Nvidia CEO Jensen Huang met with Japanese Prime Minister Shigeru Ishiba in Tokyo on April, 21. Huang urged Japan to expand its electricity generation to meet the growing power demands of AI development.

He highlighted Japan’s strengths in robotics and industrial manufacturing.

He stressed the need for new infrastructure to support energy-intensive AI technologies, saying, “Energy is essential for all industrial growth.”

Huang’s visit follows a trip to Beijing and occurs amid rising tensions in the US-China tech rivalry. The US government has barred Nvidia from selling its H20 AI chips to China, which are designed to comply with prior US export restrictions.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ The energy-AI paradox facing Japan after Fukushima

Jensen Huang’s appeal for more power in Japan highlights a critical irony: as AI drives unprecedented energy demand, Japan’s energy landscape remains constrained by the aftermath of the 2011 Fukushima disaster.

Before Fukushima, Japan relied on nuclear for 30% of its electricity with plans to increase to 50%, but the disaster led to taking approximately 50 reactors offline, causing Japan to increase LNG imports by nearly 25% and oil consumption by 85% in 2011 alone 1.

This created a massive financial burden, with Japan spending an additional ¥3 trillion (over $35 billion) on fossil fuel imports in 2011, contributing to a significant trade deficit that continues to influence energy policy today 1.

The timing of Nvidia’s request is particularly challenging as global data centers are projected to more than double their electricity consumption by 2030, driven largely by AI applications that are dramatically more energy-intensive than traditional computing 2.

This creates a strategic dilemma for Japan: how to power AI innovation while managing energy security in a nation that imports nearly all its fossil fuels and faces continued public resistance to nuclear power.

2️⃣ Global race for AI-ready power infrastructure intensifies

Huang’s meeting in Tokyo represents a pattern emerging worldwide: AI companies lobbying governments for prioritized access to reliable, abundant electricity as power availability becomes a competitive advantage in global AI development.

In the United States, data centers already consume 4.4% of total electricity demand (176 terawatt-hours in 2023), with projections indicating a need for over 800 TWh annually by 2030 – a more than 4x increase 3.

“Speed-to-power,” or the time required for data centers to secure electricity access, has become a critical bottleneck, with some regions like Virginia facing electricity supply wait times of up to 7 years for new data center connections 3.

Recent Nvidia developments