Yang Zhilin, a 34-year-old researcher trained at Carnegie Mellon University, founded Moonshot AI and created Kimi K3, an open-weight artificial intelligence model that matches the coding and autonomous capabilities of leading American AI systems while costing significantly less to operate. His achievement this week has reignited a conversation among Silicon Valley leaders about whether restrictive US immigration policies are pushing brilliant international talent toward building AI companies elsewhere.

Who Is Yang Zhilin and How Did He Build Moonshot AI?

Yang was born in 1992 in Shantou, a city in China's Guangdong province, and attended Tsinghua University before earning his doctorate at Carnegie Mellon, where he studied under prominent AI researchers Ruslan Salakhutdinov and William Cohen. During his time in Pennsylvania, he interned at Google Brain and Meta, and co-authored several research papers on how language models handle context and prompt optimization techniques.

Before launching Moonshot AI in early 2023, Yang contributed to major AI projects in China, including Huawei's PanGu model and the Beijing Academy of Artificial Intelligence's Wu Dao, a large-scale multimodal AI system. He also co-founded Recurrent AI, a startup that used artificial intelligence to analyze sales conversations and help companies improve performance.

Salakhutdinov, Yang's former PhD advisor, praised his former student's decision to return to China and build his own company. "I remember him telling me that if he didn't at least try starting his own company, he would regret it for the rest of his life. I respect that, and he was right," Salakhutdinov wrote on social media.

What Makes Kimi K3 a Game-Changer in AI?

Moonshot AI first gained attention by giving its Kimi K2 foundation model an unusually large context window, allowing it to process a trillion parameters of information. The company has since expanded into coding, research, and autonomous AI agents, attracting backing from major Chinese investors including Alibaba and Tencent.

Kimi K3 represents a significant leap forward. Vercel CEO Guillermo Rauch noted that it marked "the first time that an open model is ahead of all proprietary ones for this comprehensive web engineering benchmark," while Wharton professor Ethan Mollick called it "closest to the frontier yet". The model's coding and agentic capabilities now rival those of leading American models from OpenAI and Anthropic, but at a much lower cost.

Guillermo Rauch

Yang's philosophy on building AI systems emphasizes efficiency through scale. "If you can solve it with scale, don't solve it with a new algorithm. The new algorithm's value is to enable better scaling," he told journalist Xiaojun Zhang in an interview. The team behind Moonshot AI developed multiple AI technologies that contributed to Kimi K3's success, including Transformer-XL, RoPE, Group Normalization, ShuffleNet, MuonClip, and Mooncake.

How to Understand Moonshot AI's Technical Achievements

• Context Window Innovation: Moonshot AI's models can process an unusually large amount of information at once, allowing them to handle lengthy documents and complex tasks that require understanding long sequences of text.
• Open-Weight Model Design: Unlike proprietary models kept behind closed doors, Kimi K3 is released as an open-weight model, meaning researchers and developers can access and build upon the underlying technology.
• Cost Efficiency: Kimi K3 delivers capabilities comparable to expensive proprietary models from American labs while operating at significantly lower computational and financial costs.

Why Are US Tech Leaders Concerned About Yang's Success?

Several prominent figures in the American tech industry have expressed regret that Yang did not remain in the United States to lead an AI lab there. Legendary investor Vinod Khosla blamed the Trump administration's restrictive immigration policies, stating: "Even bigger issue is the brilliant talent we are scaring away from other countries with our immigration policies for great talent".

The Trump administration has implemented several measures that make it harder for international talent to work or stay in the US. These include a $100,000 fee for employers sponsoring certain new H-1B visa applications for foreign workers, which was later struck down by a federal judge but remains in litigation. In May, US Citizenship and Immigration Services issued a memo implying that people who could previously apply for permanent residency from inside the US may now have to leave the country while their case is being processed. This week, the administration introduced a new rule that puts an expiration date on how long people on student visas can initially remain in the US.

"He is absolutely brilliant," Salakhutdinov said of Yang in a post on social media on Friday.

Ruslan Salakhutdinov, AI Researcher at Carnegie Mellon University

However, Salakhutdinov pushed back against the narrative that immigration policy alone drove Yang's decision. While acknowledging that the US immigration process can be "intimidating" and "uncertain," Salakhutdinov emphasized that Yang was always determined to return to China and build his own startup.

Yang himself has expressed ambitious long-term vision for the AI industry. "The ultimate AGI company will dwarf today's giants, double, triple the scale. Not necessarily OpenAI, but such a company will exist," he said in his interview with Zhang. His success with Kimi K3 suggests that such a company may already be taking shape in China, challenging the assumption that American labs will maintain dominance in artificial intelligence development.