Commentary: China is betting on spreading AI faster than the US can contain it

KUALA LUMPUR, SHANGHAI, HEBEI PROVINCE, SINGAPORE, VIETNAM, SELANGOR, WASHINGTON, BEIJING, 27 AUGUST 2026 —

China’s AI Push: Cheaper, Wider Deployment May Outweigh Frontier Race, Analysts Say

Within 48 hours of its July launch, the Chinese AI model Kimi K3 overwhelmed developer Moonshot AI’s computing capacity, underscoring the rapid adoption of homegrown AI tools. By August, both Alibaba and startup Z.ai claimed their new models could rival leading Western systems on key performance benchmarks. Yet the most significant recent development in China’s AI ecosystem may not be another large language model—but a breakthrough in domestic chipmaking equipment.

In late July, a state-backed company in Shanghai reportedly began producing China’s first domestically developed immersion deep-ultraviolet (DUV) lithography machines, a critical step toward reducing reliance on foreign semiconductor manufacturing tools. While these machines are one tier below the most advanced extreme-ultraviolet (EUV) systems made by Dutch firm ASML—which remain off-limits to China due to Dutch export controls—the move signals progress in building a self-sufficient chip supply chain.

The U.S.-China AI rivalry is often framed as a race to develop the most powerful models using the most advanced chips. But analysts argue that China’s strategy prioritizes diffusion—deploying capable AI widely across the economy—over sheer frontier performance. Constrained by U.S. export controls on high-end semiconductors, China has focused on making AI cheaper, more accessible, and embedded in everyday industrial processes.

According to Artificial Analysis, a San Francisco-based research firm, DeepSeek’s V4-Flash model costs just US$0.03 per benchmark test, over 100 times cheaper than Anthropic’s Claude 3.5 Sonnet at US$3.16. This cost advantage accelerates adoption in sectors where cutting-edge performance is less critical than affordability and reliability—such as factory automation, logistics, and agriculture.

A recent example comes from Hebei province, where researchers tested an AI-powered robot on a pig farm that uses computer vision to administer needle-free vaccines. The system achieved a 93.3% success rate, demonstrating how AI can enhance efficiency in mundane but economically vital tasks. Such applications highlight a key insight: AI’s economic impact grows when it becomes invisible—embedded in systems that operate without fanfare.

This diffusion strategy does not mean China has abandoned frontier AI development. Instead, the two approaches reinforce each other. Cheaper, widely available AI models enable broader experimentation, which in turn generates data and use cases that refine future systems. Meanwhile, progress in domestic chipmaking—like the Shanghai lithography machine—helps secure the hardware backbone needed for large-scale deployment.

China’s chip ecosystem is expanding rapidly. In July, CXMT, the country’s leading DRAM memory chip producer, raised US$8.6 billion in a Shanghai IPO and is considering a new fabrication plant in Beijing. Projects in multiple cities could double its production capacity, further reducing dependence on foreign suppliers.

Washington has responded by tightening export controls, not just on advanced chips but also on downstream technologies. In July, the U.S. Federal Communications Commission (FCC) moved to block Chinese-made humanoid and quadruped robots from the American market, targeting firms like Unitree Robotics. The U.S. is also preparing restrictions on optical transceivers, components critical for data centers, where Zhongji Innolight holds an estimated 27% of the global market.

Yet analysts warn that these measures may struggle to contain China’s AI diffusion. Unlike frontier technologies, which rely on a handful of choke points (e.g., GPU designers, semiconductor fabs), diffusion involves a vast and decentralized ecosystem—from industrial robots to logistics software. As Chinese alternatives improve in cost and performance, blocking them becomes increasingly difficult.

For emerging economies in Asia, the appeal of Chinese AI may lie less in benchmark supremacy and more in practical affordability. In Singapore, autonomous vehicles from WeRide and Pony.ai are already ferrying passengers. In Vietnam, Yingshen Intelligence has signed agreements to deploy AI robots in footwear factories. In Malaysia, ZTE is a technology partner behind an AI-enabled warehouse in Selangor.

For Malaysian businesses, this trend presents both opportunities and challenges. Local firms may gain access to lower-cost AI solutions for supply chain optimization, predictive maintenance, and customer service—areas where precision is valued over raw computational power. However, reliance on Chinese AI tools could raise concerns about data sovereignty, interoperability with Western systems, and geopolitical risks.

Stakeholders in Malaysia’s tech sector acknowledge the shift. A spokesperson for MDEC (Malaysia Digital Economy Corporation) noted that the government is actively promoting AI adoption in manufacturing and logistics, with incentives for companies integrating smart technologies. “We see strong interest from both multinational corporations and SMEs in AI-driven efficiency gains,” the spokesperson said. “The key is ensuring these solutions align with our digital infrastructure and regulatory frameworks.”

Industry analysts caution that China’s strategy is still evolving. While domestic DUV lithography machines could reduce reliance on ASML, their precision, throughput, and yield must meet commercial standards—a hurdle that may take years to clear. Similarly, the cost advantages of models like DeepSeek V4-Flash do not guarantee dominance in all sectors, particularly those requiring high-accuracy, real-time processing.

Looking ahead, the competition between the U.S. and China is expanding beyond chips and models. Washington’s technology wall is growing, but China’s strategy of making AI abundant may prove harder to contain. As Diana Choyleva, founder and chief economist of Enodo Economics, argues: “America’s technology wall has not fallen, but China’s AI strategy is spreading faster than the wall can expand.”

For Malaysia and the broader region, the implications are clear. The next phase of AI adoption may not be defined by who publishes the most impressive research paper—but by who can embed AI seamlessly, affordably, and reliably into the machinery of daily economic life. In that race, China’s bet on diffusion could prove decisive.

Related: Diana Choyleva