Generative Design for Direct-to-Chip Liquid Cooling for Data Centers

2026-04-12

AI Summary

This work addresses the challenge of thermal hotspots in heterogeneous chips caused by non-uniform power distribution, a problem inadequately mitigated by conventional liquid-cooling channel designs that rely heavily on empirical guidelines. To overcome this limitation, the authors propose a generative cooling channel design framework that, for the first time, integrates a lightweight physics-informed thermal model with a constrained reaction–diffusion mechanism. This approach automatically synthesizes non-uniform flow channel geometries tailored to the NVIDIA GB200 chip while respecting practical engineering constraints, and dynamically reallocates cooling capacity through a closed-loop iterative optimization process. Experimental results demonstrate that the proposed method reduces the average chip temperature by more than 5 °C and lowers peak temperatures by over 35 °C compared to traditional parallel-channel designs, substantially enhancing liquid-cooling efficiency.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Sampling/Simulation-based Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Economic aspects and information design of Web, GenAI, and cloud computing

Abstract

Rapid growth in artificial intelligence (AI) workloads is driving up data center power densities, increasing the need for advanced thermal management. Direct-to-chip liquid cooling can remove heat efficiently at the source, but many cold plate channel layouts remain heuristic and are not optimized for the strongly non-uniform temperature distribution of modern heterogeneous packages. This work presents a generative design framework for synthesizing cooling channel geometries for the NVIDIA GB200 Grace Blackwell Superchip. A physics-based finite-difference thermal model provides rapid steady-state temperature predictions and supplies spatial thermal feedback to a constrained reaction-diffusion process that generates novel channel topologies while enforcing inlet/outlet and component constraints. By iterating channel generation and thermal evaluation in a closed loop, the method naturally redistributes cooling capacity toward high-power regions and suppresses hot-spot formation. Compared with a baseline parallel channel design, the resulting channels achieve more than a 5 degree Celsius reduction in average temperature and over 35 degree Celsius reduction in maximum temperature. Overall, the results demonstrate that coupling generative algorithms with lightweight physics-based modeling can significantly enhance direct-to-chip liquid cooling performance, supporting more sustainable scaling of AI computing.

Problem

• direct-to-chip liquid cooling
• thermal management
• non-uniform temperature distribution
• heterogeneous packages
• cold plate design

Innovation

• generative design
• direct-to-chip liquid cooling
• reaction-diffusion
• thermal management
• heterogeneous packaging