Wei Haijun's AI globalization talk: I really want to hear how he dared to invest in Musical.ly back then

An AI incubator founder hiring a mobile internet-era overseas expansion investor as a mentor sounds a bit time-travel-ish at first glance. But thinking closely, what AI global expansion lacks most right now might really not be technology, but that set of old-school common sense.

Yusheng Science recently brought Wei Haijun from Daguan Capital onto its mentor lineup to guide teams expanding overseas. The highlight of his resume is leading the angel round investment in Musical.ly while serving as Head of Investment at Cheetah Mobile in 2014. To put that in perspective, Musical.ly had only 70,000 registered users at the time. Today, that data wouldn't even pass the threshold for any AI startup's pitch deck.

But he invested anyway, and later Musical.ly became the foundation of TikTok.

I remember a quote from Wei Haijun well: Instead of expecting shortcuts or quick fixes, return to business common sense and steadily advance international innovation based on local environments, international situations, and corporate realities.

This sounds like correct nonsense, but if you've actually worked on the front lines of global expansion, you know how scarce common sense is. When I was working on medical AI products, my biggest fear was clients saying, "This works in the US, just port it over." Clinical pathways, compliance requirements, and doctor usage habits differ in every market. Porting it over leads to incompatibility. AI global expansion follows the same logic: growth models that work domestically might not even match user registration rates in Southeast Asia.

Global Expansion Isn't Just Translating the Product

In interviews, Wei Haijun distinguished expansion paths for companies of different sizes: Large enterprises should treat internationalization as a core strategy, carefully selecting starting points for global expansion; mid-sized companies need to break mental barriers, treating overseas markets equally with domestic ones; startups should leverage the development of the internet and AI to explore diverse paths.

This stratification is very product-manager-like. Different stages of products have completely different core metrics. Big tech looks at brand premium and ecosystem position when going global; mid-tier looks at user growth and retention; startups look at PMF and iteration speed. Using one methodology for all stages is like applying a top-tier hospital's treatment plan to a community health center—it's bound to fail.

I've seen too many AI startups go global by first hiring translation agencies to localize the UI, then buying traffic, and then... nothing. How is that different from not expanding? It's just running the domestic playbook in a different language environment. True localization means understanding why local users use your product, not whether your product can be opened in their language.

The Other Side of Common Sense is Calmness

Yusheng Science itself is interesting. The mentor list includes heavyweights from Youzu, UCloud, Waterdrop, and Morgan Stanley, plus Chen Danian holding things down. The lineup is indeed luxurious. But having many mentors doesn't equal high incubation success rates; the key is whether these mentors provide real resources or just lend their names.

What someone like Wei Haijun, who has experienced full cycles, brings to AI entrepreneurs might not be the latest tech trends, but rather the calmness of "I've seen this before." For example, a common trap in AI global expansion is founders getting carried away by validated domestic data, assuming larger overseas markets mean better performance. But when Wei Haijun looked at Musical.ly, 70,000 users convinced him to invest, showing he valued product logic and user stickiness, not short-term data.

Back to AI global expansion. Many current AI startups validate a logic domestically and rush abroad. But regulatory environments, user payment habits, and data compliance requirements overseas are entirely different systems from China. If you take the domestic "run first, comply later" approach abroad, you might trip a wire in step one.

Wei Haijun said the founder's mindset and determination are crucial. In the AI context, I interpret this as: Don't get blinded by technical narratives. Market research, compliance reviews, and localization adaptations cannot be skipped.

My deepest takeaway from working on AI imaging at United Imaging is that good products aren't born in labs; they're forged by shadowing doctors through surgery after surgery. AI global expansion is the same: it's not written in code, but gnawed through market by market.

Bringing Wei Haijun into Yusheng Science is at least the right direction. AI incubators don't lack technical mentors; they lack people who have actually gone global, stepped on traps, and know where business common sense lies. Looking ahead, the deciding factor in AI global expansion likely won't be model parameters, but who realizes sooner that technology can be replicated globally, but business must grow locally. This insight is worth more than any funding news.