Silicon Valley Thinks China’s Top AI Models Came Out of Nowhere — Just Check X to See the Full Story

The dominant narrative in Silicon Valley frames leading Chinese AI models like Kimi K3 and DeepSeek R1 as sudden, out-of-nowhere surprises for the global tech industry. But the truth is, you don’t need to visit China to follow what Chinese AI researchers are working on — all you have to do is log on to X. Over the past year, I’ve watched a steady, growing stream of Chinese AI researchers join the platform and engage with the global conversation around AI development and commercialization.

Take Moonshot AI, the company behind Kimi K3, as an example. Just this past Thursday, I quickly located roughly 30 active X accounts run by people who report current affiliation with Moonshot. That includes two of the startup’s co-founders, plus half a dozen former employees and outside collaborators. These are not dormant, forgotten profiles: team members share updates on Moonshot’s major launches and new incremental research papers, build connections with Western AI peers, and post details about their personal lives and hobbies off the clock.

Moonshot is arguably the most active Chinese AI lab on X, but you can easily find executives and technical staff from other top Chinese AI startups including Minimax and Z.ai on the platform too. Even several DeepSeek employees — who are reportedly barred from leaving China after authorities seized their passports — post regularly on X, commenting on the latest AI industry news and sharing open job listings to recruit new talent.

“X is the best community for AI discussions right now,” says Meng Fanqing, co-founder of startup Evolvent AI and a former Moonshot intern who worked on the Kimi K2.5 model. “You get both Chinese and overseas audiences, and the recommendation algorithm does a really good job of showing you content from your professional circle.” Compared to domestic Chinese platforms, where serious technical debates are often hard to sustain, Meng says researchers like him prefer to spend their time on X.

Meng began posting actively on X in mid-2025, just after DeepSeek’s R1 model went viral worldwide and drew global attention to China’s AI sector. Many Chinese researchers seized that moment to join X and start sharing their own perspectives. “Around that time, all Chinese researchers started thinking about building international brands for themselves,” Meng says.

Tiezhen Wang, an independent AI analyst and former Hugging Face researcher, notes that Chinese researchers are also filling a growing gap left by teams at OpenAI and Anthropic. As those U.S. firms have grown into massive industry behemoths, their in-house researchers have become far more reluctant to share their thinking publicly online. “My sense is that they see their model architectures and training processes as trade secrets now, so they post a lot less,” Wang explains. As a result, Chinese researchers are drawing more attention from X audiences hungry for insight into the fast-moving global AI sector.

Why X?

One core reason X has become popular among Chinese AI researchers is that there is no comparable domestic social platform for high-quality technical discussion about AI and other emerging technologies. Many researchers say they used to post on Zhihu, a Quora-like platform that once branded itself as the top destination for educated professionals to share ideas. But since 2020, Zhihu has increasingly shifted toward hosting casual fiction and viral low-effort content, pushing expert contributors away gradually. For example, Fields Medal-winning Chinese mathematician Deng Yu publicly announced back in 2018 that he would stop posting on the site due to the proliferation of low-quality content.

Xiaohongshu (also known as RedNote) is another major Chinese social platform that has grown popular among young, urban, educated Chinese users. Some junior AI researchers have opened accounts there recently to share work updates, but the platform’s overall audience is far less technically focused than X’s. When Moonshot published a major new research paper in March, it sparked buzz across both Chinese and Western social media, even earning praise from Elon Musk. But “almost everyone on Zhihu and Xiaohongshu was only talking about the authors’ backgrounds, ages — one author is just 17 — and their parents, all really superficial gossip,” says a current Moonshot researcher who requested anonymity, as he is not authorized to speak to media. “It was only foreign readers on X that actually discussed the technical details and praised the paper’s breakthroughs.” He added that Zhihu commenters also tend to be more hostile, and the platform is plagued by far more low-quality AI-generated spam than X. Moonshot did not immediately respond to a request for comment.

These new Chinese contributors have largely received a warm welcome from Western AI academics and industry researchers already on X, many of whom have long wanted direct access to information from Chinese AI labs. “Before all these researchers joined X, we already had many influential researchers translating Chinese AI blogs into English for the global community,” Wang says. He points to Su Jianlin, a well-respected Chinese AI researcher who long ran a popular Chinese blog sharing his industry insights — exactly the kind of voice Western AI communities have long wanted direct access to. Today, Su works for Moonshot and also maintains an active public X account.

Other Key Incentives

Beyond the ability to join open technical discussions, Chinese AI companies have also found X is an effective place to market themselves and their products globally. “X is the platform with the highest traffic and global exposure. To some extent, the AI industry works just like finance now: every individual and every company needs to build their own public brand,” Meng says. When AI researchers gain a small following as social media personalities, that also boosts their employer’s public profile.

Wang says he even advised Moonshot during a one-time 2025 consulting engagement to encourage more of its researchers to be active publicly on social media. When a startup releases a new model or paper, “it’s far more engaging to see what the individual researchers working on it want to post, instead of just reading a formal official announcement,” Wang says. “This also definitely helps them gather more useful feedback from audiences, connect with other researchers, and potentially spark new collaborations.”

Moonshot has already proven this approach works, thanks to its roster of charismatic team members. Chinese media often describes the company as one of the AI labs with the most unique internal culture: its CEO covers his office in band posters and other memorabilia celebrating his love of rock music, and the company’s Chinese name 月之暗面 is directly taken from Pink Floyd’s classic album The Dark Side of the Moon.

Wang argues that having Chinese AI researchers explain their work in their own words on X is a net benefit for the global AI community, including the U.S., because it cuts down on uninformed speculation about what Chinese AI labs are doing. “People have a lot of wrong assumptions about what China’s AI sector is doing. Right now, X is the door that lets them understand China better,” Wang says.

This is an installment of Made in China, the newsletter by Zeyi Yang and Louise Matsakis. Read previous editions here.