TL;DR:
- Shanghai aims to build a world-class AI hub in Pudong, attracting talent and investments.
- The impending US ban on exports of Nvidia’s A800 and H800 AI chips to China poses a hardware problem for Shanghai.
- Chinese AI firms currently rely on chips manufactured by Taiwan’s TSMC, which can be affected by US restrictions.
- Shanghai government implements measures to attract private capital for AI infrastructure investment.
- Huawei’s Ascend AI chips and Kunpeng CPUs have achieved breakthroughs, but US sanctions raise concerns about chip supplies.
- TSMC’s ability to continue producing AI chips for China may not compensate for the lack of tools to develop advanced chips.
- Nvidia’s investment in Taiwan and its dominance in the AI chip market further complicates Shanghai’s AI ambitions.
- The World AI Conference gains attention, reflecting the growing popularity of AI.
- Chinese market may face challenges in advancing AI capabilities due to chip conflicts and potential lag behind Nvidia.
Main AI News:
Shanghai, the bustling commercial city of China, has set its sights on establishing a world-class hub for artificial intelligence (AI) in Pudong. At the World AI Conference, key officials from Shanghai expressed their commitment to attracting top AI talent, fostering research and development, and driving the adoption of AI technologies in various sectors, such as advanced manufacturing, urban management, and robotics. However, Shanghai’s ambitious plans could encounter a major setback due to an impending chip war.
Starting this month, Shanghai may face a hardware challenge as the Biden administration prepares to ban exports of Nvidia’s A800 and H800 AI chips to China. Additionally, US investment funds are expected to face restrictions when it comes to investing in China’s high-technology sector. This decision by the US government has the potential to disrupt Shanghai’s efforts to build its AI capabilities.
While China has made significant strides in developing its own AI chips, such as those by Huawei Technologies and Cambricon Technologies, these chips rely on manufacturing processes carried out by Taiwan’s TSMC. Consequently, the prospective US curbs could impact the production of these chips, creating further obstacles for Shanghai’s AI ambitions.
Despite these challenges, the Shanghai government has taken proactive measures to attract private capital for investing in new infrastructure. It has extended the implementation period of its policy to offer an interest subsidy to private firms that construct AI facilities. The government has identified the integrated circuit, biomedical, and AI sectors as the most crucial industries that will drive Shanghai’s ascent as an AI powerhouse.
Presently, Chinese AI firms can still purchase A800 and H800 chips, introduced by Nvidia last November, as alternatives to the A100 and H100 chips, which were added to a US export control list in August. Reports indicate that the performance of A800 chips has reached approximately 70% of the A100’s capabilities. However, media speculations suggest that the US Commerce Department may soon ban the export of A800 and H800 chips to China, a move that could further complicate Shanghai’s AI plans.
Ken Hu, rotating chairman of Huawei, emphasized the company’s dedication to AI chip development and industrial applications. Huawei’s Ascend AI chips and Kunpeng central processing units (CPUs) have already achieved significant breakthroughs. According to a research report by CITIC Securities, over 20 Chinese cities have integrated Ascend chips into their AI facilities, and Huawei currently commands a 79% share in China’s AI computing center market. Chinese chips such as Ascend 910 and Cambricon’s Siyuan 370, both manufactured in Taiwan using TSMC’s advanced 7nm technology, have demonstrated commendable performance levels.
The intricate dynamics surrounding the production and availability of AI chips have raised questions about the impact of US sanctions on Huawei’s chip supplies. An article by a Hainan-based columnist highlighted that Huawei’s Ascend 910 chip remains unaffected by the sanctions due to its usage of Huawei’s self-developed Da Vinci architecture, in contrast to Kirin chips that relied on ARM architecture from the United Kingdom, as well as technologies from Cadence and Synopsys in the United States. TSMC, being a Taiwanese foundry, enjoys some latitude to maintain its business relationship with Huawei, despite the US restrictions.
However, a writer from Fujian argues that even if TSMC can continue producing AI chips for China, Chinese companies lack the necessary tools to develop more advanced AI chips. This predicament may leave them trailing behind Nvidia, a dominant player in the AI chip market. As if to underscore its commitment to AI research, Nvidia recently announced plans to invest a substantial sum of TWD24.3 billion (US$790 million) and recruit 1,000 personnel for its new AI research center in Taiwan.
The World AI Conference has gained significant attention, with this year’s event drawing even more interest. The sector’s popularity soared after OpenAI, backed by Microsoft, unveiled ChatGPT 3.5, an AI chatbot, last November. Share prices of Nvidia, which enjoys over 80% of the AI chip market, have surged by a staggering 196% this year.
Prominent figures, such as Elon Musk, have acknowledged China’s potential in AI. Musk, the founder of Tesla, commended the abundance of smart and talented individuals in China, stating that the country possesses the capability to excel not only in various sectors of the economy but also in AI. Shanghai Party Secretary Chen Jining echoed this sentiment, emphasizing the city’s commitment to fostering an open collaboration platform and creating an innovative ecosystem.
Shanghai Mayor Gong Zheng stressed the importance of embracing new technological revolutions like AI to uplift the city’s industries and transform it into a socialist modern international metropolis with global influence. Gong emphasized Shanghai’s goal of establishing itself as a leading player in the AI industry, poised for success on the global stage.
Conclusion:
The chip conflict involving the US ban on Nvidia’s AI chips poses significant challenges to Shanghai’s goal of becoming a world-class AI hub. The reliance on chips manufactured by TSMC, coupled with limited tools for developing advanced AI chips, may hinder the market’s progress. Furthermore, the investments and dominance of Nvidia in the AI chip market raise concerns about China’s ability to compete effectively in the AI industry. Efforts to attract private capital and foster collaboration will be crucial in mitigating these challenges and maintaining Shanghai’s position in the global AI landscape.
