South Korean media: China's self-sufficiency rate of AI chips will exceed 80% in three years!
On June 21, the Chosun Ilbo, a South Korean media outlet, published an article stating that according to predictions, China's self-sufficiency rate of AI chips will exceed 80% within three years. The strict U.S. semiconductor export control has stimulated the development of China's semiconductor ecosystem. China is turning its blueprint for "technological independence" into reality by building data centers driven by self-developed AI chips and launching humanoid robots equipped with domestically produced GPUs.
In its latest report titled "China Artificial Intelligence: The Awakening of the Sleeping Giant," Morgan Stanley predicts that China's self-sufficiency rate of AI chips will soar from 34% last year to 82% by 2027. Morgan Stanley analyzes that if not for U.S. sanctions, China might have continued to rely on foreign chips such as those from NVIDIA, but faced with external pressure, China has fully embarked on a strategy of complete self-reliance and is building a self-sufficient ecosystem at a faster pace than expected.
The biggest driver of explosive growth is "talent." China has approximately half of the world's AI researchers. Based on this, China has formulated a strategy to build a systematic autonomous ecosystem. In the face of top-tier chip supply disruptions, China has obtained downgraded versions of NVIDIA's mainstay AI chips and focused on improving software technology to maximize performance by stacking multiple chips. The open-source model DeepSeek-R1 has achieved artificial intelligence at a cost far below that of GPT-4, becoming a symbol of China's solution to overcome hardware disadvantages through a software strategy.
China has invested huge R&D funds in its AI ecosystem and supported domestic enterprises through "priority procurement by public institutions," leveraging its vast domestic market. Huawei has challenged NVIDIA by showcasing a supercluster composed of thousands of self-developed AI chips, "Ascend 910," which supports an AI model training system with parameters exceeding one trillion. SMIC, China's largest foundry, has broken through the barrier of 7nm process technology, making Huawei's independently designed AI chips a reality. It is assessed that China has formed a virtuous cycle of "using alternative solutions → opening up the domestic market → accelerating independent research and development → mass production."
The independence of the "AI brain" is reshaping the future industrial landscape, transcending mere chip production. The humanoid robot market is a typical example, expected to grow to a scale of $5 trillion by 2050. Morgan Stanley predicts that凭借自主采购 AI 芯片 and core components, China will capture 30% of the global humanoid robot supply. Some forecasts even suggest that if humanoid robots are produced using China's supply chain, their manufacturing costs would be only one-third of those using the global supply chain.
Enterprises leading China's AI rise span various industries. Morgan Stanley states that not only Huawei and SMIC, but also Alibaba and Tencent, which are at the forefront of AI platform competition, as well as Ecovacs, an appliance company specializing in robotic vacuum cleaners, and Horizon Robotics, a provider of autonomous driving solutions, are accelerating AI innovation. With this powerful ecosystem, it is expected that China's core AI industry will grow to 1 trillion RMB by 2030.
China's example proves that the winning formula for the battle over AI dominance is changing. Unlike the traditional view that "having the highest specification semiconductor chips can directly dominate the market," the "power of the ecosystem," which effectively combines slightly insufficient hardware with software and systems to create value, has become a key variable in competition.
Source: https://www.toutiao.com/article/1835505666962460/
Disclaimer: This article solely represents the author's viewpoint.