The race for artificial intelligence talent is intensifying, with major tech companies offering staggering salaries to secure top graduates even before they finish their studies.

As the graduation season drew to a close on June 28th, the NLP Laboratory at Tsinghua University's Computer Science Department released statistics on its graduates' destinations. Among 14 graduates, four joined ByteDance, three entered lab-incubated enterprises, one joined DeepSeek, with the remainder pursuing research at academic institutions. This list holds significant weight in China's AI industry, as the Tsinghua NLP Lab is a renowned large model research hub and a key talent pipeline for major internet firms.

Liu Zhiyuan, a professor at Tsinghua's Computer Science Department and one of the lab's academic leaders, revealed that doctoral graduates specializing in large models can command annual salaries exceeding 6 million yuan, while master's graduates can earn over 1 million yuan. This represents the pinnacle of compensation within China's current AI sector.

The competition for young talent has escalated to extraordinary levels. During recent graduation defenses at Tsinghua, HR representatives from major companies were seen bringing flowers to congratulate newly hired students. A report from Maimai's high-end recruitment division highlights a stark salary divide in the fight for elite technical talent. AI scientists and technical leads average a monthly salary of 132,800 yuan. In the first four months of this year, core roles like AIGC algorithm engineers and NLP specialists commonly offered over 60,000 yuan per month, with some exceeding 70,000 yuan.

Simultaneously, new AI job postings surged 8.7 times year-over-year, while the supply-demand ratio for underlying computing power talent plummeted to 0.26, meaning roughly four positions compete for one high-performance computing engineer. This global talent war is not confined to China; over the past year, companies like Meta, OpenAI, and Google have intensified their efforts, offering sky-high salaries and equity to poach top researchers, with founders even personally involved in recruitment.

The Early Bird Catches the Talent

The battle for AI minds is starting well before graduation. During the CVPR conference in Colorado, tech giants hosted evening events to connect with young researchers. Zhou Tianming, a 24-year-old doctoral student, submitted over 30 dinner invitations, attending events hosted by Alibaba, MiniMax, and Intel. These gatherings are less about meals and more about networking; company employees discuss research directions and projects with students, who in turn learn about the latest developments and computing resources.

This scene has become commonplace at international AI conferences. Professor Liu notes that companies no longer wait for graduation season; students often receive HR inquiries immediately after publishing papers. He mentioned one extreme case where an HR representative suggested a student abandon their degree to start work immediately—a practice he discourages but one that underscores the fierce competition.

Ba Ran, Vice President of Tongdao Liepin Group, observes a shift in hiring criteria. Beyond academic pedigree, companies now prioritize experience with large model projects, Agent development, open-source contributions, and practical business experience. The recruitment battlefield has expanded from post-graduation to internships, research collaborations, and joint training programs at much earlier stages.

The rise of DeepSeek in 2025 has accelerated this trend. Its founder noted that core members are often fresh graduates, doctoral interns, and young professionals just a few years out of school. Over the past two years, programs like Alibaba's "Ali Star," TENCENT's "Qingyun Plan," ByteDance's "Top Seed," and JD.com's youth talent initiatives have targeted undergraduates, master's students, doctoral candidates, and even current students.

Zhou Tianming observed during his internships that teams were predominantly composed of master's students, followed by undergraduates, with the youngest member being a sophomore. The key advantage for such young talent is pure potential. Some industry insiders point out that for those at the very top, formal education matters less than learning ability and innate talent, with companies fiercely competing for top-performing competition winners and gifted undergraduates.

This reach now extends to high schools. TENCENT launched a "Youth Science Training Camp" for global students aged 13-18, offering hands-on experience in core departments like FinTech and WeChat Mini-Program AI. Geely introduced a program for high school graduates to enter cutting-edge fields like new energy and AI, with CEO-level mentorship and salaries up to 300,000 yuan, with no college entrance exam score requirement.

This trend is global. Between 2024 and 2025, Google, Meta, and OpenAI collectively hired hundreds of young people who had not completed university, including some straight out of high school.

Intern Salaries Exceeding 40,000 Yuan Monthly

Zhou Tianming's internship salary increased nearly fivefold over two years. Starting in 2024 at a top tech company's autonomous driving unit with a daily rate of around 350 yuan, he moved to large model teams in 2025, where his rate rose from 400 to 1000 yuan daily. He expects a daily rate of 2000 yuan in another company's youth talent program later this year, translating to a monthly internship income of approximately 40,000 yuan—surpassing many full-time employees.

The most sought-after individuals are those with expertise in large model foundations, who command the highest market premiums. Zhou mentioned an overseas peer who secured a 3.5 million yuan annual salary for a peripheral role in a company's foundational team. The second tier includes fast-rising fields like AI Agent, video generation, and embodied intelligence.

Youth talent programs from tech giants have made the competition overt, with recruitment materials featuring phrases like "uncapped salary." Reports indicate that ByteDance's Top Seed graduate salaries rose from about 1.5 million yuan in 2024 to 3-5 million yuan in 2025, with some core roles exceeding 6 million yuan in 2026.

Beyond salary, companies demonstrate extraordinary commitment. It's reported that ByteDance founder Zhang Yiming once invited Tsinghua lab students to his home for dinner to discuss their papers in detail. Headhunters note that since 2023, clients have explicitly requested targeting talent from OpenAI, Anthropic, and Google, with "no expense spared" extending to organizational resources.

However, Ba Ran cautions that the publicized "million-dollar salaries" and "genius youth plans" represent only the very top of the AI talent pyramid. The fiercest competition is for model algorithm engineers, machine learning researchers, system architects, and AI infrastructure engineers—the technical builders who define an AI system's upper limits.

For major firms, internships are the most efficient way to lock in this talent early. Interns like Zhou gain access to vast resources, such as hundreds of H20 chips or clusters of H200 GPUs, enabling research that leads to publications. Companies pay high intern salaries to access cutting-edge research capabilities and academic perspectives that are difficult to cultivate internally in the short term.

Internships also help distribute a company's research risk. Without performance pressure, interns can explore high-risk, exploratory technical paths that full-time employees might avoid. Hiring five interns for the cost of one full-time employee allows a company to explore multiple technical avenues simultaneously.

Ba Ran adds that even scarcer than technical talent are organizational leaders with an AI mindset, who determine whether a company can truly complete an AI transformation. High-profile cases include TENCENT hiring a former OpenAI researcher as its Chief AI Scientist and Xiaomi recruiting a core DeepSeek-V2 developer, reportedly with a salary in the tens of millions, to lead its MiMo large model.

The Race Against Time

Professor Liu has noticed more undergraduates hesitating to pursue doctorates, opting instead for immediate employment, fueled by fears that the AI boom might not last. Students are anxious about missing this unprecedented window of opportunity.

While Liu believes the supply-demand imbalance won't persist indefinitely—with many universities establishing AI schools and expanding graduate programs—the talent war shows no signs of abating for now. Major companies continue to intensify their efforts, and startups are joining the fray. In June 2026, DeepSeek announced its largest recruitment drive since founding, planning to double its team size across various departments.

A headhunter noted that DeepSeek's hiring standards have shifted from broadly seeking excellent resumes to requiring deep expertise in specific sub-fields. Between late 2025 and mid-2026, several core R&D personnel left DeepSeek for giants like ByteDance, TENCENT, and Xiaomi, suggesting internal training couldn't keep pace with rapid business expansion.

As a co-founder of AI startup Zhipu AI, Liu believes large companies and startups attract different types of people. The former relies on systems and resources, while the latter attracts with technological breakthroughs and shared missions. Some top researchers will always prefer joining startup teams for more uncertain but cutting-edge exploration rather than entering the standardized systems of large firms.

He also disagrees with over-glorifying "genius youths" or viewing AI as solely a young person's game. While young researchers were initially favored for their lack of preconceptions, the rapid maturation of AI Coding is renewing the value of experience. His team has hired several veterans with nearly two decades of experience, whose knowledge of building complex systems is now amplified by AI tools, allowing them to accomplish tasks previously requiring entire teams.

Another Tsinghua-born company, Zhipu AI, recruits both young R&D talent and experienced professionals like large model technical architects requiring over five years of experience.

For young people at the forefront of this trend, seizing the moment is paramount. Zhou Tianming no longer advises juniors to pursue doctorates. He observes that in the current AI industry, the advantage of experience is diminishing, allowing young talent to compete on equal footing with veterans and achieve levels of income and status that might otherwise take years.

"This kind of unprecedented opportunity might be hard to come by again in the future," he said, reflecting on a field where five years can change everything.