China's Orchestrated Push to Build an AI Talent Superpower

In the span of a single policy cycle, China assembled the most ambitious state-directed pipeline for artificial-intelligence talent anywhere in the world, and it did so by treating the challenge not as a matter of funding schools but of engineering a system that feeds itself. A national plan set the objective; universities were restructured to meet it; technology corporations were invited inside the campus to accelerate the work; and the capabilities that resulted were turned back on the classroom that produced them. Each stage reinforces the next in a deliberate loop — policy builds institutional capacity, institutional capacity attracts corporate investment, and corporate technology reshapes how students are taught. The clearest measure of how far that loop has closed is that the same digital-human technology the pipeline was built to master is now being used to lecture the very students moving through it. The bet underlying the whole construction is that the country controlling the supply of AI talent will control the industries of the coming decades, and the speed of the build-out reflects how seriously that wager is being pressed.

The starting gun was the New Generation Artificial Intelligence Development Plan, issued by the State Council in July 2017, which declared AI the defining strategic technology of international competition and laid out a three-step roadmap: keep pace with the leading AI nations by 2020, achieve major breakthroughs by 2025, and become the world's primary innovation center by 2030. For the talent question specifically, the plan mandated AI courses in primary and secondary schools, the expansion of graduate enrollment, and the recruitment of global experts to form what it called a "talent highland," distinguishing between "vertical" specialists who master AI from theory to product and "horizontal" generalists who can apply it across economics, law, and governance. Less than a year later the Ministry of Education translated those directives into operational terms with its Higher Education Artificial Intelligence Innovation Action Plan, which instructed universities to build AI as a first-level academic discipline and to develop an interdisciplinary "AI+X" training model spanning computer science, mathematics, biology, psychology, and law. Notably, that same 2018 plan singled out smart education as a priority application domain, marking the classroom as a target for AI deployment years before the technology was ready to occupy it. What had been a strategic ambition became, almost immediately, a set of institutional instructions.

The institutional response was startling in its velocity. Before 2019 there was no formal AI undergraduate major in the Chinese system at all; when the Ministry of Education approved the first cohort of thirty-five universities to offer one that year, it opened a floodgate. A further 180 institutions gained approval the next year, the cumulative total passed 500 by the end of 2023, and by 2025 more than 626 institutions offered AI-related degrees. The leading research universities did not wait for the formal designation. Tsinghua elevated its earlier AI research institute into a full College of Artificial Intelligence in 2024 under the Turing laureate Yao Qizhi, declaring an ambition to rival MIT and Stanford within five years. Peking University consolidated its scattered units into a School of Intelligence pursuing general artificial intelligence as its long-term goal, while Renmin University took a humanities-inflected route with its Gaoling School, seeded by a 300-million-yuan private donation and ranked first globally in information retrieval for three consecutive years. Zhejiang University, which had been training AI graduate students since 1978, formalized decades of accumulated depth into a standalone AI school of its own in 2024 — a reminder that the sudden explosion of majors rested, at the strongest institutions, on foundations laid long before the national plan existed. Zero majors to more than six hundred institutions in roughly six years is the kind of figure that only centralized mobilization can produce.

The feature that most distinguishes China's approach, however, is also the least visible: the structural fusion of corporate AI research with university operations. Chinese policymakers describe it through the "triple helix" of university, industry, and government, and its most vivid expression is Zhejiang Lab, founded in 2017 as a joint venture of the provincial government, Zhejiang University, and Alibaba, funded with thirty billion yuan and housed on a dedicated campus. Baidu anchors the government end of the same arrangement, leading the country's first national-level AI engineering laboratory and, through its open-source deep-learning platform, partnering with more than five hundred universities and helping over two hundred of them stand up credit-bearing AI courses. Tencent, Alibaba, and SenseTime have layered joint laboratories across the top institutions, embedding corporate researchers, corporate tooling, and corporate funding directly into the machinery of graduate training. The result blurs the boundaries between public research, private innovation, and state direction in a way that has no clean parallel in Western higher education, and it means the pipeline does not simply graduate students into industry — industry is already inside the pipeline.

By 2025 the system was reaching down past the university into childhood. Xi Jinping used a Politburo study session that spring to call for AI education across every stage of schooling, and the State Council followed with an "AI+" action demanding what it termed super-conventional cultivation of leading talent. Beijing's municipal education commission operationalized the mandate in a three-year plan requiring a minimum of eight hours of AI instruction per year for every student from elementary through high school, the training of a hundred expert and a thousand backbone teachers — and, in a clause that reveals where the whole enterprise is heading, the use of virtual digital-human technology combined with large models to create high-level AI teachers. The classroom was no longer only the place where AI talent was made; it was becoming a place where AI itself would do the teaching.

That is where the through-line stops being figurative. At Zhejiang University, the School of Marxism built a system that delivers lectures through digital-human instructors modeled on real professors, serving more than five thousand first-year students and standing as the first large-scale domain-specific AI model deployed in the country's ideological curriculum. Individual professors have begun cloning themselves outright: a recently retired finance scholar recorded a photorealistic avatar that lectures from text, converts its own delivery into English and French, and answers questions in real time through a connected language model, while a computer-architecture instructor built a digital teaching assistant from recordings of his past classes. The pattern extends well beyond lectures. A digital avatar of one of China's most prominent cardiologists, powered by a cardiovascular-specific model and speaking in his own regional accent, correctly diagnosed a rare condition during its public demonstration. A traditionally styled museum-guide character loaded with knowledge of more than a hundred thousand cultural relics now works in over thirty museums and converses in scores of languages, and a primary school in Nanjing has fielded a digital teaching assistant that fields student questions across subjects. Each of these is still, in its way, a demonstration — but the demonstrations are multiplying, and the policy language pushing them into teacher-shortage regions is explicit.

The commercial ground beneath these experiments is expanding fast enough to suggest they will not stay experimental for long. One market assessment put the value of China's AI digital-human sector at 41.2 billion yuan in 2024, an increase of more than eighty-five percent in a single year, and projected it would reach 250.5 billion yuan by the end of the decade. The regulatory apparatus has moved in step rather than trailing behind, with the education ministry issuing classroom-specific guidance that openly endorses generative digital-human teachers for regions short of qualified staff, naming pilot schools and drawing provincial departments into coordinated deployment. What emerges from all of this is not a single program but a layered system whose components — strategic plans, restructured universities, embedded corporate laboratories, regional industrial bases, and a parallel body of rules — are engineered to reinforce one another. The most consequential piece may remain the quietest: the fusion of corporate research with university operation that blurs the line between who funds, who trains, and who directs. And the digital human is the point at which the whole system turns recursive, because the technology the pipeline exists to produce has begun, quietly, to produce the next cohort in turn.