China to begin AI-chip production as it skirts U.S. sanctions

Chinese scientists are considering building large facilities for designing semiconductor chips to create artificial intelligence (AI) systems. This is reported by South China Morning Post.

As part of the initiative, a particle accelerator 100-150 metres in diameter is planned, which will be used in the production of AI processors. Researchers from Tsinghua University are already in talks with authorities in Xungan Province about site selection.

Particle accelerators will replace the role of lithography machines at stages of producing semiconductors for high-level AI systems. The light beam produced by the device is extreme ultraviolet (EUV), used in creating chips smaller than 7 nm.

“One of the potential applications of our research is a light source for future EUV lithography machines. I think that’s why the international community has taken a close interest in them,” said the project leader Professor Tan Chuansian.

The Chinese research is based on a new luminescence mechanism called Sustainable Microgrouping (SSMB). The technology is an “ideal” light source and has higher average power and efficiency at low cost.

The scientists’ report states that indigenous technologies will help China circumvent existing U.S. sanctions.

At present, Dutch Advanced Semiconductor Materials Lithography is the only company that owns the technology to produce 2 nm chips. In January the United States banned the firm from selling its products on the Chinese market; later the restrictions also affected Nvidia.

“The path to developing EUV lithography machines of our own remains long, but light sources based on SSMB offer us an alternative to sanctioned technology,” the scientists stressed.

On 15 August, the AI law came into effect in the PRC, which requires prior approval from the Cyberspace Administration of China (CAC) before launching relevant products.

Earlier, the CAC launched a campaign against fake AI news. The agency targeted methods of creating content that evoke emotions in audiences to attract attention and traffic for “nefarious purposes”.

In January, a law prohibiting the creation of deepfakes without the subject’s consent or to harm national interests came into force in China.

In April, the CAC published a draft regulation aimed at regulating generative artificial intelligence algorithms.

As of 30 August, local companies — Baidu, Baichuan Intelligent Technology, SenseTime and Zhipu AI — launched their own AI-based chatbots after government approval.