Nvidia looks to Mitsubishi Heavy for cooling and power as Japan's AI buildout accelerates

Nvidia’s pivot to Mitsubishi Heavy Industries for cooling and power infrastructure reveals that AI’s true bottleneck has shifted from chip performance to thermal and energy constraints. This triggers a cascade: liquid cooling, phase-change materials, and modular power systems will accelerate into mainstream deployment, raising the capital barrier for new data centers. Japan’s push for AI sovereignty—bolstered by its nuclear restart policy—makes industrial giants like MHI geopolitically compliant partners, reducing Nvidia’s supply chain exposure in Asia-Pacific. In response, AMD and Intel may fast-track alliances with European energy tech firms like Siemens Energy to counterbalance the U.S.-Japan tech axis. Over the next 12–24 months, data center competitiveness will be defined not by GPU density alone, but by watts-per-teraFLOP efficiency and waste-heat reuse—ushering in an era where energy intelligence outweighs raw compute.