China’s AI Ambitions Hit a Snag as Data Centers Sit Idle

China’s aspirations to lead in the AI sector are facing a significant hurdle, as an overwhelming number of newly built data centers remain unused. According to the MIT Technology Review, despite the country investing billions into infrastructure to support AI development in 2023 and 2024, as much as 80% of this new capacity is currently sitting idle.

The Boom That Didn’t Last

The hype surrounding AI and the allure of GPU rentals led to a flurry of construction across the nation. In an effort to position itself as a global AI leader, the Chinese government encouraged local officials to expedite data center projects. By the end of 2024, more than 500 projects were announced, with at least 150 completed. However, the investment was premised on ever-increasing demand, which has since dwindled, causing operators to struggle to keep their heads above water.

Challenges in Central China

What’s more, location has become a double-edged sword. Many facilities built in central and western China, where electricity is more affordable, are now grappling with latency issues. In places like Zhengzhou, desperate operators are resorting to handing out free compute vouchers just to attract users. This scenario has led some developers to begin offloading GPUs to avoid accumulating losses. A project manager, Xiao Li, noted a significant shift in the conversation within WeChat groups that once boasted about Nvidia chip deals. “It seems like everyone is selling, but few are buying,” he lamented.

Should these idle assets flood the market, it could present a dire situation for data center developers already facing challenges. An oversupply could push prices down further, creating a bumpy road ahead.

A Shift in AI Demand

A major reason for the declining demand for data centers is the recent popularity of DeepSeek, which upended the tech landscape with its launch in January 2025. This open-source reasoning model, R1, delivers performance matching that of ChatGPT o1 but at a fraction of the cost. This shift has redirected interest from model training—which requires massive infrastructure—to real-time AI use, a landscape that necessitates different infrastructure capabilities.

Unfortunately, many of the new data centers constructed during the AI boom were designed for large-scale training operations rather than the low-latency needs of real-time inference.

Despite the glum outlook, the Chinese government remains undeterred. An AI symposium held in early 2025 showcased commitment to this sector, with major players like Alibaba and TikTok owner ByteDance announcing substantial investments in AI.

Closing Thoughts

For early investors who anticipated high demand, the reality has been disheartening: while the infrastructure is in place, the expected demand simply hasn’t materialized. This situation serves as a lesson about the unpredictable nature of tech investments and the necessity of aligning infrastructure with actual market needs.

The AI Buzz Hub team is excited to see where these breakthroughs take us. Want to stay in the loop on all things AI? Subscribe to our newsletter or share this article with your fellow enthusiasts.