China has faced problems with its state-supported data center buildout. People have jumped into data center construction without real industry knowledge. Utilization rates for some AI data centers were around 30% last year. “However, not all of the investment attracted by subsidies has been rational. Jin Yadong said that many computing centers rushed into construction are typically funded by a fragmented mix of private and municipal capital. “These investors tend to have plenty of money but little understanding of the industry, making it difficult for them to find suitable customers,” he said. These “outsider” investors, focused on short-term returns, often lack a deep understanding of trends in AI technology and fail to develop the necessary complementary industrial planning. As a result, some computing centers are built around a single performance metric, only to become “outdated as soon as they are completed.” They struggle to meet the market’s increasingly complex demand for AI computing, leading to an oversupply of low-quality computing capacity.” This piece from Science & Technology Daily is from last year but still relevant now:

Chinese startups are rushing to develop nuclear small modular reactors (SMRs). One startup exec imagines SMRs will replace old coal plant sites across China as coal gets phased out. Fascinating piece by @edwardwhitenz

“Shortly after Anthropic released the report, the Cyberspace Administration of China summoned representatives of the seven companies identified by the AI firm for questioning. The regulator later zeroed in on DeepSeek and Moonshot.”

My NEW op-ed for The New York Times "Rather than jump straight into debates over a global slowdown or an arms control treaty for AI, the US and China should start with something much more basic: realizing & acknowledging how the other side sees AI risks."

Xi Jinping warning about “technological loss of control” at a Politburo study session on “future industries” earlier this year. Not just about AI, but can be read as including AI. “Development and security must be coordinated. China should explore scientific and effective approaches to regulation and build systems for technology monitoring, risk early warning, and emergency response, so as to proactively address emerging risks such as loss of control over technology, ethical failures, and data misuse. The aim is to ensure that innovation is given sufficient room to flourish while also being effectively governed, creating a favorable environment for technological innovation and industrial development.”

Three times when China paused a new technology it was racing to build: 1) High-speed rail projects after 2011 Wenzhou crash 2) Nuclear plant approvals after Fukushima 3) Robotaxi licenses after Baidu Apollo Go robotaxi freeze in Wuhan this April In these cases, China wasn’t racing with the US like with AI. But it’s still interesting to consider these cases where a major incident caused China to unilaterally decide to pause temporarily. And one incident—Fukushima—didn’t even take place in China.

People’s Daily piece pushing back on the NSA report on AI distillation. What’s interesting to me is that Beijing never denies doing distillation and instead argues it’s a common industry practice. Headline: “Politicizing technology issues, an old tactic of US hegemony” “AI is a new frontier of human development. It is not the preserve of major powers, still less should it be reduced to an arena for zero-sum competition. Recently, the U.S. National Security Agency, Cybersecurity and Infrastructure Security Agency, and Federal Bureau of Investigation jointly issued an advisory accusing Chinese artificial intelligence companies of conducting “industrial-scale” distillation against the United States to obtain the capabilities of U.S. frontier models, while recommending related defensive measures to U.S. companies. The U.S. accusation is unsupported by facts and lacks any legal basis. In essence, it politicizes and instrumentalizes distillation—a normal technical and commercial practice in the industry—and applies a double standard in practice. This once again exposes the United States’ longstanding approach of pursuing technological hegemony in artificial intelligence, monopolizing computing power, and suppressing competition. China firmly opposes this. Distillation is a common practice through which models in the artificial intelligence field learn from one another. It is, in essence, a neutral technical method used by model developers around the world, including U.S. companies. By using a high-performing “teacher model” to guide a lighter-weight “student model,” distillation can improve the efficiency of model training and enable more efficient use of human knowledge. It has positive significance for countries seeking to develop their artificial intelligence industries, unlock the potential of the technology, narrow development gaps, and extend its benefits more broadly to developing countries and the general public. China encourages open source development, openness, cooperation, and sharing. It has made open-source models available to companies around the world, including U.S. firms, facilitating the development of related models. Reports on model development by U.S. companies have also disclosed that they make extensive use of distillation from Chinese models. By the United States’ logic, when U.S. companies use distillation it is a “standard industry practice,” but when Chinese companies use it, it becomes an “attack.” This is a blatant double standard. By elevating a standard industry practice into a national security issue, the United States is using opposition to distillation as a pretext for pursuing industrial monopoly. For a long time, certain U.S. artificial intelligence companies have abused their dominant competitive positions by inserting sweeping geographic restrictions and other overbearing provisions into their terms of service, in an attempt to exclude competitors from countries such as China from the global technology ecosystem and monopolize the benefits of artificial intelligence development. This time, U.S. security agencies themselves issued the advisory, tying the interests of particular companies and capital to national security and using the coercive power of the state to interfere in normal commercial activity. This undoubtedly amounts to official endorsement of such overbearing provisions. For some time, U.S. executive and legislative authorities have repeatedly threatened sanctions and other forms of suppression against Chinese companies, with restrictions expanding step by step from controls on chips and blockades on computing power to areas such as access to models. The United States has not hesitated to wield state power to preserve its technological hegemony and monopoly over computing power. At root, this reflects deep anxiety over the prospect of losing its technological dominance. The logic it continues to follow is one in which “only America is allowed to win…”