Artificial intelligence companies in China have formed two new alliances aimed at reducing the industry’s reliance on the United States. President Donald Trump has imposed multiple restrictions to limit China’s access to American tech, and is teeing up a 55% tariff on most Chinese imports.
The Model-Chip Ecosystem Innovation Alliance, launched by the AI model developer StepFun, is a consortium that connects AI companies and chipmakers — including Enflame, Huawei, Biren, and Moore Threads — to develop end-to-end AI technologies entirely within China.
The second industry group is the Shanghai General Chamber of Commerce AI Committee, which aims to accelerate the adoption of AI across Shanghai’s private sector, especially among manufacturing and other mature industries that tend to rely on legacy IT. As of 2023, only 15% of Chinese enterprises had adopted AI, with the manufacturing sector trailing at only 5%.
Members of the Shanghai General Chamber of Commerce AI Committee include model developers StepFun, SenseTime, and MiniMax, as well as chipmakers Metax and Iluvatar CoreX. StepFun has been repeatedly sanctioned by the US since 2019, when the company focused on surveillance technology.
The alliances hope to weather the storm of the US’s chip-related regulations
The US has sought to maintain its sovereignty in AI by restricting China’s access to cutting-edge hardware developed by American companies, like NVIDIA, which is essential for training and deploying powerful AI models. In addition to financial motivations, Washington has cited national security concerns, particularly China’s potential use of AI for military applications.
The US implemented its first set of chip-related export controls back in 2022, prompting NVIDIA to develop modified semiconductors that could circumvent the restrictions and continue sales in China. However, these regulatory loopholes were later closed, significantly affecting NVIDIA’s China-derived revenue.
China and the US have continued to impose reciprocal export restrictions on semiconductors and related technologies. One of Trump’s first export controls during his second term required US firms to obtain licenses before selling chips to China, costing NVIDIA an estimated $5.5 billion in lost sales. He also blacklisted approximately 80 foreign entities — most of them Chinese — suspected of acquiring US technology to support military advancement.
More restrictions were going to come into effect on May 15 as a result of the AI Diffusion Rule, introduced by Trump’s predecessor Joe Biden, but this was rescinded two days prior. The Under Secretary of Commerce for Security deemed it “ill-conceived and counterproductive,” while the Bureau of Industry and Security said a replacement rule is coming that better supports innovation and diplomatic relations.
In the same month, Trump ordered American semiconductor software firms to cease sales of Electronic Design Automation (EDA) tools to Chinese companies, building on restrictions that had already been implemented under Biden. He had also barred Huawei from accessing US EDA tools during his first term.
Trump’s tariffs also push China to shift AI supply chain from US
It’s not just the chip-specific regulations imposed by the US president that have prompted the formation of these alliances, but also his erratic approach to reciprocal tariffs. In April, Trump announced a 10% baseline import tariff on all US trading partners, with additional country-specific tariffs aimed at offsetting bilateral trade deficits.
China’s reciprocal rate was initially set at 34%, but following a series of tit-for-tat measures from the two nations, this had surged to 145% by April 10. In May, both countries reached a temporary agreement to reduce tariffs: China lowered its rate to 10%, while the US cut its rate to 30%, effective for a 90-day period while they negotiated a permanent deal.
Then, last month, they reached a trade truce under which nearly all Chinese imports into the US face a unified 55% tariff.
This measure is expected to raise operational costs for US chipmakers such as NVIDIA, which may pass those costs onto consumers, thereby encouraging Chinese buyers to opt for domestic alternatives. The new alliances can collaborate to ensure China’s AI industry can support such demand.
Earlier this month, NVIDIA CEO Jensen Huang announced plans for a new chip for Chinese clients that is designed for robotics and smart factories and is fully compliant with US regulations.
