Japan Faces Energy Uncertainty as DeepSeek Reshapes AI Power Demands

New AI Developments Challenge Japan’s Energy Forecasts

TOKYO, Jan 31 (Reuters) – Japan’s carefully crafted energy strategy, designed in part to accommodate rising artificial intelligence-driven electricity needs, has been shaken by the unexpected rise of Chinese AI startup DeepSeek. The emergence of its seemingly more energy-efficient models has prompted experts to reassess the anticipated power demands of the sector.

Previously, Japan had moved away from long-standing forecasts predicting declining electricity consumption due to its aging population. Instead, it began incorporating projections for increased energy use from AI-driven data centers and semiconductor manufacturing.

In December, the government released a draft of its basic energy plan, projecting a 10-20% increase in electricity generation by 2040. While officials are unlikely to rush into revising these forecasts, DeepSeek’s innovations have sparked widespread debate on the future of AI energy consumption.

"Ignoring this development would be risky," said Andrew DeWit, professor at Rikkyo University's School of Economic Policy Studies in Tokyo.

DeepSeek recently introduced a free AI assistant that operates with significantly lower data usage and costs compared to existing models. By Monday, it had surpassed U.S. competitor ChatGPT in downloads from Apple’s App Store, triggering a global tech stock selloff.

Japan’s energy sector, which had been buoyed by expectations of soaring electricity demand due to AI expansion, also felt the impact, as investors recalibrated their expectations in light of DeepSeek’s potentially more energy-efficient technology.

However, some analysts suggest DeepSeek’s success may not necessarily reduce overall AI power consumption. Instead, by lowering the entry barrier for AI developers, it could encourage a surge of new players in the field, ultimately driving higher electricity demand.

Japan, which produces only 13% of its energy domestically—the second-lowest among OECD nations—must carefully navigate these shifting dynamics. "If AI development becomes cheaper and more accessible, its widespread adoption could actually accelerate electricity demand rather than reduce it," said Yuriy Humber, CEO of Tokyo-based energy consultancy K.K. Yuri Group.

Despite acknowledging AI’s influence on power demand, Japanese policymakers have been slow to adjust their long-term forecasts. "The AI boom was evident two years ago, but official energy plans have been slow to reflect this," Humber noted, adding that the government will likely monitor these developments closely.

The Ministry of Economy, Trade and Industry, which oversees Japan’s energy policy, declined to comment on the matter. However, a senior ministry official speaking anonymously noted that while DeepSeek’s technology was impressive, it did not significantly advance existing AI models. He suggested that markets had overreacted to its emergence.

Japan’s largest power provider, Tepco, stated that it is closely monitoring DeepSeek’s potential impact but has not yet assessed the full implications.

AI and Japan’s Energy Future

For years, Japan’s energy forecasts predicted gradual declines in electricity demand due to population shrinkage and energy-efficient technology adoption. However, in 2024, the outlook was revised to reflect a projected 5.14 million kWh increase in demand from data centers and chip manufacturers by 2034.

AI-driven power consumption has also been used as a rationale for restarting Japan’s nuclear reactors—a politically sensitive issue in a nation still haunted by the 2011 Fukushima disaster.

Mika Ohbayashi, director at Tokyo’s Renewable Energy Institute, views DeepSeek’s rise as evidence that AI could evolve to be less energy-intensive. She criticized government officials for using AI energy needs to justify nuclear expansion, arguing that Japan should instead prioritize renewable energy development.

The debate over whether AI advancements will increase or decrease energy consumption remains unresolved. Analysts stress that DeepSeek’s models must undergo rigorous evaluation before policymakers make any significant changes to energy strategies.

Japan has faced costly miscalculations in its energy planning before, warned DeWit of Rikkyo University. He pointed to the late 1980s when Japan, then a semiconductor leader, overbuilt its power infrastructure under the assumption of continued dominance—only for demand to collapse after the country’s economic bubble burst.

"Policymakers must take these developments seriously to avoid repeating past mistakes," he cautioned.

The Reuters Power Up newsletter provides essential insights into the global energy sector. Sign up here.

Reporting by John Geddie and Katya Golubkova; Additional reporting by Tim Kelly and Yuka Obayashi; Editing by Kim Coghill.

Our Standards: The Thomson Reuters Trust Principles.