Lenovo Establishes Japan's First Water-Cooled AI Infrastructure Verification Center, 'Neptune® Lab'

Lenovo Japan, led by President Taro Hiyama, has established 'Neptune® Lab,' the first of its kind in Japan, within the 'MC Digital Realty Innovation Lab (DRIL)' at the NRT12 data center in Inzai, Chiba, operated by MC Digital Realty (MCDR). This move addresses the cooling and power challenges arising from high-density GPU environments used for AI inference workloads. The facility will showcase water-cooled servers in operation and serve as a hub for discussing and promoting water-cooling technology in Japan.

At this lab, Lenovo’s 'Lenovo Neptune' water-cooling technology is combined with Nidec’s Coolant Distribution Units (CDUs), GPU servers, racks, networking, and monitoring systems to validate performance, power, cooling, and operational viability for AI inference and training workloads. This allows enterprises to transition from Proof of Concept (PoC) to full-scale production environments by performing thorough infrastructure design and TCO analysis.

While enterprise AI adoption is shifting from experiments to production,インフラ (infrastructure) design—including power, cooling, availability, operations management, and TCO—is becoming as critical as model performance. Lenovo’s 'CIO Playbook 2026' report shows that the percentage of Japanese enterprises in trial or full AI adoption is expected to rise from 21% in 2025 to 68% in 2026, with 93% planning to increase AI investment within the next year. However, only about half of the companies expecting a positive ROI from AI have successfully transitioned to real-world operations, highlighting the challenge of scaling AI from PoC to production.

With the increasing adoption of servers equipped with high-performance GPUs, power consumption and heat generation per rack have surged, often exceeding the capabilities of traditional air-cooling designs. Consequently, there is growing anticipation for water-cooling technologies that balance cooling and power efficiency to support the stable operation of high-density AI infrastructure.

'Lenovo Neptune' is designed for AI and HPC, providing integrated design encompassing servers, racks, cooling, power, and management software to support density, operational efficiency, and sustainability. Through this lab, customers can verify configurations, evaluate performance, and confirm power/cooling efficiency based on their own workloads and requirements before final implementation, utilizing validated solutions featuring Nidec’s CDUs.

Starting with this lab, Lenovo aims to expand its collaboration with customers, ISVs, cloud providers, system integrators, and data center operators to form a community that shares expertise on water-cooled AI infrastructure.