data center management software so operators can have a unified monitoring and control solution.

How HyperCool Delivers ROI & Sustainability at Scale

HyperCool transforms data center economics and environmental impact. Measurable savings and ambitious sustainability goals are within reach via a simple retrofit or easily scalable system for new data center construction.

50% Lower Total Cost of Ownership (TCO)

HyperCool reduces energy use, operating costs, and infrastructure requirements, ensuring rapid ROI and long-term financial benefits.

• Operational Savings: HyperCool reduces energy usage by up to 80%, translating into substantial annual savings. For a 2MW data center, this can save operators up to $1 million annually.
• Capital Savings: Eliminate chillers, CRAC units, and massive airflow infrastructure, cutting CAPEX by up to 50%.
• 82% Energy Efficiency Gains: Optimized energy use drastically lowers overall power consumption, enabling PUEs as low as 1.04.
• Rapid ROI: With energy savings, scalability, and reduced maintenance costs, HyperCool delivers a return on investment in under two years. New AI factory deployments can experience immediate cost benefits.

10x Compute Density

Scale seamlessly, accommodate growing workloads, and increase profitability without expanding square footage or power usage.

• Compact Design: HyperCool achieves more than 250kW per rack without expanding physical space. Its 1U profile fits seamlessly into new builds or retrofits.
• AI Processor Support: HyperCool handles processors exceeding 2,800W, ensuring compatibility with next-generation chips like NVIDIA’s Grace Blackwell.
• Resource Efficiency: Operators achieve 3x more performance per square foot with half the power consumption of traditional cooling systems.

Zero Water, Zero Risk

Unlike water cooling, the waterless heat transfer fluid behind HyperCool delivers unmatched efficiency, safety, and performance in a compact package.

• No Water Dependency: HyperCool eliminates water use entirely with a non-toxic, waterless heat transfer fluid, removing risks of leaks, corrosion, and water scarcity.
• Low Fluid Volume: Compared to immersion cooling, which requires up to 400 gallons, HyperCool uses less than 4 gallons per system.
• Safe Operation: The heat transfer fluid evaporates harmlessly and poses zero risk to IT equipment, even in the event of a leak.

100% Heat Reuse & Sustainability

HyperCool makes it possible for even the most advanced AI facilities to reduce emissions, conserve resources, and achieve net-zero goals.

• Circular Energy Economy: HyperCool enables 100% heat reuse, repurposing waste heat for district heating, office buildings, and industrial processes.
• Ultra-Low PUE: Achieve industry-leading PUE as low as 1.04, maximizing efficiency and lowering carbon emissions.
• Sustainability Goals: Cut greenhouse gas emissions by reducing energy waste and enabling a circular energy system.
• Compact, Efficient Design: With less infrastructure and no water requirements, HyperCool reduces land use and supports sustainable data center growth.

Where’s the Proof? Real-World HyperCool Scenarios

A 2MW data center is under immense pressure to keep pace with rising AI and HPC demands. A series of NVIDIA GB-equipped server racks have come online, and the heat they generate is unlike anything the facility has dealt with before. Operational costs are climbing, energy consumption is surging, and sustainability targets loom overhead.

The facility’s manager has looked into upgrading air-cooled systems, but they couldn’t handle the heat loads efficiently. Immersion cooling was also taken into consideration, but it would take extensive infrastructure changes and costly maintenance.

By implementing HyperCool, this 2MW facility transforms into a high-efficiency powerhouse:

• Energy and maintenance costs plummet, freeing up $1M in annual savings for growth.
• Expensive chillers and bulky infrastructure become a thing of the past — a change that drives a 50% CAPEX Reduction.
• Power-hungry air cooling systems are replaced with streamlined, waterless liquid cooling that now use 82% less energy.
• Waste heat from the facility’s new AI chips captured and repurposed with 100% heat reuse to support on-site heating or nearby facilities and reach ambitious carbon emission reduction goals.

In this realistic scenario, the data center finds rapid ROI, immediate scalability, and a clear path to meet AI-driven workloads, all without compromising on performance or environmental goals. HyperCool has done more than cool. It has effectively future-proofed the data center.

Scenarios like this are not just theoretical; they’re already happening. Explore the real case study of the University of Münster’s implementation of HyperCool to see how it helped deliver the most computing power for the University’s money. Extensive tests by Storage Review — performed at the Centersquare Data center in Reading, Berkshire, UK, on a Supermicro server with dual AMD EPYC Bergamo CPUs — have also showcased the data-driven potential of HyperCool to push hardware to its limits. 

The Next Step: Transform Your Data Center With HyperCool

The time for incremental improvements has passed. AI, HPC, and high-density workloads require a revolution in cooling technology. HyperCool delivers the performance, reliability, and sustainability that modern data centers need to thrive.

Are you ready to redefine your data center? Connect with ZutaCore today to learn how HyperCool can maximize your compute density, reduce operational costs, and achieve net-zero sustainability goals. Together, we can build a future where data centers drive progress without compromising the planet.