---
格式版本: 2
标题: "Rethinking data center cooling for AI: The rise of direct-to-chip liquid cooling - DCD"
原文链接: "https://www.datacenterdynamics.com/en/opinions/rethinking-data-center-cooling-for-ai-the-rise-of-direct-to-chip-liquid-cooling/"
发布日期: "未标注"
发现时间: "2026-06-17T11:12:49+08:00"
入库时间: "2026-06-22T10:33:03.021Z"
来源平台: "Exa"
搜索渠道: "exa_search"
搜索词: "Liquid Cooling"
匹配关键词:
  - "Liquid Cooling"
  - "GPU"
  - "Cold Plate"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "XCrawl Scrape"
清洗工具: "XCrawl Markdown + LLM 正文裁剪"
原始附件:
  []
AI优质: "否"
AI打分: 39
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "文章为厂商营销人员撰写的液冷科普，缺乏具体架构、参数等技术细节，无商业部署信号，且未聚焦超节点/机柜级AI系统，属于低价值内容。"
AI质检模型: "deepseek-v4-flash"
AI质检时间: "2026-07-30T01:23:27+08:00"
AI主题相关性: 12
AI来源权威性: 10
AI新颖性: 5
AI技术细节: 5
AI商业部署信号: 0
AI完整性: 7
采集批次: "2026年6月22日17点39分25秒"
采集批次ID: "20260622-173925-591"
去重键: "https://www.datacenterdynamics.com/en/opinions/rethinking-data-center-cooling-for-ai-the-rise-of-direct-to-chip-liquid-cooling"
---

## Rethinking data center cooling for AI: The rise of direct-to-chip liquid cooling

![Christopher Leonard](https://media.datacenterdynamics.com/media/images/cropped-Headshots-Motivair-Corp-4-.2021072a.fill-180x180.jpg)

Christopher Leonard is director of marketing at Motivair by Schneider Electric

As the demand for AI and HPC infrastructure continues to grow, the move from air cooling to liquid cooling will become increasingly important for cooling AI workloads

As artificial intelligence (AI) and high-performance computing (HPC) continue to push the limits of processing power, traditional approaches to data center cooling must evolve. Data centers must innovate to manage the substantial heat generated by increasingly powerful GPUs and CPUs.

One of the most effective emerging solutions is direct-to-chip liquid cooling, which supports AI workload cooling by delivering efficient heat management while enhancing sustainability and performance.

## Why traditional cooling is no longer enough for HPC infrastructure cooling

Traditional [air-cooling systems](https://blog.se.com/datacenter/2025/11/18/air-vs-liquid-cooling-finding-right-strategy-ai-ready-data-centers/) are struggling to keep up with the heat generated by today’s AI-driven workloads. Air cooling is reaching its limits as GPU heat density increases. Even at higher fan speeds, air lacks sufficient thermal capacity to dissipate heat efficiently, leading to hot spots, thermal throttling, and increased failure risk in AI and HPC environments.

In addition, air cooling uses significant electricity as fans run continuously to manage rising GPU heat. [Data centers already consume an estimated two percent of the world's electricity](https://www.datacenterdynamics.com/en/news/global-data-center-electricity-use-to-double-by-2026-report/), a figure expected to increase as AI adoption grows. A different approach is needed for HPC infrastructure cooling.

## Why liquid cooling is crucial for AI data centers

Liquid cooling is a critical solution, being up to [3,000 times more effective than using air cooling](https://www.gartner.com/en/information-technology/glossary/liquid-cooling), enabling higher compute density while reducing energy usage. By directly absorbing and dissipating heat from the hottest components, such as GPUs, liquid cooling provides a better approach for cooling AI workloads. It maintains chips at optimal operating temperatures, enhancing performance and reliability.

## What is direct-to-chip cooling in AI and HPC?

[Direct-to-chip cooling](https://www.motivaircorp.com/products/liquid-cooling-systems/) is highly efficient for cooling generative AI models, in which coolant – typically water – flows directly across the hottest components, such as CPUs and GPUs. The coolant absorbs heat via cold plates attached to these chips and transfers it outside the data center to a heat exchange system. This approach minimizes the need for large fans, reducing energy consumption and freeing up valuable space for higher computing density.

## Why single-phase direct-to-chip cooling is the preferred choice

In single-phase cooling, the coolant remains in its liquid state throughout the process, ensuring stable and predictable thermal transfer. Unlike two-phase cooling, where the liquid turns to vapor, single-phase systems offer simpler maintenance and higher reliability.

This makes [single-phase cooling](https://www.motivaircorp.com/news/what-dells-study-means-for-data-center-cooling/) the go-to method for many data centers, especially those running AI and HPC workloads.

## HPC thermal solutions that enable direct-to-chip cooling

Successfully implementing direct-to-chip cooling relies on a set of integrated HPC thermal solutions designed to manage the extreme heat generated by AI and high-performance computing workloads. Together, these technologies form a complete liquid-cooling ecosystem that delivers reliable, high-density data center cooling.

- [Coolant distribution units (CDUs):](https://www.motivaircorp.com/products/CDU/)

CDUs control the temperature and flow of the coolant, ensuring it reaches the servers under the right conditions. They are essential for single-phase cooling systems, helping balance the entire liquid-cooling infrastructure.

- [In-rack manifolds:](https://www.motivaircorp.com/products/in-rack-manifold/)

These manifolds distribute the coolant throughout the rack, connecting to each cold plate with leak-proof, color-coded quick disconnects for easy maintenance. They are crucial for maintaining stable cooling across high-density computing setups.

- [Cold plates:](https://www.motivaircorp.com/products/dynamic-cold-plates/)

Cold plates are mounted directly on CPUs and GPUs to dissipate heat more efficiently than traditional heat sinks. By using high-conductivity materials and direct contact with the hottest components, cold plates enable the precise heat removal required to address AI workload cooling.

Cold plates are mounted directly on CPUs and GPUs, drawing heat away from these components more effectively than traditional heat sinks. Their superior thermal conductivity is vital for handling the high-power densities present in modern AI and HPC environments.

- [Rear door heat exchangers:](https://www.motivaircorp.com/products/chilleddoor/)

Positioned at the rear of server racks, these exchangers dissipate heat from the coolant before it is recycled, ensuring efficient cooling across the data center.

## The future of AI and HPC infrastructure cooling: A shift toward direct-to-chip solutions

As the demand for AI and HPC infrastructure continues to grow, the move from air cooling to liquid cooling – specifically direct-to-chip systems – will become increasingly important for cooling AI workloads. This technology offers the performance needed to handle data center cooling while significantly reducing energy consumption and CO2 emissions.

Organizations aiming to future-proof their AI and HPC infrastructure should consider adopting direct-to-chip cooling solutions to meet growing computational demands while staying energy-efficient and sustainable. To learn more, access the executive report “ [Optimizing AI Infrastructure: The Critical Role of Liquid Cooling](https://www.se.com/ww/en/download/document/SPD_EB3_EN/) ” to see how a comprehensive, system-level approach integrates liquid cooling with existing data center infrastructure to support high-density AI workloads.