China’s Industrial Software Gap - Trust, Margins, and Customer Concentration

After moving back to Shenzhen, I realized China has many more problems worth solving than I expected. I have been in Shenzhen for almost a month now, and have met some factory owners and people in the supply chain. I wanted to write down a few thoughts.

1. Dongguan factories are deeply tied to large domestic customers

Even for a listed company making advanced materials, a high technical barrier does not necessarily mean real pricing power. Based on last year’s financial report, more than 90% of this company’s revenue came from its top five customers, and all revenue came from China. More extremely, the top two customers alone contributed more than half of revenue, accounting for 39.03% and 33.72% respectively. This means the company is still deeply pulled by the procurement cycles and project needs of a small number of large customers. If a major customer switches suppliers, pushes prices down, develops the capability in-house, or changes its project cycle, the company’s revenue could be directly pulled away.

More importantly, this structure can temporarily hide many hard problems. When there are few customers, concentrated scenarios, and a purely domestic market, regulation, certification, standardization, and sales can all look easier than they really are. But that does not mean the company already has the ability to replicate across customers, industries, and markets.

2. In the U.S., I felt there was nothing to do. After returning to China, I feel there is a lot to do, but many real-world constraints hold things back

When I was in the U.S., I often felt that many things would soon be replaced by AI agents. Many software products were already very well built. The U.S. also keeps innovating, so new markets become competitive very quickly.

After coming back to China, I found the opposite feeling.There are many things one could do. But these opportunities are not as simple as “there is demand, so you can build and sell a system.” There are many constraints in the middle.

First, factories have low margins and are extremely cost-sensitive. Factory margins are generally thin, even in high-precision manufacturing. Competition is intense, prices are transparent, and differentiation is often not large enough. Because of this, owners usually will not easily spend a lot of money to hire people, or buy a system whose output is uncertain. This is understandable. A factory is an asset-heavy business. Even if it does nothing on a given day, it is still continuously burning cash. As long as the return is uncertain, any additional cost will naturally be treated with caution.

Second, trust costs are very high. If a forward deployed engineer or outside consultant enters a factory and gains a deep understanding of its core processes, customers, pricing, and supply-chain relationships, the owner will naturally worry: Will you go to a competitor a few months later? Will you secretly sell my factory’s information?

This kind of thing has happened too often in China, which is why you see many different stores and factories selling almost the same things. Behind this is not only a lack of innovation, but also a problem of business trust and boundaries. On one hand, this problem requires the government and legal system to give business owners more protection. But it also should not suppress innovation through non-compete agreements in the way many internet companies have done. There should be a better balance.

At the same time, the culture itself also needs to change. Competition should not always default to close-quarters, zero-sum imitation. When someone is already working on a problem, the better first question is: What needs are still unmet? Unless you can be dramatically better, it is usually wiser to find a different direction.

That is the biggest difference I feel after coming back: China still has many real-industry problems worth solving, but the opportunity is not simply to build better software. The hard part is finding a narrow vertical where cost, trust, incentives, and daily business reality actually allow a product to work.

I’m still trying to understand what those verticals look like in practice. If you are building or working in manufacturing, supply chain, or industrial software in China, I’d be curious to hear what problems you think are genuinely worth solving—and what constraints make them difficult.