File Under: Understanding business in China and why it’s more alien than you think. Part of a project with Coco Liu around Chinese entrepreneurship.

I.

My friend manages intellectual property for a British semiconductor company. He’s the “China guy”—handles the accounts there, flies out to Shenzhen three to four times a year. He’s been making the trip for the last three years.

One day, he was visiting a customer and immediately sensed that something was off.

150 engineers were sitting at desks, spinning lazily in their chairs. Some were literally twiddling their thumbs. More than a few sat in front of gently rolling screensavers. It was quiet in a way offices shouldn’t be quiet.

Not focused quiet.

Empty quiet.

This didn’t make sense. China is the country of 996—9 a.m. to 9 p.m., six days a week. This isn’t a laidback part of the world. Semiconductors are the technology that allows AI to work. And AI is the most intense, high-stakes industry on the planet right now. Which is why Nvidia is practically holding up the American stock market. Every major economy is racing to secure chip supply chains. This should be a pressure cooker.

So why was everyone just sitting around?

He asked his host what was going on. The answer he got required understanding how fundamentally different Chinese business operates from the rest of the world.

II.

He learned that the semiconductor industry in China has split into two ecosystems: privately funded companies and government-funded companies. It used to be maybe 80% private, 20% government. Today, it’s flipped—closer to 20% private, 80% government. This happened after the U.S. passed restrictions limiting American capital flowing to Chinese VCs. Private money dried up. Public money rushed in to fill the gap.

But public money in China works differently. Rather than a robust national level-funding like in the United States, it’s handled almost entirely by provincial governments. Each province competes with the others to fund startups, trying to look good to the central government by hitting national targets: proliferation of 5G, electric vehicles, advanced manufacturing.

And the provinces compete hard. Startups will shift their legal registration and even rebrand themselves to align with whichever province just funded them. My friend told me this happens repeatedly. Provinces compete fiercely for prestige, and companies play musical chairs, moving headquarters and renaming themselves to follow the money.

For the provinces, success has two metrics: achieving growth milestones and creating jobs. Job creation tends to be more important than technical goals. As long as a startup exists, employs several hundred people, and keeps those people on payroll, that’s often enough. Whether the startup is actually making progress? Secondary.

III.

The company my friend visited had raised over $100 million from the provincial government to build sophisticated chip designs. Work that requires operating at nanometer precision, manipulating matter at scales most people can’t conceptualize. This is some of the highest of high technology.

One year in, they still hadn’t purchased the software design tools their engineers needed to do their jobs. In the semiconductor industry, you need specialized software—EDA tools (electronic design automation software)—the same way an architect needs AutoCAD. Without these tools, you literally cannot design chips.

So these engineers were just waiting. Sitting there. Still getting paid.

In the United States, a $100 million semiconductor startup resembles an engine pushed past the redline. Every engineer would be sprinting to hit milestones and targets. The idea of paying 150 engineers to sit idle for a year and a half because you haven’t bought them the tools to work would be insane. The board would riot. Investors would pull out. The company would collapse.

In Silicon Valley, you measure startups by product-market fit, burn rate, user growth. In China’s government-funded ecosystem, the metrics are different: jobs created, provincial prestige, alignment with national goals. My friend watched an entire floor of engineers sit idle for a year and a half because those metrics had already been satisfied. The company looked successful. And by the measures that mattered to the people funding it, it was.

The question of course, is: how long will it stay this way?