A Hong Kong Founder on China's AI Edge

Today I had a coffee chat with a college friend who runs an AI startup in Asia, in Hong Kong, rather than in the US. My friend is out there surviving alone and building directly with AI. Here are my notes from someone who is experiencing AI in a different country from Korea, in a different way.

1. Hong Kong is an AI Galapagos

Surprisingly, you still can't use Claude, ChatGPT, or Gemini in Hong Kong. You can get around it with a VPN, but then you need an American credit card. There are even services that buy subscriptions on your behalf for people who don't have one. Most people in Hong Kong don't have a VPN either, so they use workaround services like Poe.com or ChatLLM.

I once had a similar conversation with another friend who does consulting in Hong Kong. They described Hong Kong as an AI Galapagos: you can't use American LLMs, and Chinese LLMs aren't easy to use there either. I was struck that there's still a place where Gemini and Claude aren't available. For us, these are tools we take completely for granted. On the other hand, DeepSeek, a Chinese LLM, does work there, and on the Chinese side people also use Kimi, Manus, and ByteDance's Seed Dance and Rare Stack. My friend said Chinese LLMs give much better insight than Claude for research on Asia.

2. Non-developers now build alone

What stood out most was that my friend, a non-developer, is building everything solo thanks to AI. My friend left a crypto company last November and got a grant of about $8,000 from the Hong Kong government. That covers a full year of AI costs with money to spare. With it, my friend handles marketing automation, product development, and pitch decks all alone. A presentation in five minutes, a prototype built on the spot. The era of the one-person company really is here.

Friends were so impressed watching my friend build things firsthand that it turned into an AI class my friend now runs. In April there's a consulting project lined up to upgrade a Chinese manufacturer's internal operations with AI. The fun part: my friend built a maritime intelligence tool for a friend at a law firm, and that friend's reaction was, "We were thinking about hiring developers, but it was too expensive. If you've already built it, we'll license it." We now live in a world where a non-developer who knows the domain and builds with AI is faster and more practical than adopting enterprise AI or hiring developers.

Still, my friend said AI adoption outside the tech community is pretty slow. Those of us inside are all moving fast, but at ordinary companies and in non-tech fields, it's still at a beginner level. That's why my friend called this the golden window. Not enough people are using these tools yet, so most people have no idea how easy this can be. Who knows about a year from now, but right now there's clearly an opportunity for whoever moves first.

3. How China develops differently

I was curious why China can develop AI on its own terms. My friend pointed out a few things. First, the Hong Kong and Shenzhen Greater Bay Area ranks first in the world in patents and first in robotics research. The government has poured huge investment into research and robotics for more than ten years.

Then there's the insane level of competition inside China. If you build something innovative in Hong Kong, a Chinese company copies it and pushes it faster, bigger, and harder. Hong Kong used to have Foodpanda and Deliveroo, but then Keeta, run by Meituan, came in and pushed Deliveroo out completely within a year, offering free delivery for every neighborhood shop. That, my friend said, is the Chinese way of doing business.

There are three ways to win: be faster, be cheaper, or offer something more end to end and vertical. Storytelling and branding are things only Americans think about. And among 1.4 billion people, there are so many who will do whatever it takes to survive. Go out on the street in China and it's overflowing with delivery riders, everyone hustling to get by. In the US there's still a sense of "I want to get out of this blue-collar work." In China the mindset is simply to work hard, make money, and survive.

Above all, thanks to WeChat, the whole country is smartphone native. Twelve years ago, even in the countryside in the middle of Hunan province, grandmothers and grandfathers were already using WeChat Pay, making video calls, and collecting coupons through online games. Credit cards spread late, but people skipped that stage entirely and jumped straight to mobile payments. So when news came out recently that the Chinese government is installing OpenClaw for rural farmers, the basic access to technology across the population was already in place, more than you'd expect.

4. The soft power wall

China's weaknesses were clear too. My friend said Chinese people themselves know all too well that the country lacks soft power. The technology is ahead, but trust and branding can't make it out of China. My friend described the UI and UX of Chinese apps as distinctly Chinese, and for overseas consumers that becomes a barrier. In B2C, Korean and Japanese brands still carry a big premium. In Shenzhen, people line up for Korean-style bakeries, and in Hong Kong, French bakeries are popular. Coming from abroad is a premium in itself.

That's leading to an interesting phenomenon. As Hong Kong has become more Chinese lately, foreign capital has started to shy away from investing in Hong Kong companies. So Hong Kong companies set up headquarters in Korea, Japan, Singapore, or Taiwan first and then enter Hong Kong in reverse, a kind of white labeling. They package themselves as not Chinese to get foreign investment, and at the same time they wear the premium of a Korean or Japanese brand to stand out in the Chinese market.

5. Bridging Europe and Asia

My friend runs a community for women founders, and they once invited a trademark lawyer to speak. What came out of that was interesting too. Hong Kong and Chinese companies register trademarks and launch in Korea or Japan first, then enter China and Hong Kong in reverse. That way they can get foreign investment and also stand out inside China.

The European side was interesting as well. My friend went on a business trip to the UK last October, and British brands are obsessed with the US market. The default strategy is to launch, roll out nationwide, and then go to America. They have zero interest in the Asian market. But as my friend sees it, a lot of what's already a trend in the West is still an open market in Asia, and Asian spending power keeps rising, so there's real opportunity. My friend told British brands to consider Asia, and most didn't listen. Asia is too diverse, the languages differ, and there's a perception that purchasing power has traditionally been low. That's changing, and they just don't know it yet.

As a result, Hong Kong and the rest of Asia have a lot of marketing agencies run by French and Italian people, and they have a near monopoly on helping European brands enter Asia. My friend thought this market could also be flipped by AI.

6. In the end, only humanity survives

The conversation ended up at the same conclusion. No matter how much AI raises the quality of what gets produced, people consume more than the output. They consume the person behind it. AI content is going to pour out even more from here, but in the end people consume the artist's story along with the content.

My friend reached a similar conclusion. Moving from crypto to AI, my friend realized that even if AI takes over execution, only people can create IP and brands. My friend is now building original IP that draws on their own domain knowledge. AI helped with the development, but the idea and the storytelling are my friend's own, and in the AI era that will be an even bigger differentiator.

People's experience, their humanity, their brand. AI can't copy these no matter how far it advances, and their value can only go up from here.

Writing the Claude Blue series, I seem to land on a similar conclusion every time. But when the person I'm talking with is different, the same conclusion feels new. Watching a Hong Kong founder's survival-driven pivot, I felt that even in an AI Galapagos, the people who move first end up making a path.

As it happens, an American outlet reached out for an interview after reading the Claude Blue post I wrote on Reddit. I also recently did an interview with Korea's Dong-A Ilbo, a major daily newspaper, and it was my first interview as an individual rather than on behalf of the company. It's still a small independent American outlet, but if content keeps piling up like this, maybe someday I'll be in a magazine like The New Yorker. All I can do is keep building the story.

Originally written in Korean on Brunch, March 31, 2026.