Late on July 23, the Fields Medal, awarded once every four years and widely regarded as mathematics’ highest honor, was announced. Two Chinese mathematicians won it this time: Wang Hong and Deng Yu.
Both were classmates in the 2007 undergraduate cohort at Peking University’s School of Mathematical Sciences. The report says Wang solved the century-old three-dimensional Kakeya conjecture, while Deng resolved the restricted version of Hilbert’s sixth problem. It was also the first time that mathematicians who completed their undergraduate studies in mainland China received the Fields Medal, and the first time two such winners appeared in the same year.
Shortly after the results were released, Shing-Tung Yau said at a public event at Tsinghua University: “I sincerely invite Wang Hong and Deng Yu to come back and teach. Tsinghua’s Qiuzhen College will always welcome them.”
That remark quickly pulled a long-running contest between Tsinghua and Peking University back into public view. As laid out in the original article, the rivalry did not begin with this year’s prize and does not stop at a handful of star scholars. It stretches across talent pipelines, faculty hiring, research institutions, and, more recently, AI for mathematics.
The 1952 restructuring left Peking University with an early lead
The article traces the starting point to the nationwide restructuring of university departments in 1952. Tsinghua’s mathematics department was moved in its entirety to Peking University, along with faculty and library materials. Tsinghua’s mathematics program then fell into a gap that lasted close to 30 years, until an applied mathematics department was rebuilt in 1979.
After that shift, Peking University’s mathematics school held a dominant position for years, according to the report. The country’s strongest mathematics students were widely expected to choose Peking University. The article names Zhang Yitang, Xu Chenyang, Yun Zhiwei and Zhang Wei among the scholars who came out of that system, and says a large share of younger Chinese mathematicians who later won the Sloan Research Fellowship, the Ramanujan Prize or the New Horizons Prize were also Peking University graduates.
That edge also showed up in Olympiad recruitment. Math competition gold medalists were said to treat Peking University as the default destination when filling out applications. The article cites Wei Dongyi as an example, saying his direct admission to Peking University after two perfect-score appearances at the International Mathematical Olympiad was seen at the time as entirely natural.
Yau’s relationship with Peking University went from support to rupture
The report says Yau’s original intention was not to oppose Peking University. When he returned to China to lecture in 1979, his first stop was Peking University. He wanted to help build a world-class mathematics center in China. Tian Gang was one of the students he recruited during that period and later brought to Harvard for doctoral study, becoming one of Yau’s best-known protégés.
For roughly the next two decades, Yau moved back and forth between China and overseas academia. The article says he helped establish mathematics institutes at the Chinese Academy of Sciences and Zhejiang University, donated books to Peking University, and taught there as well. At that point, he believed the roots of Chinese mathematics would likely be planted in Yanyuan.
By around 2000, though, disagreements over academic standards and institution-building had grown sharper. According to the article, Yau wanted platforms built around international standards and objected to what he viewed as academic inbreeding. He was also dissatisfied with the number of in-house graduates retained by Peking University and with what he saw as a hierarchy-driven evaluation system.
The conflict erupted publicly in 2005. The article says Yau openly criticized Peking University’s mathematics establishment and named Tian Gang directly, accusing him of academic misconduct and problems in academic style. He also said Peking University’s mathematics research center was operating as a “small circle.” A retired Peking University professor responded, according to the report, by saying: “Shing-Tung Yau wanted to become the overlord of China’s mathematics world and failed.” From that point on, the teacher-student relationship broke down completely, and the article says Yau never again stepped into Peking University’s mathematics school.
After moving to Tsinghua, Yau built a platform of his own
In 2009, Yau became director of the Yau Mathematical Sciences Center at Tsinghua University. The article highlights one detail in particular: from his first return lecture tour in 1979 onward, Yau spent more than four decades teaching and building centers in China without drawing a salary from domestic institutions, and he personally donated $1 million to Tsinghua to support mathematics talent development.
In 2010, he funded the Yau College Student Mathematics Contest out of his own pocket. The report says the competition was modeled on the Putnam Competition, one of the best-known and most difficult university-level mathematics contests in the world, and that its level was well above ordinary domestic collegiate math contests. In the article’s framing, it was both a screening mechanism and a recruitment tool.
Tsinghua then pushed its training pipeline to younger and younger students. In 2018, it launched the Yau Mathematical Sciences Talent Class, which admits no more than 30 high school juniors and seniors each year without requiring the gaokao. In 2021, it went further by establishing Qiuzhen College and introducing a “Mathematics Leading Program” aimed at students as early as the ninth grade. That track combines undergraduate and doctoral study over eight years, takes up to 100 students per year, and provides English-language instruction and top faculty resources.
The article gives an example from July 2026. In a preparatory class of 94 students, 20% were removed after failing two core courses, mathematical analysis and higher algebra. More than 20 parents reportedly protested at the school, and several vice deans urged Yau to ease up because stricter standards could hurt the next year’s recruitment. He rejected that suggestion and was quoted as saying: “Students scoring in the 20s and 30s are a waste of resources if they stay.”
That episode is used in the article to illustrate how hardline Yau’s standards have been in trying to close the gap with Peking University’s long-built mathematics base.
Peking University still held a deep lead in student reserves
Even with Tsinghua’s heavier investment, the article argues that Peking University’s accumulated advantages remained difficult to dislodge. It says 80% of the country’s annual math Olympiad gold medalists ultimately went to Peking University’s mathematics school.
Wei Dongyi appears again as the clearest example. The report says he earned two IMO gold medals with perfect scores and won repeated honors in the Yau competition as an undergraduate. It also cites a line that circulated in math circles at the time: when Wei competed, everyone else was fighting for second place. After graduation, he stayed at the Beijing International Center for Mathematical Research, directed by Tian Gang, and became one of the most visible symbols of Peking University mathematics.
The competition expanded from students to professors and resources
In the article’s telling, the rivalry did not stop at undergraduate recruitment. It spread into faculty hiring, research funding and academic resources.
On the student side, once Tsinghua pushed its recruitment threshold down to the ninth grade, Peking University followed with its own talent class and began targeting younger candidates too. Around the release of results for the Chinese Mathematical Olympiad, the report says, admissions teams from both universities would wait outside exam venues and move quickly on gold medalists, trying to secure commitments before the other side could.
Some participants told stories, according to the article, of leaving the exam room only to be approached by Tsinghua staff and then pulled into discussions by Peking University staff. Others were reportedly kept in hotel meetings by the two schools in rotation from afternoon into late night. In some cases, admissions teams were said to promise broad freedom in choosing a major and even an extra direct-admission quota for a student’s high school if the student enrolled.
The faculty battle was just as intense. Tsinghua recruited a number of young mathematicians from overseas in recent years with high salaries, formal positions and startup research support. Peking University, for its part, leaned on its alumni network to bring back top scholars it had trained earlier.
In 2021, Tsinghua recruited Fields Medalist Caucher Birkar, then 40, to the Yau Mathematical Sciences Center. The article says this was the first time a Chinese university directly brought in a Fields Medal winner still in his prime, and that Tsinghua offered its highest-level treatment package.
The report also names Jun Liu, a tenured professor in the statistics department at the University of Pennsylvania, and Liu Zhengwei of Vanderbilt University among the younger scholars Tsinghua hired from overseas. Liu Zhengwei, it notes, was himself a Peking University undergraduate.
Peking University answered on its own side. In 2024, it brought Jian Ding, a tenured professor at the Wharton School of the University of Pennsylvania, back to the School of Mathematical Sciences as a chaired professor. The article says Ding was an undergraduate in Peking University’s 2003 cohort and had won the Rollo Davidson Prize, described in the report as the top prize in probability theory. Tsinghua had also made him an offer, but he chose to return to Yanyuan.
Earlier, Peking University had already brought back figures from its so-called golden generation, including Xu Chenyang, Yun Zhiwei and Zhang Wei. Xu later moved to MIT, but the article says Peking University never stopped relying on alumni return as a strategy. In some cases, the two schools were said to offer almost identical packages to the same Sloan-winning young scholar, leaving personal ties to advisers as the decisive factor.
AI became the newest battleground in mathematics
The sharpest new front, the article argues, is AI. It quotes Yau as saying: “Without breakthroughs in mathematics, large models will forever be embroidered pillows stuffed with parameters.” In his view, as described by the report, the bottleneck in large-model reasoning is fundamentally a bottleneck in mathematical capability. Whoever pushes AI math ability forward first may end up controlling entry into the next generation of major models.
Tsinghua is portrayed as moving faster. According to the report, Tsinghua worked with Tencent Hunyuan on an algorithm called STARE designed to address a common failure mode in mathematical reasoning, where one mistake in the middle invalidates everything that follows. On the AIME24 and AIME25 benchmarks, described as leading math competition benchmarks, accuracy improved by 4% to 8%.
The article also points to Zhipu AI, built by a team led by Tang Jie in Tsinghua’s computer science department, as one of the front-rank Chinese large-model groups in mathematical performance. It says support from people in Tsinghua’s mathematics department sits behind that capability.
Peking University’s side of the race includes TongGeometry, a geometry-focused large model developed with the Beijing Institute for General Artificial Intelligence. The report says it built the world’s largest geometry theorem database, targets the hardest geometry problems from the IMO, and can even discover new geometric theorems on its own.
It also mentions MiniMax’s M3 model, described as having achieved IMO gold-medal level performance, and says members of Peking University’s mathematics school were part of the core team behind it.
At the 2025 WAIC World Artificial Intelligence Conference, Yau personally wrote problems for several leading Chinese large models and had them sit a closed-book test on site. In the account given by the article, the scores ended up extremely close, with no clear winner.
The rivalry also helped lift Chinese mathematics as a whole
The article says that 20 years ago, Chinese mathematics was not internationally prominent, strong students often left for the United States after graduation, and China lacked first-rate mathematics research centers at home. It argues that the picture looks different now.
According to the report, one-third of the award-winning scholars on the ICCM list now come from mainland China, triple the share from 10 years earlier. Qiuzhen College at Tsinghua and the School of Mathematical Sciences at Peking University are both said to produce papers in the field’s four top journals every year, and to turn out several scholars annually at the level of Sloan award winners.
In that sense, the article frames the long contest between the two schools as a force that strengthened both. Yau may have criticized Peking University harshly, it says, but he also knew that Tsinghua would not have pushed so hard to build mathematics centers and training systems without such a strong rival nearby. Peking University, in turn, kept adjusting its own pipeline and fighting for the strongest young talent because Tsinghua was pressing from behind.
At the 2026 Yau contest, Peking University swept the top four places
On June 28, 2026, the 17th Yau College Student Mathematics Contest held its award ceremony at Tsinghua University’s Mengminwei Concert Hall. Yau stood on stage, with many of China’s top student math talents in the audience.
In the most important overall individual award, the Yau Award, the gold medal went to Shi Haojia of Peking University. The silver medals went to Lu Junpeng and Guo Huankun, also from Peking University, and the bronze medal went to Ye Yuxing of Peking University. The article emphasizes that the top four places were all taken by Peking University students.
Among the six individual gold medals, the Hua Luogeng Award, the Chern Shiing-Shen Award, the Wei-Liang Chow Award, the Chiao-Lin Li Award, the Pao-Lu Hsu Award and the overall award all went to Peking University as well. The team gold, by contrast, was shared by Tsinghua and Peking University.
Even so, the report says Yau did not appear visibly upset while handing awards to Peking University students. In his remarks, he called on young scholars in China to work across university lines and push Chinese mathematics forward together.
In the end, AI may put both schools on the same new starting line
The article closes by returning to the intersection of AI and mathematics. On July 20, it says, an Anthropic researcher spent several hours running the Fable 5 model while watching the World Cup final and then posted a simple three-line polynomial argument that overturned the Jacobian conjecture, a problem the article says had stood for 87 years. It also notes that Zhang Yitang’s doctoral thesis had been stuck for seven years in this same research direction.
Against that backdrop, the report argues that Tsinghua is ahead in this new AI-math race. It says a substantial share of China’s leading large-model groups are tied to Tsinghua, including Tang Jie and Moonshot AI, founded by his former student Yang Zhilin. Moonshot’s Kimi K3 model, the article says, now performs at roughly the same mathematical level as Fable 5 and GPT-5.6 Sol, making it one of the strongest Chinese models in mathematics.
Peking University is not absent from the field, the article adds, but many of its teams still lean more toward pure academic research. In commercial model deployment and broader public recognition, it says, Peking University has not produced products on the level of Zhipu AI or Kimi.
The report also cites a Stanford talk by Terence Tao three months earlier. Tao said mathematics had moved from an era of “proof scarcity” into one of “proof surplus.” AI, in his description, can generate dozens of proofs at top-journal level in a single day, faster than human mathematicians can read or verify them. For that reason, he said, he no longer tries to keep up with every new paper. The article notes that just a little more than a year earlier, Tao had still criticized AI as too “stupid,” saying it could get eight out of 10 problems wrong.
That is where the original piece leaves the story. Tsinghua and Peking University have spent decades fighting over students, professors, prizes and institutional prestige. But in the face of AI, those older scorecards are starting to matter less. No matter which side won more medals or hired more star faculty, both now face the same next stage: a period in which AI may begin to shape mathematics research itself.
This article was originally published on the WeChat public account Zimu AI, written by Miao Zheng and edited by Wang Jing, and republished by MarsBit.