Issue #164

Why a Freelance Coder Delivers Food Every Night

A Chengdu developer's delivery earnings match his rent to the yuan—a window into freelance wages and Korea's platform labor data.

BusinessWhy a Freelance Coder Delivers Food Every Night

Two Rooms on a Rooftop, Delivery Income That Matches the Rent to the Yuan

On the roof of an old residential building in Chengdu, China, there’s a home that doesn’t officially exist. It’s an unpermitted two-room structure built on top of the building, renting for 1,000 yuan a month—about ₩200,000 (~$144). There’s no lease; the money just gets transferred to the downstairs neighbor every month. A 28-year-old developer couple has lived there for three years now.

The husband, Jared, used to be an engineer building AI at a big tech company. After grinding through 9961 hours, he quit, and now he codes freelance projects up on the rooftop during the day. Then at 6 p.m., he puts on a delivery vest and gets on his motorcycle. Four hours of riding until 10 p.m. earns him about 1,000 yuan a month—the exact same number as his rent.

I recently watched a documentary following him through one of his days, and I couldn’t just file it away as “one eccentric young man’s story.” Watching his case made me curious about how people who make a living selling their technical skills are responding to shifts in job availability and pay rates. There’s a part of Korea’s platform-labor statistics worth looking at alongside this, too.


Coding by Day, Delivering by Night

Calling Jared’s household frugal, as shown in the video, doesn’t do it justice. In a country where 98% of urban households own an air conditioner, he has none — in summer he cools his walls by spraying them with a mister. Each meal is a boxed instant meal costing 5 yuan, roughly ₩1,000 (~$0.7). He fixes broken appliances himself and even builds his own speakers from parts he buys. With income this unstable, “don’t buy what you can make” has become his survival rule. Together with his wife, who works at a hospital pharmacy, the couple earns about $1,000 a month — roughly ₩1,400,000 (~$1,000).

deliveryWhat’s worth noting is that this income has actually fallen sharply compared with his first two years as a freelancer. The work itself hasn’t disappeared — it’s the rates that have collapsed. Jared, who used to build AI himself, points to AI as one of the causes behind the falling rates. He puts it this way: “The more I understand AI, the more it scares me. Getting it to do something has become too easy. All it takes is one prompt. And anyone who figures out that prompt can build the exact same thing.”

Since coding income alone isn’t enough to live on, he delivers food every evening. During the evening peak hours, his hourly pay runs 23–30 yuan, roughly ₩5,000 (~$3.6). His best-ever hourly rate — 50 yuan, about ₩10,000 (~$7.2) — came during a rainstorm so heavy that other riders had given up.

One scene captured during filming stood out. Jared made a delivery to a hospital, and the person who’d placed the order turned out to be a fellow rider who’d broken his leg in a delivery accident. As of May 2026, when the video was filmed, China’s occupational injury insurance2 for platform workers — essentially China’s version of workers’ compensation — was still only a regional pilot program. Nationwide rollout wasn’t set to begin until July 1, 2026. That rider was injured two months before the program took effect, and now faces months without income while covering roughly 20% of his medical costs out of pocket.

Even so, Jared doesn’t blame the platforms. “There’s an oversupply of riders, so platforms push rates down as far as they can. Cutting costs is capital’s whole job, after all.” And then he adds: “Once you’ve tasted freedom, you can’t go back.” The video adds one line here. Freedom without an economic foundation can always turn out to be fake freedom. The moment you have to go deliver food just to make rent, you’ve already surrendered part of that freedom.

Two Forces Pressing Down on Rates: Less Work, More People Chasing It

If you read Jared’s shrinking income as personal laziness, you miss the point. You need to look at both the demand for work and the number of people competing for it.

The first force comes from the demand side. AI is shrinking the pool of gigs itself. A recent study in the management journal Management Science measured this with some precision. The research team tracked 1.4 million job postings on a global freelance platform over two years starting in July 2021, and found that after ChatGPT launched, requests for jobs vulnerable to automation fell by an average of 21%. Writing dropped 30.4%, and software/web development—Jared’s line of work—fell 20.6%. After image-generation AI arrived, graphic design requests fell 17%, with no sign of recovery by the end of the study period. This data comes from an online outsourcing platform, so it doesn’t represent the entire market, but it does let researchers compare before-and-after AI’s arrival by tracking changes in postings.

The second force comes from the supply side. China’s flexibly employed population is estimated at roughly 200 million, of which about 84 million are “new-form employment” workers who get gigs through platform apps. At Meituan alone, 7.45 million riders took orders over the course of a year—and interestingly, nearly half of them worked fewer than 30 days a year. For many people, delivery isn’t a main job but something they pick up briefly when they need extra cash. With millions of people standing by, ready to jump in whenever they need money, rates are structurally hard to push up. Layer on top of this the youth employment situation. China’s unemployment rate for ages 16–24 stood at 15.6% as of May this year, and recent reports have even projected it could reach 20% as AI spreads further.

Put it all together: as AI shrinks some outsourced work, more people are turning to side gigs because their regular employment is unstable. When there are plenty of people willing to work, it becomes hard for any individual to command a high rate. Full-time riders working high frequency in first-tier megacities like Beijing and Shanghai earn a monthly average of 7,354 yuan—about ₩1,400,000 (~$1,010). That’s close to what Jared and his wife bring in each month, but the average for a full-time rider in a major city and the income of one household in Chengdu, a city in southwestern China, are cases with very different conditions.

More Freelancers, Falling Rates: Korea’s Own Version

Korean statistics also show a rising trend of people working as independent contractors or on platforms. But these numbers alone can’t tell us exactly how much AI has driven down freelance rates in Korea.

The number of income earners who report under the 3.3% withholding tax category for personal services3 — the “broad freelancer” population captured in tax statistics — grew from 3.26 million in 2008 to 8.47 million in 2022. This happened even as the share of wage employees in employment statistics actually rose over the same period. Labor market researchers read this gap as a signal that “employment-like non-employment” is expanding: people who work as if they’re inside a company but are contracted as individual business owners, quietly multiplying in the background.

Platform workers also grew — reaching 883,000 as of 2023, an 11% increase in a single year. What’s worth noting is the composition. Delivery and driving actually declined, while workers in IT and professional services surged 141%. That means platform labor is expanding not just in delivery but in IT and professional services too. Meanwhile, the number of people juggling a side gig on top of their main job — so-called “N-jobbers” — hit roughly 680,000 in 2025, an all-time high.

The delivery scene looks even more like Chengdu. Korea’s delivery market hit an all-time high of roughly ₩40 trillion (~$29 billion) in 2025, yet riders on the ground report that their actual take-home pay per delivery has fallen from the ₩5,000 (~$3.6) range down to the low ₩3,000s. In a large-scale 2023 survey, 63% of riders said their income had declined. The market grows bigger while the actual share going to the people doing the delivering shrinks — the exact same structure I saw in Chengdu.

Of course, there are differences too. Korea moved faster on policy. In July 2023, Korea scrapped the “exclusivity”4 requirement for workers’ compensation insurance, extending coverage to most delivery riders — three years ahead of China, which only expanded its program nationwide in July 2026. But being ahead on policy doesn’t mean the pricing structure is any different. The time lag between what’s happening in Chengdu and what’s happening in Korea looks like a matter of a few years at most — maybe even less.

Finally, I want to point out one scene that flatly contradicts these numbers. These days, Instagram and Threads are flooded with people calling themselves “solopreneurs,” flaunting proof of ₩10 million (~$7,200) months and revenues in the hundreds of millions of won a year. Bios describing them as mentors, high-net-worth individuals, or successful CEOs are common too. But if you look at the closest thing we have to real statistics on this group — the National Tax Service’s reporting data on solo media creators — for income attributed to 2024, only 34,806 creators reported any income at all. The bottom 50% earned an average of ₩24.63 million (~$17,900) a year, which works out to roughly ₩2 million (~$1,450) a month in gross revenue — before expenses. The top 1% — just 348 people — took home 18% of all income. Even within this creator population, earnings are heavily concentrated at the top. This data doesn’t represent the income distribution of the full 8.47 million personal-service income earners. So why do high-income cases show up so disproportionately often in social feeds? Is it really just a coincidence that these supposedly rare high-earners all happen to be gathered on Threads and Instagram?

Oswarld’s Lens

Honestly, this video didn’t feel like someone else’s story to me. I’ve been running what’s effectively a one-person consulting business since 2024. People often tell me, “You must feel so free,” but after 2 years of this, I’ve learned that freedom only lasts as long as you can cover your fixed monthly costs. Jared’s fixed cost was ¥1,000 in rent, and delivery driving was simply his means of covering it.

Let me answer the question I raised earlier. No, it’s not a coincidence. When I teach data courses, the first question I have students ask is: “How was this sample drawn?” A feed is not a random sample. It’s a self-selected sample made up only of people who have a reason to show off success. Sometimes income screenshots are even used as advertising to sell courses or e-books. When you look at posts like this, you need to distinguish whether the money came from an actual product or job, or from selling the method of making money itself. Of course, some people really do earn that much. But you can only know how many by looking at statistics, not at a feed.

That’s why I think the buzzword “solopreneur” badly needs a dividing line. Someone who sells their time and skills, and someone who owns a product, customers, and distribution, are running completely different businesses — even if both are technically a company of one. If the output of outsourced work can easily be replaced by AI, whoever takes on that work will face pressure to cut their rates. The latter type uses AI as a tool to build their own products and services. When Sam Altman talks about the “one-person unicorn,” he’s talking about the latter — not the future of the former. We should also ask whether course ads adequately explain this distinction. If you show someone doing outsourced work a product founder’s success story, you also need to explain the conditions required for customer acquisition and product development. At the end of the day, a solopreneur is just a “sole proprietor.” It’s worth not getting too swept up in the fantasy. Sometimes you just need to assess it plainly and coolly. (Call it whatever you like — that’s everyone’s own business.)

Jared instinctively understands this distinction too. He says that instead of continuing to take on outsourced work, he wants to pivot toward building his own product, or a product for his company. That’s the right direction. But here’s the difficulty: making the leap from outsourced work to your own product takes time — and to earn the rent, he still has to spend his evenings delivering food.

Closing

Let me sum up. Freelancers in the AI era aren’t going broke because they’re lazy. In some outsourced fields, demand has actually shrunk since AI arrived, and in platform work, huge numbers of people are competing over the same jobs. In Korea, 8.47 million people reported income from personal services, so the question of per-project rates and living costs for independent workers deserves to be examined on its own terms. The success stories flooding your feed don’t overturn this statistic. If anything, they show that selling the success story has itself become a product. And freedom only lasts as long as you can cover the fixed costs that come due every month. Without a steady way to earn that money, it becomes — just as the video put it — fake freedom.

I’d encourage you to try this exercise this week: split your income into exactly two lines. “Money earned by selling my time” and “money earned by assets I built (products, content, systems).” If the second line reads zero, there’s a good chance the freedom you have right now only lasts as long as the hours you put in.

How about you? If you’ve stepped outside an organization, or you’re weighing whether to go independent, tell me in the comments what “fixed cost of freedom” caught you off guard. Sharing the results of your two-line income breakdown works too. I’ll carry this conversation forward with readers’ stories in the next issue.


📨 If you know a colleague weighing whether to go independent, pass this issue along.


Your take shapes the next issue

What resonated most in this issue, or where has your experience been different?

Any registered reader can comment for free.

References & Further Reading

Primary sources

  • No Boss, No Money: The Raw Reality of China’s Gen-Z Freelancers ··· A video documenting a day in Jared’s life, filmed in Chengdu in May 2026.
  • Demirci, O., Hannane, J. & Zhu, X., “Who Is AI Replacing? The Impact of Generative AI on Online Freelancing Platforms”, Management Science, 2025. Link ··· The backbone research for today’s piece. It measures which kinds of work shrank, and by how much, after ChatGPT, using 1.4 million job postings.
  • Meiri Jingji Xinwen (via China Economic Net), “美团745万骑手收入大曝光” (“Meituan’s 7.45 Million Riders: Income Exposed”), Sept. 2024. Link ··· A report on Meituan’s data covering the real earnings of 7.45 million delivery riders, and why the rumor that “80,000 master’s degree holders deliver food” is an exaggeration.
  • Ministry of Employment and Labor & Korea Employment Information Service, “Platform Workers Reach 883,000, Up 11.1% Year-on-Year,” Aug. 2024. Link ··· The official scale of platform labor in Korea. Best read alongside the ETNews report on the surge in IT-sector platform work.
  • Park Young-sam, “The Rapidly Growing 3.3% Personal Service Income Bracket, and the Widening Gap Between Employment and Tax Statistics,” Maeil Labor News. Link ··· The source of the “8.47 million personal-service earners” figure. A column that reads the invisible freelancing hidden in the gap between two mismatched statistics.

Background

  • Xinhuanet, “2026年扩至全国!职业伤害保障试点扩围启动” (“Expanding Nationwide in 2026! Occupational Injury Insurance Pilot Rolls Out Further”), July 2025. Link ··· Traces how China’s platform-worker injury insurance moved from a 2022 pilot program to nationwide rollout in 2026.
  • Korea Institute for Health and Social Affairs, “Policy Challenges for Protecting China’s Platform Workers,” Global Social Policy Brief. Link ··· A look at China’s institutional struggles in the era of 200 million people in flexible employment.
  • Korea Business Review, “The Paradox of the ₩40 Trillion (~$28.7B) Delivery Market: Why Are Rider Incomes Moving in Reverse?”, June 2026. Link ··· A briefing on how Korea’s delivery market size and rider per-order pay have moved in opposite directions.
  • Seoul Shinmun, “YouTubers’ Average Annual Income Tops ₩70,000,000 (~$50,400) — And the Top 1% Earn How Much?”, Feb. 2026. Link ··· National Tax Service statistics on solo media creators (for income year 2024). The gap between the top 1% and the bottom half puts numbers to the “gap between the feed and the statistics.”
  • The Scoop, “The Scream of a Generation for Whom Side Hustles Became the Default: What It Means to Be a 2030s Multi-Jobber,” Dec. 2025. Link ··· An article on the generational landscape of Korea’s 680,000 multi-jobbers, based on National Data Office figures.
  • Kyunghyang Shinmun, “Substitute Drivers and Delivery Riders Also Covered by Workers’ Compensation Insurance Starting July,” June 2023. Link ··· Shows what changed once the “exclusivity requirement” was abolished.
  • Newsis, “China’s Youth Unemployment Rate Hits 15.6% in May,” June 2026. Link · Ajunews, “China, World’s No. 1 in AI Adoption, Youth Unemployment Rate Set to Near 20%,” July 2026. Link ··· Chinese youth employment indicators that form the backdrop to the oversupply of labor.
  • AI Times, “Is the ‘AI Solo Unicorn’ Becoming a Reality?”, April 2025. Link ··· The opposite extreme of the solopreneur discourse — the story of “a one-person, $1 billion company.” Reading it alongside today’s piece sharpens the contrast.

Past issues worth reading alongside this one


Illustrated portrait of Kwangseob Ahn (Oswarld)

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

Footnotes

  1. 996: A grueling work culture in China’s tech industry — 9 a.m. to 9 p.m., six days a week. Courts have ruled it illegal, but it persists as common practice.

  2. Occupational Injury Insurance (职业伤害保障): China’s version of workers’ compensation insurance for delivery, ride-hailing, and freight platform workers. It launched as a pilot in 7 provinces in July 2022, had enrolled 12.34 million people cumulatively by June 2025, and expanded nationwide on July 1, 2026.

  3. Personal Service Business Income (3.3%): Korea’s income-reporting method for individuals who work without an employment contract, under which 3.3% of their pay is withheld at source. People captured in this statistic are, in effect, “freelancers” in the broad sense.

  4. Exclusivity Requirement: A condition under which workers’ compensation insurance applied only if a worker was primarily affiliated with a single company. Riders who worked across multiple apps were excluded from coverage because of this rule — it was abolished in July 2023.