On a Monday morning in late September 2026, in a fourth-floor lab on the Tsinghua University campus in north-west Beijing, a young Chinese postgraduate in a white coat pipetted a clear solution into a small vial. Across the bench, a colleague adjusted the fine focus of a microscope. At the far end, an older researcher in a pale blue apron wrote on a clipboard. There was nothing unusual about the morning. The number of mornings like it, in labs like this one, across Chinese universities, is what is unusual. The most-cited recent figure for Chinese output in science, technology, engineering and mathematics, 5 million graduates a year, was restated by the State Council Information Office in May 2026 as a rounded current reality, not as a target. The figure is now embedded in the way Chinese ministries, foreign universities and Western policy shops talk about the country. A second figure sits underneath it: 77,000 STEM PhDs a year, the largest single national doctorates total in the world, and almost twice the figure for the United States. Both numbers are large; they are also shaped by choices of definition that make the comparison with other countries harder than it looks. This dispatch walks through what the published counts actually count, where they come from, what they leave out, and what they do and do not say about the shape of the global research workforce.
Five million, or thereabouts
The most-cited recent figure for Chinese STEM output is the 4.7-million-per-year total that the World Economic Forum attributed to China in 2016. The same figure is still quoted in 2026 by aggregators that depend on the WEF methodology, and it is broadly consistent with the State Council Information Office’s 11 May 2026 release, which says that “China now produces more than 5 million graduates in science, technology, engineering and mathematics (STEM) disciplines each year.” The State Council figure is the larger of the two and the more recent. The 0.3-million gap between 4.7 million and 5 million is well inside the methodological variation that the WEF itself flags in any cross-country comparison.
The 2026 WorldRankd dataset reports 3.57 million Chinese STEM graduates for the most recent year, materially lower than the State Council figure. The discrepancy is not a data error in either source. It reflects a narrower field definition. WorldRankd follows the OECD‘s Fields of Education and Training classification, which excludes agriculture, medicine and some engineering sub-fields that the State Council includes. Both numbers are correct on their own terms. They are measuring two different things and calling both of them “STEM graduates”.
This is a recurring problem in the cross-country comparison. UNESCO, which is the closest the world has to a global tertiary-education statistics body, did not publish data for China in the most recent comparative STEM tables because Chinese reporting follows a different taxonomy. The World Economic Forum workaround, the OECD workaround and the US National Science Foundation workaround each handle the gap differently. The NSF’s most recent count put Chinese science and engineering graduates at about 1.6 million in 2014, a much smaller figure than the WEF and State Council totals, because the NSF applies the Western definition of “science and engineering” rather than the Chinese one.
The cleanest summary: depending on which definition is in play, China is producing between 3.5 million and 5 million STEM graduates per year, and the figure that matters most for global STEM workforce planning is closer to the higher end of that range, because China’s domestic definition is the one its own employers use. Anyone quoting a single cross-country number without saying which definition they have used is reading from the wrong end of the telescope.
The seventy-seven thousand
The doctorate count is more comparable across countries than the all-degree count, because PhD programmes are more standardised and field definitions line up more closely. The most-cited recent figure for Chinese STEM PhD output is the 77,000-per-year projection in a 2024 report by the Center for Security and Emerging Technology (CSET) at Georgetown University, which the report says is supported by current enrolment patterns. The same 77,000 figure is cited by independent Chinese-talent analyses and by working scientists discussing US-China STEM competition on professional networks.
For comparison, the United States produces around 40,000 STEM PhDs per year, but 42 per cent of those go to temporary-visa international students who then return to their home countries after completing their PhDs. Strip out the international contingent and the US domestic STEM PhD output is about 23,000 per year. That puts the Chinese-to-American domestic-PhD ratio at roughly 3.3 to 1.
The Indian figure is harder to pin down. India’s domestic PhD production has been growing fast, and the most recent CSET-adjacent estimates put Indian STEM PhD output in the 25,000 to 30,000-per-year range, well below China’s. The European Union’s combined output is around 40,000 to 50,000 STEM PhDs per year, distributed across 27 member states. Germany alone produces around 12,000 to 15,000 STEM PhDs per year, more than any single EU member state but well below the China total.
The doctorate component is the part of the STEM pipeline that most directly drives research output. A country that graduates 77,000 STEM PhDs per year is, all else equal, the country whose research output grows the fastest. The Nature Index, which tracks authorship in 82 high-quality natural-science journals, has tracked this trajectory. The Index ranked China second to the United States in 2019; by 2024 to 2025 it ranked Chinese institutions ahead of American ones by share of authorship, with eight of the top ten contributing institutions now Chinese. The PhD pipeline of 2015 to 2020 became the research output of 2020 to 2025.
The twenty-five years behind the number
The tenfold growth in Chinese STEM output over the past quarter century is not a single policy choice. It is the product of three overlapping decisions taken at different times, each of which became irreversible once made.
The first decision was the 1999 expansion of university admissions, in which the central government more than doubled the undergraduate intake over five years. The expansion was originally conceived as a response to the Asian financial crisis and to the coming demographic peak in college-age population, but it had a second effect that nobody at the time labelled: it produced a generation of students for whom the bachelor’s degree was the floor rather than the ceiling, and a much larger pool from which the later PhD expansion could draw.
The second decision was the long-running build-out of the Chinese research university system, in the form of Project 211 (1995), Project 985 (1998) and the newer Double First-Class programme. These gave selected universities ring-fenced funding for research infrastructure, faculty hiring and graduate programmes. The 39 “985” universities and the 115 “211” universities, between them, account for the great majority of Chinese STEM PhD output and almost all of the output that is internationally visible. The build-out is now mature: most of the 985 universities have research groups with the kit and the postdoc pipeline to compete with good Western departments.
The third decision was the post-2009 expansion of the professional master’s degree, the zhuanye shuoshi, which sits between the Western master’s and the first year of a PhD in terms of research output. The professional master’s degree was originally created in the 1990s and expanded sharply after 2009 to serve the engineering and technology workforce. It now accounts for the great majority of new master’s enrolments in engineering, and is one of the reasons the all-degree STEM total has grown faster than the PhD total. The professional master’s is not a Western degree; the international comparisons treat it inconsistently.
On top of these three decisions sits a fourth, which is the policy weight given to science and technology as a national priority. The 2006 National Medium- and Long-Term Programme for Science and Technology Development, which ran from 2006 to 2020 and has been renewed, set quantitative targets for R&D spending, for researcher density and for the share of GDP going to basic research. Those targets have largely been met. Chinese gross domestic R&D spending crossed 3.5 per cent of GDP in 2023, ahead of the European Union and approaching the United States. The PhD pipeline and the R&D spending move together. The 77,000-PhD figure is the human-capital expression of the R&D-spending trajectory of the past decade.
Where the data comes from
It is worth saying where the published numbers come from before saying what they show. The State Council Information Office’s 2026 release is an English-language restatement of figures published annually by the Chinese Ministry of Education in its Educational Statistics Yearbook. The yearbook is the canonical source; the State Council release is a translation for foreign readers. Both are based on a national roll-up of university-by-university degree conferrals, which the Ministry of Education collects from the higher-education student information system.
The CSET 77,000 figure, by contrast, is a projection rather than a direct count. CSET, part of Georgetown University, builds the figure from current PhD enrolment data plus an estimate of time-to-degree drawn from the published completion rates. The same methodology has been used to project other countries’ STEM PhD output and is reasonably reliable for one to three years out. The 40,000 US figure, on the other hand, comes from the US National Science Foundation’s Survey of Earned Doctorates, which is a near-census of every research doctorate conferred by a US university in a given year. The 42 per cent international-student share is also from the same survey.
The European Union figures, the German figures and the OECD aggregations come from the OECD’s Education at a Glance publication, which is itself a roll-up of national submissions to the joint UNESCO-OECD-Eurostat data collection (the UOE). The 3.57-million WorldRankd figure is taken from the same UOE data, narrowed to the OECD field classification. The Asian Times tenfold-growth figure is editorial rather than a single-source count; the underlying trajectory is well established in the Chinese statistical yearbooks.
Two practical consequences. First, the State Council 5-million figure and the CSET 77,000 figure are not directly comparable: the first is a count of conferred degrees in a year, the second is a projection of PhDs in a one-to-three-year window. Both are useful, but they answer slightly different questions. Second, the Asia Times tenfold-growth claim is a real trajectory rather than a single-year figure, and the way it is presented in some Western commentary collapses a 25-year trend into a one-year comparison. The trajectory is the result; the year is just the most recent data point on the curve.
What counts as a STEM graduate
A useful way to read the headline number is to ask what counts as a STEM graduate in the Chinese system. The State Council’s 5-million-plus figure includes students in:
- Science: the natural sciences, mathematics and applied mathematics, statistics.
- Engineering: all branches, including civil, mechanical, electrical, chemical, computer, materials, biomedical, nuclear and aerospace engineering.
- Agriculture: agronomy, horticulture, veterinary science, forestry and fisheries.
- Medicine: clinical medicine, public health, traditional Chinese medicine, pharmacy, nursing and stomatology.
It excludes humanities, social sciences, education, the arts, management and law. The WEF and OECD counts are narrower. The NSF Western count is narrower still. The 3.57-million WorldRankd figure excludes medicine and most of agriculture, which alone removes well over a million graduates from the count.
A second caveat is the inclusion of two- and three-year junior college diplomas, the dazhuan, which UNESCO does not classify as tertiary degrees at the same level as four-year bachelors. Chinese reporting includes these in many STEM totals, while international comparisons often strip them out. A dazhuan-trained software engineer in Shenzhen is a real working engineer, but the international comparison counts them differently. The gap is not fictitious, but it is real.
A third caveat is the master’s degree. The 60-per-cent figure for STEM master’s enrolments in 2023 includes both academic master’s and the much larger professional master’s category that was expanded after 2009. Professional master’s are two- to three-year applied-research degrees with a thesis component. International comparisons vary in how they treat them.
A fourth caveat is the foreign-student share of the Chinese PhD pipeline. Unlike the US, where the foreign-student share is large and well-documented, the Chinese PhD pipeline is overwhelmingly Chinese-domestic, with a much smaller international contingent. The CSET report’s domestic-pipeline comparison strips out the international share on the US side and finds the 3.3-to-1 ratio. If the same adjustment is made on the Chinese side, the ratio widens to roughly 3.5 to 1, because the international share of Chinese STEM PhDs is in the low single digits.
A fifth caveat is the quality dimension. PhD output is not the same as research output, and research output is not the same as research impact. The 77,000-PhD figure tells you how many people started research careers in a given year; it does not tell you how many of them stayed in research, published in high-impact journals, supervised successful students or produced findings that mattered. The Chinese system has been criticised, by both Chinese and Western analysts, for a long-standing preference for paper-count metrics and for an output-heavy promotion system that can overproduce papers of variable quality. The Nature Index 8-of-top-10 ranking is the upside of the same trajectory.
Where the PhDs end up
The pipeline question that follows the output question is the destination question. The most recent reliable data on Chinese STEM PhD destinations comes from CSET and a small number of follow-up studies. The headline finding is that the Chinese system retains most of its own output: about 90 per cent of Chinese STEM PhD graduates stay in China for at least their first post-PhD position. Of those, the largest single employment sector is academia (universities and research institutes), followed by industry (especially the technology and energy sectors), and a much smaller but still significant civil service and policy sector.
The US-side destination question is the one that gets more policy attention. Of the roughly 40,000 STEM PhDs awarded each year in the United States, the largest single employment destination is industry, followed by academia, with a small but significant postdoc and foreign-return share. The international-student share, 42 per cent of the US STEM PhD total, has been the focus of US-China STEM competition policy since 2018. Restrictive visa rules and the post-2020 China-US tension have produced a modest but measurable shift in the destination choices of Chinese students in US PhD programmes, with a higher share now returning to China after completing their degrees.
The net effect on the global STEM workforce is small in any single year but compounding. A 2024 CSET report estimates that for every 100 Chinese STEM PhDs produced, 90 stay in China, 6 go to the United States or the European Union, and 4 go to other Asia-Pacific countries. For every 100 American STEM PhDs produced, 23 stay in the US as domestic-trained PhDs, 17 are temporary-visa internationals who then return home (often to China or India), and the rest go to industry. The asymmetry, China retaining nearly all of its STEM PhD output while the US retains only a portion of its own, is the most-cited structural fact in current US STEM workforce analysis.
Per head of population
The 5-million figure is large in absolute terms. It is also worth comparing it to the size of the population it sits in. China has roughly 1.41 billion people, so 5 million STEM graduates a year works out to about 0.36 per cent of the population, or 3.6 graduates per 1,000 people. The United States, with 1.6 to 2.0 million STEM graduates against 335 million people, is at 0.5 to 0.6 per cent, or 5 to 6 per 1,000. The European Union, around 1.0 million STEM graduates against 450 million, is at 0.22 per cent, or 2.2 per 1,000. India, around 2.5 million against 1.43 billion, is at about 0.18 per cent, or 1.8 per 1,000.
The per-capita comparison tells a different story from the absolute one. The United States still produces more STEM graduates per head than China does; the European Union produces fewer per head; India produces much fewer. What China has done is run a much larger absolute production through a much larger population, and run it for long enough that the cumulative stock of STEM graduates is now well past the US stock in the 25-to-34 age bracket. The pipeline is not just large, it is large relative to the size of the cohort coming through the universities, and the cohort itself is large relative to the working-age population.
The cohort size matters because research output is not just a function of the number of PhDs, it is a function of the number of PhDs who stay in research for ten or twenty years after graduating. The Chinese system is now mature enough that the 2010s PhD cohorts are the senior postdoc and junior faculty cohort of the 2020s, and the 2020s PhD cohorts will be the senior cohort of the 2030s. The compounding effect is the point at which the absolute 77,000 number becomes a research-workforce number, not just a graduation-rate number.
The per-capita comparison is sharpened by looking at the other East Asian research economies. South Korea, with a population of about 52 million, produces around 14,000 STEM PhDs a year, or about 27 per million people. Japan, with 124 million, produces around 6,500 STEM PhDs, or about 5 per million. Taiwan, with 24 million, produces around 5,500 STEM PhDs, or about 23 per million. Singapore, with 6 million, produces around 1,500 STEM PhDs, or about 25 per million. China, with 77,000 STEM PhDs against 1,410 million people, is at about 55 per million. The Chinese per-capita PhD production is now ahead of the East Asian comparators, and approaching the US figure of about 120 per million. The Chinese pipeline is the largest in the region; it is also the youngest. The 1990s and 2000s expansion of Chinese PhD programmes is still playing out, and the cumulative effect on the regional research workforce over the next decade is the next thing to watch.
Other comparison points
Two more comparison points are worth setting down, because they round out the picture.
The first is the share of the country’s own university output that goes to STEM. The 45-per-cent figure for new college entrants studying STEM subjects in China, and the 60-per-cent figure for postgraduates, are well above the corresponding shares in the United States, the European Union and Japan. A high share at the front of the pipeline, sustained over a quarter century, produces the 77,000-PhD figure at the back of the pipeline.
The second is the research-and-development spending. China’s gross domestic R&D spending crossed 3.5 per cent of GDP in 2023, ahead of the European Union and approaching the United States. The bulk of the spending increase has gone to applied research and experimental development, with a smaller but still growing share going to basic research. The PhD pipeline and the R&D spending move together; the 77,000-PhD figure is the human-capital expression of the R&D-spending trajectory of the past decade.
What the figure does and does not say
The Chinese STEM graduate and PhD numbers are large. They are larger than any other country’s. The exact size of the number depends on which definition is in play, and the definition is a choice, not a discovery. The 5-million figure is real on its own terms, and so is the 3.57-million figure; they are measuring different things.
The doctorate component is the part that drives the global research output. The 77,000-PhD figure is a near-term projection, supported by current enrolment patterns and corroborated by independent analysts. The trajectory is real: tenfold growth in 25 years, against a flat-to-slowly-rising trend in the United States and the European Union. The Nature Index 8-of-top-10 ranking is the publication-output expression of the same trajectory.
The figures do not tell you whether the Chinese research system is more efficient than the American or European one, whether the average quality of a Chinese STEM PhD is the same as the average quality of an American STEM PhD, whether the Chinese system produces more or fewer high-impact findings, or whether the long-term sustainability of the pipeline is sound. Those questions require a different set of data: publication-citation analysis, comparative funding trends, retention and attrition rates by field, and a careful look at the quality distribution within the 77,000. A useful illustration of the gap is the so-called paper-count incentive that has been a feature of the Chinese academic promotion system for the past two decades. A faculty member at a 985 university is rewarded by a formula that counts publications in indexed journals, particularly in English-language journals, and the formula has been credited with both the 8-of-top-10 Nature Index ranking and with a parallel growth in mid-impact, low-citation output that does not move the global needle. The system is now being rebalanced, with more weight given to patent quality and to industry collaboration, but the older incentive structure is still the one that produced the cohorts now in their late thirties and forties.
What the figures do tell you is the size of the pipeline. The pipeline is enormous. The pipeline is the largest single national pipeline in the world, and it has been growing faster than any other comparable national pipeline for a quarter century. The figures also tell you that the doctorate component is where the global research output comes from, and that the Chinese doctorate component is now the largest in the world. Those two facts are enough to explain most of the recent shape of global STEM workforce analysis, and they are the part of the data on which the cross-country comparisons should rest.
The next decade will show whether the pipeline, now mature, turns into a comparable advantage in the very high-impact research that the Nature Index 8-of-top-10 ranking only partly captures. The Chinese system’s next stage is not about more graduates, it is about whether the graduates it already produces stay in research, win the international prizes, take the editorial positions at the major journals, and lead the kinds of long-horizon projects that have, until now, been the preserve of US and European institutions. The 77,000-PhD figure is the size of the engine. The question that follows is what the engine, once it is running at full speed, actually moves. That is the story for the 2030s. This dispatch is the baseline.
| Country or region | STEM graduates per year (broad definition, 2023) | STEM PhDs per year (2023 to 2025 estimate) | Domestic STEM PhDs retained | Notes |
|---|---|---|---|---|
| China | About 5.0 million (or about 3.57 million, narrow definition) | About 77,000 | About 90 per cent | Includes agriculture, medicine, professional master’s, dazhuan. Largest single national pipeline. |
| United States | About 1.6 to 2.0 million (NSF, 2014 baseline) | About 40,000 | About 58 per cent after international-student return | Excludes international temporary-visa PhDs from the domestic retention figure. |
| India | About 2.5 million (UNESCO and WEF) | About 25,000 to 30,000 | About 75 per cent, rising | Fast-growing; second-largest in the medium term. |
| European Union | About 1.0 million (combined) | About 40,000 to 50,000 (combined) | About 70 per cent | Distributed across 27 member states; Germany is the largest single producer. |
| Japan | About 250,000 (Ministry of Education) | About 6,000 to 7,000 | About 80 per cent | Slow decline since 2005; ageing demographic. |
References
- State Council Information Office, “From classrooms to talent hubs: China looks to education to drive”, English-language release, 11 May 2026. http://english.scio.gov.cn/m/chinavoices/2026-05/11/content_118486938.html
- Saif M. Khan, “China is rapidly becoming a leading innovator in science and technology”, testimony summary and CSET report, Center for Security and Emerging Technology at Georgetown University, August 2024, on the 77,000-per-year STEM PhD projection and the 40,000 US comparison. https://cset.georgetown.edu/article/should-the-us-fear-rising-number-of-stem-phds-in-china/
- WorldRankd, “Countries With the Most STEM Graduates [2026]”, accessed October 2026, on the 3.57 million per year Chinese figure (OECD field definition). https://www.worldrankd.com/education/countries-most-stem-graduates-2026
- Asia Times, editorial and reporting on Chinese STEM-pipeline growth, 2026. https://asiatimes.com/2026/06/university-research-at-china-speed-brings-sea-changes-to-science/
- Nature Index, global science ranking, top institutions, 2024 to 2025 release, on the 8-of-top-10 ranking of Chinese institutions. https://www.nature.com/nature-index/
- National Science Board, Science and Engineering Indicators, on the methodological difference between Chinese and US STEM field definitions and the under-count that results on the US side. https://ncses.nsf.gov/pubs/nsb20211/
- World Economic Forum, “Which countries’ students are getting most involved in STEM?”, 2016 baseline, on the 4.7-million-per-year Chinese figure and the UNESCO non-publication note. https://www.weforum.org/stories/education-and-skills/which-countries-students-are-getting-most-involved-in-stem/
- Stephen Bradley, public discussion of CSET findings, on the 77,000-PhD figure and the Nature Index trend. https://www.linkedin.com/posts/stephen-bradley-31661b4a_china-produces-77000-stem-phds-per-year-activity-7446527831083229184-ZoEz
- IDA Paper P-23835, Challenges to China’s Academic STEM Research Ecosystem, on the historical and comparative context for Chinese STEM PhD output. No current public URL; the paper is in the IDA archives.
- OECD, Education at a Glance, on EU and member-state STEM PhD output (Germany about 12,000 to 15,000 per year). https://www.oecd.org/education/education-at-a-glance/
