Top statistician Jun Liu has left Harvard University after 25 years to accept a full-time chair at Tsinghua University, formalized in August 2025, capping a year marked by his election to the U.S. National Academy of Sciences (NAS) and expanding leadership in statistics and AI in China [1][4]. Liu cited patriotism, shifting U.S. policies under the Trump administration, and a commitment to education and science as factors shaping his decision, underscoring a high-profile eastward move with outsized academic and strategic implications [1].

In April 2025, the NAS named 120 new members and 30 international members; the roster included “Jun S. Liu,” then listed as professor of statistics at Harvard, a signal of his stature just months before the transition became public [4]. Tsinghua and Chinese media have emphasized Liu’s deep publication record, foundational work in MCMC and statistical learning, and longstanding roles shaping Tsinghua’s data-science infrastructure since 2015, culminating in this summer’s full-time appointment [2][3][5].

Key Takeaways

– Shows Jun Liu concluded a 25-year Harvard tenure by accepting a full-time Tsinghua chair in August 2025, marking a decisive institutional realignment [1]. – Reveals NAS elected Liu on April 30, 2025 among 120 new members and 30 international members, highlighting peak recognition before his move [4]. – Demonstrates scholarly impact with over 300 publications and more than 90,000 Google Scholar citations reported as of May 2025, spanning decades [2][3]. – Indicates administrative leadership since 2015 at Tsinghua, including heading the preparatory committee for a new statistics and data science department in July 2024 [3]. – Suggests alignment with a May 30, 2025 trend: top scholars with Harvard ties move to China amid competitive incentives in AI, biomedicine, and data science [5].

Jun Liu’s career milestones and metrics

Liu’s academic path bridges elite institutions and landmark contributions: a BS from Peking University in 1985, followed by a PhD from the University of Chicago in 1991, established the foundation for his later prominence in Bayesian computation, MCMC methods, and statistical learning [2]. His recognition includes the 2002 COPSS Presidents’ Award, regarded as the most prestigious early- to mid-career prize in statistics, reflecting early, sustained influence in theory and applications alike [2].

By 2025, Tsinghua reported that Liu’s publication record exceeded 300 papers, with a citation footprint surpassing 90,000 on Google Scholar, indicating both productivity and broad adoption of his methods across disciplines [2][3]. These figures place Liu among the most cited statisticians globally, aligning with the NAS’s emphasis on original research and sustained achievement exemplified by his body of work [4].

The NAS’s April 30, 2025 announcement listed Liu as a Harvard professor at the time of election, situating his peak U.S. recognition only months before he accepted a full-time appointment at Tsinghua in August 2025 [4][1]. That sequencing underscores the degree to which elite acknowledgment and international mobility can converge in a single year, reshaping where high-impact research leadership resides [4][1].

Timeline: Jun Liu’s roles at Tsinghua and Harvard

Liu’s relationship with Tsinghua predates his formal move, with leadership roles documented since 2015, indicating a long-running engagement with the university’s statistics and data science ambitions [3]. In July 2024, Tsinghua named him head of the preparatory committee for a new Department of Statistics and Data Science, an organizational milestone that foreshadowed deeper integration [3].

The NAS election on April 30, 2025 showcased his global stature, with Liu recognized among 120 members and 30 international members in that class, a pivotal accolade preceding his transition [4]. In August 2025, he accepted a full-time chair at Tsinghua, ending a 25-year tenure at Harvard and formalizing a move that had been building through administrative and research leadership ties [1].

The 25-year figure highlights the significance of the shift, marking the end of a long Harvard era for Liu even as he enters a rapidly expanding research ecosystem at Tsinghua [1]. The timing aligns with Tsinghua’s active build-out of a new department under his preparatory leadership, suggesting a continuity of vision between his prior advisory roles and his new full-time mandate [3].

Why Jun Liu’s move matters for AI and statistics

Liu’s expertise in MCMC and statistical learning sits at the core of modern AI workflows, from Bayesian inference to probabilistic modeling and scalable posterior sampling, translating directly into cutting-edge machine learning research [2]. SCMP’s reporting frames his decision as driven by patriotism and policy shifts in the United States, along with a stated devotion to education and science, pointing to a confluence of personal values and institutional opportunities in Beijing [1].

CGTN’s May 30, 2025 analysis highlights a trend: top scholars with Harvard affiliations heading to China, attracted by competitive packages, infrastructure, and targeted investment in AI, biomedicine, and data science [5]. Liu’s move interlocks with that pattern, but with additional weight given his NAS election and his role in shaping a new statistics and data science department at Tsinghua [4][3][5].

As an NAS member with over 300 publications and more than 90,000 citations as of May 2025, Liu brings both symbolic and practical capacity to catalyze research, hiring, and cross-border collaboration in AI-intensive statistical science [2][3][4]. The appointment also signals how institutions are vying for leadership in core methods that underpin generative models, causal inference, and uncertainty quantification across industry and academia [2][5].

Comparing academic signals: NAS election and institutional alignment

The NAS class of 2025 recognized Liu’s lifetime contributions while listing his affiliation as Harvard, a snapshot taken just as his institutional alignment was poised to shift eastward [4]. Only months later, the acceptance of a full-time Tsinghua chair in August 2025 reframed that recognition as a credential now anchored in Beijing’s premier STEM university [1].

This rapid sequence underscores how global talent markets function at the highest level: recognition by a U.S. academy, followed by immediate deployment of that prestige into institution-building roles abroad [4][3]. Tsinghua’s citations and leadership appointments since 2015 make clear that Liu’s role is not merely honorary; it is operational and strategic within the university’s data-centric expansion [3][2].

Factors behind the decision, in Liu’s words and policy context

According to SCMP’s account, Liu pointed to patriotism and U.S. policy shifts under the Trump administration as central to his thinking, while emphasizing a “love for education and science” that he expects to pursue through his Tsinghua role [1]. The framing suggests an alignment of personal motivations with programmatic opportunities inside a university investing heavily in statistical and AI infrastructure [1][5].

CGTN’s reporting provides broader context, noting that competitive academic packages, advanced research infrastructure, and national priorities in AI and biomedicine are drawing senior scholars to China, including those with Harvard ties [5]. The combination of national strategy and institutional build-outs helps explain why Tsinghua could convert Liu’s advisory leadership since 2015 into a full-time chair by August 2025 [3][5][1].

What to watch next: department build-out, hiring, and collaboration

Liu’s July 2024 appointment as head of the preparatory committee for Tsinghua’s new Department of Statistics and Data Science indicates continued growth in curriculum design, faculty recruitment, and lab-scale infrastructure oriented toward AI and statistical learning [3]. Given his NAS recognition and long record in MCMC, one should expect sustained emphasis on probabilistic methods that translate into machine learning pipelines and data-centric discovery [4][2].

Tsinghua’s communications emphasize Liu’s extensive publication record and high citation impact as of May 2025, implying that mentorship and team-building will be core to his remit as the department scales [2][3]. Meanwhile, SCMP’s report that he left Harvard after 25 years, formalized in August 2025, suggests an institutional pivot with long-term implications for trans-Pacific collaboration, graduate training pipelines, and joint research programs [1].

CGTN’s coverage of incentives and infrastructure points to a policy environment favoring rapid build-out of AI and data science capacity, a context in which Liu’s move functions as both signal and accelerant for similar returns by elite researchers [5]. The NAS election, coming on April 30, 2025, provides a high-visibility credential that may help Tsinghua attract additional senior hires and industry partnerships in the near term [4].

Sources:

[1] South China Morning Post – Top Harvard mathematician Liu Jun leaves US for China: www.scmp.com/news/china/science/article/3324637/top-harvard-mathematician-liu-jun-leaves-us-china” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.scmp.com/news/china/science/article/3324637/top-harvard-mathematician-liu-jun-leaves-us-china

[2] Department of Statistics and Data Science, Tsinghua University – Professor Jun Liu Elected to the U.S. National Academy of Sciences: www.stat.tsinghua.edu.cn/en/info/1054/1391.htm” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.stat.tsinghua.edu.cn/en/info/1054/1391.htm [3] Department of Statistics and Data Science, Tsinghua University – Warm Congratulations to Professor Jun Liu on Election to the National Academy of Sciences: www.stat.tsinghua.edu.cn/en/info/1101/1421.htm” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.stat.tsinghua.edu.cn/en/info/1101/1421.htm

[4] National Academy of Sciences – National Academy of Sciences Elects Members and International Members: www.nasonline.org/news/2025-nas-election/” target=”_blank” rel=”nofollow noopener noreferrer”>https://www.nasonline.org/news/2025-nas-election/ [5] CGTN – Harvard’s loss, China’s gain: Top talents turn eastward: https://news.cgtn.com/news/2025-05-30/Harvard-s-lost-China-s-gain-Top-talents-turn-eastward-1DK93pzHCI8/index.html TARGET_KEYWORDS: [Jun Liu, Jun Liu NAS 2025, Jun Liu Tsinghua 2025, Jun Liu Harvard 25 years, 120 NAS members 2025, 30 international members NAS, Jun Liu publications 300, Jun Liu 90,000 citations, Peking University 1985 BS, University of Chicago 1991 PhD, COPSS Presidents’ Award 2002, Tsinghua Department of Statistics and Data Science, July 2024 preparatory committee, AI and statistics leadership, MCMC contributions Jun Liu, Harvard to Tsinghua move, Chinese academic return trend 2025, NAS election April 30 2025, Harvard statistics professor Jun Liu, Tsinghua full-time chair August 2025] FOCUS_KEYWORDS: [Jun Liu, Jun Liu NAS election, Jun Liu Tsinghua chair, Jun Liu 25-year Harvard, Jun Liu 90,000 citations, Jun Liu 300 publications, Jun Liu MCMC] SEMANTIC_KEYWORDS: [Markov chain Monte Carlo, Bayesian computation, statistical learning, citation impact, academic mobility, research infrastructure, generative models, uncertainty quantification, probabilistic modeling, faculty recruitment, data science department, international recognition] LONG_TAIL_KEYWORDS: [Why did Jun Liu leave Harvard, Jun Liu NAS class of 2025 details, Tsinghua statistics department leadership 2025, Jun Liu publications and citations 2025, Harvard to Tsinghua academic move timeline, COPSS Presidents’ Award Jun Liu 2002, Chinese incentives for AI scholars 2025, Jun Liu Peking University 1985 background, University of Chicago 1991 statistics PhD Jun Liu] FEATURED_SNIPPET: Jun Liu left Harvard after 25 years to accept a full-time chair at Tsinghua University in August 2025. On April 30, 2025, the NAS elected him among 120 new members and 30 international members, underscoring his global stature. Tsinghua cites 300+ publications and over 90,000 citations as of May 2025, alongside leadership roles since 2015, including heading a new statistics and data science department’s preparatory committee in July 2024. [1][4][2][3][5]

Image generated by DALL-E 3