Lenovo Neptune

Overview

Lenovo Neptune is Lenovo's proprietary direct water cooling platform for high-performance and AI server systems, derived from IBM mainframe liquid cooling practices acquired with the 2014 server business purchase. Now in its 6th generation (2024), Neptune uses 45°C distilled water as the coolant — eliminating the need for mechanical chillers and chemical additives — flowing through copper piping, aluminum cold plates, and thick EPDM hoses to achieve 80% GPU+CPU cooling efficiency. The platform covers H100, H200, and B200 GPU systems with GPU and CPU covered in the same closed loop, and extends to DIMMs, networking, storage, and PSUs in full configurations.

Neptune delivers quantified performance advantages: 40% less energy consumption, 4:1 server consolidation, and 6.1x compute throughput versus prior-generation air-cooled systems. GPU junction temperature variation is kept below 1% with proper water flow and power delivery — a critical guarantee for AI training and inference workload consistency.

The Neptune deployment timeline: 2012 (first large-scale deployment at LRZ Barcelona), 2018 (Neptune brand launch), 2020 (100% water-cooled solution covering GPU/CPU/DIMMs/networking/storage/PSUs), 2024 (6th generation), 2025 (next-generation AI factory cooling).