Why AI Isn't Making You Money: It Grows One Order at a Time

A lot of people say they can’t make money with AI. Most of them aren’t failing — they just don’t think the money is worth their time. Using AI to write songs, make images, restore old photos, edit talking-head videos — open a shop on a second-hand marketplace and it’s like setting up a street stall. Everyone can do it. They just look down on the money.

A college student went home for Chinese New Year and set up a laptop at a mall entrance doing AI portraits — ten to fifty yuan per order, ten to twenty orders a day. From New Year’s to the Lantern Festival, he made over ten thousand yuan, at very low cost. A second-hand marketplace published its numbers: roughly ten million AI service orders in the first half of this year, but sellers averaged less than 900 yuan a month. The money is real, just not a lot. This is earning through labor — same as driving ride-hailing or delivering food, except you don’t have to brave the elements. It’s just that the repetitive work is tedious.

What do people who look down on small money end up doing? One group goes after something that looks like big money — learning to make AI comic dramas. The models are so capable now that everyone’s output looks about the same unless you’re exceptionally skilled, which drives everyone into the same tier of price competition. And you don’t even know if you’ll make money when you’re done. A media outlet interviewed Mr. Huang, who was unemployed and spent 2,680 yuan on a course to learn AI comic drama production. His videos earned him 0.1 to 2 yuan each. In his own words: he wanted to learn a trade to earn some living expenses, and ended up spending his living expenses on someone else.

Another group always wants to hold out for something big — build a grand product. When it’s finished, the market is flooded with similar things. Meanwhile, the small jobs that pay you the same day you do them? Looked down on. People pass them up to bet on something they may never earn a thing from.

The order should actually be reversed. Taking orders means the demand comes first: someone finds you, pays you, and then you deliver. Take editing talking-head videos — that work goes for at least ten to twenty yuan per minute. At first the costs aren’t low either, and you barely clear a few yuan per minute. Once you’ve got it running, the volume picks up, costs come down, and you’re making several more yuan on every minute. A video maker in Xuzhou started with it as a side gig, earning three to four thousand yuan a month. When orders piled up beyond what he could handle, he taught others and farmed out the work, taking a cut — earning more than when he did it all himself. Eventually his AI income surpassed his main job. Steady order flow means you’ve cracked customer acquisition. Add automation, and it’s a real small business.

AI is an amplifier. People who were already making money don’t need AI to keep making money. People who weren’t making money won’t start just because they learned AI. Amplify zero, and you still get zero.

Start by doing services, delivering with human effort. Find the business first, nail the business, and then use AI to accelerate, to scale up, to automate. Look down on this, look down on that — in the end, nothing ever gets started. Money doesn’t come from holding out for something big — it grows one order at a time.

EP0104 · 2026-09-23 · 02:41. The picture uses a digital human, and the voice uses my cloned voice. This is an English translation of the approved Chinese script used to create the narration.

A lot of people tell me they can’t make money with AI. Here’s what I’ll tell you — most of them aren’t failing to make money. They just don’t think the money is worth their time.

Using AI to write someone a song, using AI to make images, all those AI old photo restorations, using AI to edit someone’s talking-head videos. Open a shop on a second-hand marketplace, and it’s like setting up a street stall. You know how to do this, right? Of course you do. You just don’t think the money is worth it.

There’s a college student who went back to his hometown for Chinese New Year, set up a laptop at a mall entrance doing AI portraits. Ten to fifty yuan per order, ten to twenty orders a day. From New Year’s to the Lantern Festival, he made over ten thousand yuan. He said it himself — the costs were really low.

A second-hand marketplace published its own numbers: in the first half of this year, AI service orders hit roughly ten million, but sellers averaged less than 900 yuan a month. The money is real, it’s just not a lot — so people look down on it. This is earning through labor. It’s the same as driving ride-hailing or delivering food, except you don’t have to brave the elements. It’s just that the repetitive work is tedious, and you’re simply not willing to grind for that kind of money.

So what do people who won’t earn small money end up doing? One group goes after things that look like they could bring in serious cash — like learning to make AI comic dramas. The models are so powerful now that people of different skill levels, unless someone’s truly exceptional, all produce pretty much the same stuff. Everyone ends up in the same tier, competing on price. And when you’re done, you don’t even know if you’ll make any money. One media outlet interviewed a Mr. Huang, who was unemployed and spent 2,680 yuan on a course to learn AI comic drama production. The videos he made? Each one earned him 0.1 to 2 yuan. In his own words: he wanted to learn a trade to earn some living expenses, but ended up spending his living expenses on someone else.

Then there’s the other group — always wanting to hold out for something big. Build some grand product: AI presentations, AI this product, AI that product, AI investing. But once it’s done, the market is already flooded with similar things.

Those small jobs from earlier — you do them today, you get paid today. People look down on that, then turn around and bet on something they don’t even know will ever pay off.

The order should actually be reversed. When you take orders, the demand comes first, then the service — someone comes to your door, pays you, and then you do the work. Take editing talking-head videos again — that work goes for at least ten to twenty yuan per minute. At the start, costs aren’t low either; you barely clear a few yuan per minute. Once you’ve truly got it running, the volume picks up, costs come down, and every minute earns you several yuan more. There was this interview — a video maker in Xuzhou started as a side gig, making three to four thousand yuan a month. Later the orders piled up beyond what he could handle, so he taught others, farmed out the work, and took a cut — and actually made more than when he was doing it all himself. After that, his AI income surpassed his main job. When orders come in steadily, it means you’ve cracked the customer acquisition chain. Add automation development on top, and you can really call it a small business.

I said in an earlier video that AI is just an amplifier. People who were already making money don’t need AI to keep making money. People who weren’t making money — learning AI doesn’t change that. If you want to amplify something, you need something turning first. Amplify zero, and you still get zero.

So where does the turning thing come from? Start by doing services, delivering with human effort. Find the business first, nail the business, and then use AI to accelerate, to scale up, to automate. Look down on this, look down on that — in the end, none of it ever gets started. Money doesn’t come from holding out for something big — it grows one order at a time.

That’s all for today. See you tomorrow. Bye.