Xi says China must 'overcome' AI chip challenges

President Xi Jinping said China must "overcome" the challenges of developing core AI technologies including high-end chips, state media reported Saturday, as Beijing seeks to become a world leader in the rapidly developing industry.China aims to dominate the artificial intelligence sector, a goal complicated by the trade standoff with Washington that could further deprive Chinese industry of certain key technologies.The world's two leading economies are locked in an escalating tit-for-tat trade battle triggered by US President Donald Trump's new levies on Chinese goods, which have reached 145 percent on many products. Beijing has responded with new 125 percent duties on imports from the United StatesIn this context, Xi called for "continuing to strengthen basic research, focusing our efforts on overcoming challenges in key technologies such as advanced chips and core software, and building an autonomous AI system," according to Xinhua news agency.Xi made the remarks during a quarterly meeting of the Politburo, the inner circle of China's top leadersSince the launch of ChatGPT in November 2022, generative AI models have proliferated in the United States and China.Chinese startup DeepSeek shook up the AI world in January with its R1 chatbot, matching the performance of its US competitors at a lower cost'Promote self-reliance'But Xi acknowledged Friday that the Chinese industry still had "gaps". It was "essential" to "promote self-reliance" in the field, he added.Getting Up at 5 AM Every Day: I tried it for a week and by day five...Getting Up at 5 AM Every Day: I tried it for a week and by day five... Political support was essential to achieve this, Xi stressed, citing in particular "a combination of policies such as intellectual property rights, taxation, public procurement, and the opening up of infrastructure".Under Trump and his predecessor Joe Biden, Washington has banned or restricted exports to China of advanced processors which are known for helping develop high-end AI models.The Trump administration has imposed new licensing requirements to export to China some chips used in AI, which US firms Nvidia and AMD have said will hit them hard.Nvidia CEO Jensen Huang visited Beijing this month and said he was "willing to continue to plough deeply into the Chinese market and play a positive role in promoting US-China trade cooperation", Xinhua reported.Washington's controls are officially imposed in order to prevent China developing military technologies, but they also allow the United States to maintain its competitive edge.China's AI ambitions have prompted concern in numerous countries worried about the handling of personal data, particularly the possibility that such information could be transferred to Chinese authorities

HISTORY

Founding and early years (2016–2023)

edit

In February 2016, High-Flyer was co-founded by AI enthusiast Liang Wenfeng, who had been trading since the 2007–2008 financial crisis while attending Zhejiang University.The company began stock trading using a GPU-dependent deep learning model on 21 October 2016; before then, it had used CPU-based linear models. By the end of 2017, most of its trading was driven by AI.

Liang established High-Flyer as a hedge fund focused on developing and using AI trading algorithms, and by 2021 the firm was using AI exclusively,[25] often using Nvidia chips.

In 2019, the company began constructing its first computing cluster, Fire-Flyer, at a cost of 200 million yuan; it contained 1,100 GPUs interconnected at 200 Gbit/s and was retired after 1.5 years in operation.

By 2021, Liang had started buying large quantities of Nvidia GPUs for an AI project,[26] reportedly obtaining 10,000 Nvidia A100 GPUs before the United States restricted chip sales to China.Computing cluster Fire-Flyer 2 began construction in 2021 with a budget of 1 billion yuan.

It was reported that in 2022, Fire-Flyer 2's capacity had been used at over 96%, totaling 56.74 million GPU hours. 27% was used to support scientific computing outside the company.

During 2022, Fire-Flyer 2 had 5000 PCIe A100 GPUs in 625 nodes, each containing 8 GPUs. At the time, it exclusively used PCIe instead of the DGX version of A100, since at the time the models it trained could fit within a single 40 GB GPU VRAM and so there was no need for the higher bandwidth of DGX (i.e., it required only data parallelism but not model parallelism). Later, it incorporated NVLinks and NCCL (Nvidia Collective Communications Library) to train larger models that required model parallelism.

On 14th April High-Flyer announced the launch of an artificial general intelligence (AGI) research lab, stating that the new lab would focus on developing AI tools unrelated to the firm's financial business. Two months later, on 17 July 2023, that lab was spun off into an independent company, DeepSeek, with High-Flyer as its principal investor and backer.

On 20 January 2025, DeepSeek launched the DeepSeek chatbot—based on the DeepSeek-R1 mod