When I wrote about China’s future of jobs for the World Economic Forum last year, I called it a story about fault lines. That was too gentle (sorry, all Forum writing guideline’s fault!). What is happening is a controlled demolition of one economic model and the planned construction of another, and the world is not watching closely enough.
Start with the demolition. The implicit promise of Chinese higher education: study hard, pass Gaokao, get a university degree, and do a white-collar job is broken. A record 12.7 million graduates enter the job market this year, while youth unemployment sits at 15 to 17%. While the China analysts can read it as a crisis, the more accurate reading is category error: China has too many people with the wrong credentials chasing jobs that no longer exist at scale, while manufacturers compete desperately for skilled technicians they cannot find. A mismatch running at an industrial scale.
Beijing responds with a rewrite. A new eight-level worker system directly links vocational qualifications to wages — making skilled trades financially rewarding in a way that is rare globally. More than 34 million young people are enrolled in technical and vocational programs, with “micro-majors” in robotics, renewable energy, and intelligent manufacturing rolling out nationally.
However, the stigma around vocational education persists, and parents still discourage children from vocational institutes despite the government essentially paying them not to. Beijing can redesign a curriculum overnight. Convincing a Chinese parent that a robotics technician out-earns a finance graduate from Fudan is a longer project. The Asian education system has spent three generations optimizing children for a single exam score, and no five-year plan has a chapter on reprogramming that.
The Great Jobs Rewrite
Last year, China’s new growth sectors looked like ambition dressed in policy language. In 2026, they have an institutional architecture. China does not wait for markets to create jobs. It designates them into existence.
I ran a text analysis of both the 14th and 15th Five-Year Plans to see what actually changed in the language. The shift is not subtle. Artificial Intelligence appears 51 times in the 15th FYP against 6 in the 14th, a 750% increase, and 41 of those 51 appearances are in the front third of the document, meaning AI is now the framing logic of the entire plan.
More striking: terms like embodied intelligence, brain-computer interfaces, and the low-altitude economy did not exist in the 14th FYP vocabulary at all. They become new categories of economic life.
The mechanism is best illustrated by the low-altitude economy: drones, air taxis, and unmanned logistics below 1,000 meters. China has officially recognized “drone flight planner” as a profession. Demand for drone pilots alone could reach one million. The drone economy matters as a proof of concept: central policy designation flows into provincial mandates, which flow into official occupation recognition, which flows into vocational training pipelines. A job category is invented, legitimized, and trained simultaneously. No market signal required.
The long-run version of that same logic is humanoid robots. In Shanghai, 23-year-old Chen Wenhai spends his days in a VR headset teaching a robot to grind coffee and pour water, a job that barely existed three years ago. China shipped 90% of the world’s humanoid robots in 2025. The 15th Five-Year Plan gives embodied intelligence its own dedicated box among the plan’s top ten new industry tracks. Quantum technology, biomanufacturing, 6G, and brain-computer interfaces follow as the next tier, seeded now and due to mature across the 2030s.
My text analysis also shows something less flattering: in both the 14th and 15th FYPs, employment mentions cluster overwhelmingly in the back third of each document - implementation chapters, instead of vision statements. Vocational education appears only in the back of the 15th FYP, absent entirely from the 14th. The plan builds sectors. Jobs are expected to follow. Whether that expectation is warranted is the open question that 12.7 million graduates will answer over the next five years.
What Geopolitics Did to Chinese Jobs
April 2, 2025, Liberation Day was supposed to be the moment the West decoupled from Chinese manufacturing for good. US tariffs on Chinese goods reached 145% within weeks. For a vertiginous few weeks, the decoupling that had been a slow-motion policy aspiration for a decade looked like it might happen all at once. It didn’t.
A year on, Chad Bown of the Peterson Institute puts it plainly: the US has “unwound the supply chains that were easy to unwind, and the ones that are left are the ones that are really, really hard to break away from.” The tariff war sorted Chinese manufacturing rather than dismantling it. It separated low-margin assembly work that could move from the strategically entrenched layer that Beijing had spent years making impossible to relocate.
When China introduced export controls on heavy rare earth elements in April 2025, two days after Liberation Day, carmakers across the US and Europe cut production rates as permanent magnet supplies dried up. This was exactly a demonstration of what “impossible to move” looks like in practice. The IEA notes that China is the leading refiner for 19 of 20 important strategic minerals, with an average market share of 70%.
What the tariff shock did, in effect, was hand Beijing a political forcing function to do faster what the 15th Five-Year Plan was already calling for: shed low-value assembly, concentrate state resources on sectors that cannot be decoupled, and redesign manufacturing around disruption-proofing rather than pure efficiency. My text analysis of both Five-Year Plans makes this visible in the language itself. “Resilience” appears three times more often in the 15th FYP than in the 14th, and in every instance it sits beside supply chain and economic security language. The manufacturing chapter is now designed to withstand pressure.
The plan’s own language confirms it. When I counted every term across both Five-Year Plans, the words that built China’s industrial identity, including “manufacturing,” “supply chain,” “digitalization”, shed mentions. What grew in their place are “Digital-intelligent transformation”, a conceptual merger of digital and AI-driven, and “New quality productive forces”, which is Xi Jinping’s term for an economy that competes on innovation.
Buried in the implementation chapters, two terms appear that would have seemed bureaucratic before April 2025 but now read as urgent: industrial machine tools and industrial software, the specific chokepoints that export controls exposed. Beijing did not need Liberation Day to know these gaps existed. It needed Liberation Day to justify closing them at speed.
The risk aversion is visible and prevalent across China’s workforce, with mid-career workers prioritizing stability over salary and mobility becoming more cautious. The IMF called AI “a tsunami hitting the labour market” at Davos this year, with 60% of jobs in advanced economies facing enhancement, elimination, or transformation. The mechanism is selective: workers who can harness AI gain; those who cannot fall further behind. China’s two bets, build a vocational floor that AI cannot yet replicate, and become the country that manufactures the AI doing the displacing, are both rational responses to that pressure. Neither is guaranteed.
Whether a state can engineer its way out of AI-driven labour polarization faster than markets can adapt, at a scale no one has attempted, is the most important labour market question of this decade. China is the only place running the experiment at full scale. The results will matter everywhere.
Cover photo by Shengpengpeng Cai on Unsplash