DropboxHack Week 2025: How these engineers liquid-cooled a GPU server
Read Full ArticleSummary
The article details a project undertaken during Hack Week 2025 at Dropbox, where engineers developed a custom liquid cooling system for GPU servers to address the increasing thermal demands of AI workloads. The project demonstrated significant thermal performance improvements, achieving 20–30°C lower operating temperatures under stress tests compared to traditional air cooling systems. The engineers utilized scaled-down components typical of data center setups, including radiators, pumps, and sensors, to monitor and optimize performance. This initiative not only aims to enhance current infrastructure but also prepares Dropbox for future high-performance computing needs as AI workloads continue to grow.
Key Learnings
- 1Liquid cooling systems can significantly reduce operating temperatures and improve efficiency for high-powered GPU servers.
- 2Custom-built liquid cooling setups can be more effective than pre-assembled systems, allowing for tailored solutions to specific thermal challenges.
- 3The project highlights the importance of proactive infrastructure planning to accommodate next-generation hardware requirements.
- 4Experimentation during Hack Week fostered innovation and collaboration among engineers, leading to practical applications of emerging technologies.
- 5Understanding the thermal dynamics of AI workloads is crucial for optimizing data center performance and sustainability.
Who Should Read This
Senior Systems Engineers designing thermal management solutions for next-generation GPU infrastructure.
Test Your Knowledge
What are the specific thermal advantages of liquid cooling compared to air cooling in GPU servers?
How does the design of a custom liquid cooling system differ from pre-assembled solutions in terms of performance and adaptability?
What challenges might arise when implementing liquid cooling in existing data center infrastructures?
Why is it important to consider future power requirements when designing cooling solutions for high-performance servers?
How can the insights gained from this project influence future infrastructure strategies at Dropbox?
Topics
More from Dropbox Engineering
View Dropbox engineering blogs →Using LLMs to amplify human labeling and improve Dash search relevance
The article outlines how Dropbox Dash utilizes a retrieval-augmented generation (RAG) approach to enhance search relevance by integrating large language models (LLMs) with human labeling. It explains...
How low-bit inference enables efficient AI
The article discusses the advancements in large machine learning models and the challenges associated with their deployment, particularly focusing on low-bit inference techniques that enhance...
Insights from our executive roundtable on AI and engineering productivity
The article provides insights into Dropbox's approach to enhancing engineering productivity through the adoption of AI tools. It highlights the importance of aligning AI initiatives with business...
Engineering VP Josh Clemm on how we use knowledge graphs, MCP, and DSPy in Dash
In this article, Josh Clemm discusses the technical architecture behind Dropbox Dash, focusing on the integration of knowledge graphs, retrieval methods, and the use of large language models (LLMs)....
Inside the feature store powering real-time AI in Dropbox Dash
The article delves into the implementation of a feature store that powers the AI-driven Dropbox Dash, focusing on how it manages and delivers data signals for effective ranking and retrieval of...