China’s AI Chip Project, New Manhattan Project

China is increasingly asserting itself in the global race for artificial intelligence (AI), particularly in semiconductor and AI chip technology. Traditionally, companies such as U.S.-based Nvidia and Taiwan’s TSMC dominate chip manufacturing, while Samsung, Micron, and SK Hynix supply memory chips. Recently, Google has also emerged as a strong competitor thanks to its proprietary TPU chips.

A less widely known but critical player is the Dutch company ASML, often described as the backbone of the world’s most advanced chips. ASML manufactures extreme ultraviolet (EUV) lithography machines, which are essential for modern semiconductor production. The United States has imposed strict restrictions on China to prevent these advanced technologies from reaching its hands, including limits on Nvidia chips and related software exports.

Despite these restrictions, China has persisted. It stockpiled older chips and, in some cases, acquired them through illicit channels. This enabled the country to develop top-tier AI models such as DeepSeek and Qwen. Moreover, Huawei’s recent successes—particularly its Kirin chip line—and the rise of local semiconductor startups have heightened global attention and concern.

According to Reuters, Chinese scientists have developed a prototype lithography machine capable of producing EUV light, though it is not yet capable of manufacturing functional chips. The project, involving former ASML engineers, is being referred to as China’s own “Manhattan Project,” echoing the historic U.S. initiative to build atomic weapons during World War II.

Huawei’s role in the project is pivotal. Despite being blacklisted, the company continues to develop its Kirin chips, Harmony operating system, and Ascend AI chips. Its “SuperPod Connect” integrates thousands of computing units for AI training, rivaling Nvidia’s AI chips in performance.

Chinese engineers have also patented a “Quadruple Patterning” method, enabling the production of 2-nanometre chips using older DUV machines. This represents one of the most advanced methods for AI and high-performance computing chip production today. The main challenge remains how quickly these initiatives can produce functional chips and scale them commercially.

Experts warn that if China succeeds, it could significantly shift the balance of power in AI and semiconductor industries worldwide.