Meta and ByteDance Escalate AI Talent War in Silicon Valley

The exodus of top AI researchers from ByteDance to Meta has reignited attention on the fierce competition for talent between leading Chinese and American technology companies.

Recent departures, including Qiao Siyuan and Song Yang, underscore how global tech giants are aggressively pursuing engineers and scientists with expertise in artificial intelligence and recommendation algorithms.

Qiao, a core member of ByteDance's AI Seed team and former Google DeepMind researcher, left the company earlier this year to join Meta. Shortly after, Song Yang, who headed TikTok's main site video recommendation algorithm, reportedly followed suit, also joining Meta to oversee Instagram Reels' recommendation operations. Both departures have fueled speculation that ByteDance, despite its rapid growth, is becoming a primary target for Silicon Valley headhunting.

According to LeiFeng.com, ByteDance has seen at least seven Seed team members leave this year. Not all exits were due to external recruitment. Yang Jianchao, leader of Doubao's visual multimodal team, reportedly stepped down for family reasons, while Qiao Mu, formerly responsible for Doubao's large language model, was dismissed following an infidelity scandal. In contrast, Qiao Siyuan's departure appeared abrupt, signaling a more strategic poaching effort.

Industry observers note that Meta's moves reflect a broader, well-financed strategy to secure top AI talent. Mark Zuckerberg's company has reportedly offered compensation packages of unprecedented scale—up to $300 million over four years for key hires, with first-year packages exceeding $100 million. Such financial muscle, paired with high-profile recruitment efforts, illustrates the intensifying competition for talent capable of developing next-generation AI.

ByteDance, the Chinese internet giant behind TikTok, has emerged as a formidable player in both social media and AI. In the first quarter of 2025, the company generated approximately $43 billion in revenue, surpassing Meta's quarterly results for the first time and positioning ByteDance as the world's largest social media company by revenue. A recent secondary market transaction, in which CapitalToday acquired a stake from BOC Group, pushed the company's valuation to roughly $480 billion.

The company has simultaneously expanded its AI team via the Seed program, recruitment drives, and stock incentive plans while maintaining operations in Silicon Valley. Chinese tech firms, including ByteDance, are increasingly engaging in talent competition directly on U.S. soil, signaling the globalization of China's AI talent strategy.

Qiao Siyuan's career trajectory exemplifies the fluidity of talent between Chinese and U.S. tech companies. After completing his Ph.D. at Johns Hopkins University in 2021, he joined Google, contributing to both Google Brain and DeepMind. Earlier this year, he moved to ByteDance alongside former DeepMind VP Wu Yonghui, only to transition to Meta within months—a "triple jump" across three AI powerhouses in under a year.

Jiang Lu, another Seed team member, similarly left ByteDance to join Apple after leading video-generation AI projects. These cases highlight a recurring pattern: Chinese researchers gain experience at major Silicon Valley firms, return to China to lead AI projects, and then become highly sought-after by U.S. tech giants seeking expertise in foundational models and large-scale recommendation systems.

Song Yang's profile adds nuance to the narrative. While his work at ByteDance focused on recommendation algorithms and short-video operations, he also has a research-oriented background. Prior to joining Kuaishou and ByteDance, Song Yang served as research and engineering manager for Google AI's Deep Dialog team, working on advanced conversational systems. At Meta, his role is expected to leverage both his AI research expertise and operational experience to optimize Instagram Reels' content delivery.

Meta's hiring strategy has historically targeted OpenAI and Apple employees, but ByteDance is now a primary focus. Since July 2024, more than 800 U.S.-based TikTok employees have departed, with roughly 10% joining Meta, according to data from Live Data Technologies. For Meta, recruiting former ByteDance talent is not merely opportunistic—it reflects a broader recognition of Chinese researchers' centrality in the global AI ecosystem.

Chinese companies are also actively competing in Silicon Valley. Alibaba, for instance, has recruited AI researchers in Sunnyvale to support the globalization of its digital commerce and search engine projects, while ByteDance has established multiple U.S.-based AI project teams collaborating with teams in China and Singapore.

"Chinese researchers in Silicon Valley are highly sought after, not just for their technical skill, but for their familiarity with both Chinese-language and international business contexts," said Michael Parekh, a Silicon Valley investment advisor. "Many receive frequent recruitment calls from ByteDance, Alibaba, and Tencent, often with compensation offers that exceed their current salaries at American firms like Meta."

Meta's aggressive hiring tactics, however, are not without challenges. Reports indicate that some researchers who joined Meta's new Super Intelligence Laboratory (MSL) left within months due to mismatched expectations or cultural fit issues. OpenAI CEO Sam Altman criticized Meta's approach in an internal memo, warning that "high salaries alone can bring about deep-rooted cultural issues" and emphasizing that long-term impact depends on mission-driven teams rather than mercenary-style compensation.

Nonetheless, Meta's financial power remains a potent tool in the talent war. The company has reportedly poached at least 16 top scientists or engineers from OpenAI, Apple, Google, and Anthropic, raising overall compensation levels across the AI industry. Some offers have been labeled "free agent–level contracts," personally approved by Zuckerberg, underscoring the lengths to which Meta will go to secure elite talent.

In response, ByteDance has intensified retention and recruitment incentives. The Seed department recently introduced a stock option program granting long-term equity to researchers and engineers working on foundational models, with monthly vesting over 18 months. The "Doubao Long-Term Incentive Plan" further expands coverage, linking LLM and AI employee rewards to company performance and valuation.

ByteDance has also pursued external recruitment aggressively. In mid-2025, the company launched a global recruitment campaign, offering more than 5,000 positions across R&D, product, and management roles. Demand for expertise in large language models and AI research has increased 23% year-over-year.

Company leadership has become personally involved in talent acquisition. Reports suggest that Zuckerberg engaged directly in recruitment calls with top Chinese AI researchers, while ByteDance founder Zhang Yiming has actively participated in headhunting Chinese engineers from competitors. Earlier this year, Zhang appeared at the opening ceremony of ByteDance's Zhichun Innovation Center in Shanghai, signaling the company's commitment to cultivating the next generation of AI talent.

The competition for AI talent reflects broader geopolitical and economic stakes. The White House Council of Economic Advisers highlighted in a January 2025 report that human capital is central to maintaining U.S. AI leadership, yet domestic supply falls short of demand. Nvidia CEO Jensen Huang noted that approximately half of the world's AI researchers are Chinese, underscoring the critical role Chinese talent plays in the U.S. AI ecosystem.

At Meta, Chinese researchers constitute a significant portion of elite teams. Data compiled by Silicon Valley investor Deedy shows that among 44 top "geniuses" in MSL, half were originally from China. In the early days of the division, seven of 11 core members earned their undergraduate degrees at Tsinghua University, Peking University, and other top Chinese institutions. This "internal circulation" within the Chinese AI community illustrates the increasingly transnational nature of the talent pool.

The cross-border competition between ByteDance and Meta is reshaping the global AI labor market. Chinese companies are not only defending against poaching but actively expanding into Silicon Valley, while U.S. firms leverage unprecedented financial incentives to secure elite researchers. Salaries for senior AI engineers in China now average more than $160,000 annually, narrowing the gap with Silicon Valley by roughly 15–20%.

For both sides, the stakes extend beyond corporate growth. Talent is a strategic resource underpinning AI innovation, economic competitiveness, and technological leadership. In the coming years, the "never-ending stream" of Chinese researchers navigating between China and Silicon Valley is expected to remain central to the sector, influencing product development, AI research agendas, and international collaboration.

The departures of Qiao and Song are more than individual career moves—they are emblematic of a larger, systemic battle for human capital in AI. As the global competition intensifies, the flow of talent between China and the U.S. will continue to shape the trajectory of the AI industry for years to come.