Experts: China Trails the US in AI Development but Could Quickly Narrow the Gap

Standing before an eager audience in Hangzhou, a notable tech hub in eastern China, Eddie Wu, the CEO of Alibaba, delivered a significant announcement that caught many by surprise. “We are on the brink of an AI-driven intelligent revolution,” Wu stated during a developer conference held in September. “The emergence of Artificial General Intelligence (AGI) will not merely enhance human intelligence, but it will also unlock previously unimagined potentials, paving the way for the subsequent arrival of Artificial Superintelligence (ASI).”

Wu elaborated on the transformative possibilities of ASI, predicting it could give rise to a cohort of “super scientists” and “full-stack super engineers,” capable of addressing complex, unsolved scientific and engineering challenges at unprecedented velocities.

In a further show of commitment to AI advancement, Wu announced an ambitious investment plan of 380 billion yuan (approximately £40 billion) earmarked for AI infrastructure development over the next three years. This announcement had an immediate impact, causing Alibaba stocks to surge to their highest levels in almost four years.

Wu’s remarks, which delved into existential themes typically associated with prominent Western tech executives like OpenAI’s Sam Altman and DeepMind’s Demis Hassabis, piqued the curiosity of many analysts. “Wu’s discussion of ASI marks a watershed moment,” noted tech writer Afra Wang in her AI-focused newsletter, Concurrent. “Prominent Chinese corporations are starting to articulate expansive visions that echo the prophetic aesthetics of the future.”

AGI, often referred to as a theoretical scenario where highly autonomous systems can perform tasks typically managed by humans, has emerged as a central focus for tech giants in the U.S. like OpenAI and DeepMind, igniting a fierce competition to be the first to achieve this breakthrough. In a recent US Senate hearing on AI, Microsoft’s president Brad Smith reflected on this rivalry by asserting that the “race for international influence between the U.S. and China will likely be determined by who makes it across this frontier first.”

Within Washington, these concerns have taken root, leading authorities to influential decisions. The U.S.-China Economic and Security Review Commission has even recommended that Congress initiate a program akin to the Manhattan Project to advance towards establishing an AGI capability—an endeavor reminiscent of the WWII-era project that ultimately birthed nuclear weaponry.

Contrarily, many observers within China interpreted Wu’s address as a showcase of Alibaba’s unique vision rather than a reflection of the broader Chinese AI landscape.

“While there are indeed research initiatives working towards AGI in China, the majority of AI enterprises are focused on creating practical applications,” remarked Ya-Qin Zhang, the dean of Tsinghua University’s Institute for AI Industry Research and former CEO of Baidu.

China’s national AI policy is characterized by its pragmatic focus on harnessing AI for immediate, real-world applications rather than pushing the boundaries of cutting-edge research, largely influenced by constraints like limited computing capabilities.

August saw the unveiling of the Chinese government’s much-anticipated “AI+ strategy,” a policy manifesto outlining how AI could propel the nation’s development ambitions—aiming to enhance medical diagnostics and streamline supply chains—yet conspicuously lacking any mention of AGI.

“The Chinese government is strongly oriented towards leveraging AI for immediate benefits in sectors like the economy, society, and defense,” explained Julian Gewirtz, a former senior director at the White House National Security Council. “While China’s ambition may be to ‘catch up and surpass’ the U.S., it is important not to assume that the Communist Party views AGI as an urgent priority.”

“Official narratives provide little clear acknowledgment of the AGI goal,” added Selina Xu, a tech analyst specializing in China. She observed that President Xi Jinping has historically prioritized tangible economic aspects over more ethereal innovations.

“This narrative diverges significantly from the AGI race envisioned by many in Washington,” Xu noted.

A significant factor affecting this strategy is the pervasive U.S. sanctions that hinder Chinese firms from acquiring advanced semiconductors, essential for pioneering AI research.

In an effort to stifle China’s AI advancement, Washington has prohibited the sale of high-tech microchips to China. In response, chipmaker Nvidia has since rolled out less advanced semiconductors tailored for the Chinese market. However, Beijing has reportedly advised customs agents against allowing the import of these chips as it seeks to reduce dependence on foreign technology.

China’s sentiment encapsulated in the saying “necessity is the mother of invention” suggests a strong belief that U.S. restrictions will only serve to ignite more innovative solutions. A noteworthy example is DeepSeek, whose founder, Liang Wenfeng, shared Wu’s interest in AGI.

Nonetheless, until Chinese companies can produce cutting-edge semiconductors independently and on a large scale, many are inclined to optimize the existing hardware they possess for AI applications rather than delving into the complex realm of AGI.

Another significant factor driving the U.S.-China tech competition is the availability of data centers and the resources to fuel them. In November, Nvidia’s CEO Jensen Huang projected that China might “win the AI race,” attributable in part to energy subsidies provided for data centers.

These energy subsidies were reportedly enacted after domestic semiconductor production caused an uptick in electricity costs, prompting complaints from Chinese tech firms. Indications that the Chinese government is resolute in its mission to eliminate reliance on imported technology are evident, especially when it comes to state-funded centers being mandated to use domestic chips exclusively.

Such policies could diminish Nvidia’s advantage within China while providing a boost to its domestic competitors, such as Huawei.

It has been reported that since 2021, China has invested approximately $100 billion in AI data center support.

Yet, signs indicate that this exuberance may be excessive. A recent study from the China Academy of Information and Communications Technology highlighted that the overall utilization rate of AI data centers in China stands at a mere 32%.

In an op-ed featured in China Economic Weekly, Rao Shaoyang, the director at the China Telecom Research Institute, warned that in certain regions, the expansion of the computing power sector mirrors the challenges faced by China’s troubled property market: constructions precede the identification of demand. He urged a cautious approach to “blindly building intelligent computing centers” and emphasized the need for aligning new developments with actual local demand.

Despite a surplus of general computing power, experts maintain that China’s chip capabilities remain insufficient for tackling advanced AGI research. However, many analysts suggest that the current climate is highly dynamic and could shift rapidly.

“The existing landscape is fluid, and Xi Jinping has explicitly vocalized intentions to lead in AI development,” stated Gewirtz. “Thus, the way China interprets that goal at this particular moment may not reflect how it will approach it in a year.”

Additional research by Lillian Yang