I quit my job without a backup plan this June. Yesterday I calculated that my income for the year has already exceeded 1 million RMB.
I am just an ordinary programmer. Earning 1 million annually was a dream I had held for many years, and I never expected to achieve it this year.
Below, I will break down every single source of my income for this year and share my personal insights.
1. Salary: ~150k RMB
I resigned on June 8th. This year's salary, plus a year-end bonus, totals approximately 150k RMB. My monthly take-home pay was around 25k, and the year-end bonus was equivalent to one month's salary.
First, a disclaimer: I don't keep strict accounting records, so the numbers below are mostly estimates. They aren't perfectly precise, but the magnitude is correct.
This was my previous main job. During this period, my realization was simple:
Working a full-time job really gets in the way of making money.
It was also during this time that I started working on side projects.
2. Self-Media: ~280k RMB
The first and most important side project was self-media (content creation).
I started posting on X (Twitter) last September. Why choose X? For two reasons.
First, I wanted output to force input—pushing myself to constantly learn about AI, study AI, and use AI.
Second, AI is developing too fast, especially in programming. It can quickly replace me. As technology advances, the speed at which programmers develop software increases, technical barriers disappear, and simply building products becomes harder.
The bigger barrier is traffic, and self-media is the best way to acquire it. A large portion of my income this year came from monetizing self-media traffic.
This section breaks down into three parts:
1. X Ad Revenue Share: >$3,000 (~20k RMB).
2. Sponsored Posts: >210k RMB.
This is the bulk of self-media income. I think this benefits from the current era: models were released frantically in recent months, there were many AI products, and everyone was promoting them, leading to many sponsored posts.
However, I believe this income may decrease in the future. You should be able to feel that sponsored posts are dwindling recently. Everyone was fighting hard before, whether in AI applications or other areas, but now things are cooling down. Major companies might be wondering: Is throwing money like this right? Even domestic models haven't been releasing much this month.
3. Cloud Server Referral Commissions: >50k RMB.
I recommended cloud servers to some B2B clients I know. Whenever they used cloud servers monthly, I received a commission. So far this year, this segment has brought in over 50k RMB.
These are all based on self-media monetization. Is that it? Not quite. Next is the biggest chunk.
3. API Relay Station: 580k RMB
You might know that I run an API relay station with Wuyun @Pluvio9yte, but I haven't publicly disclosed the revenue before. Friends outside say we make over 200k a month; I can only say that while the relay station is indeed my highest-earning segment, it's not that high.
We started in March and have monetized a total of 580k RMB so far. This includes two parts: selling tokens via the relay station itself, and two rounds of agency/coaching services worth over 40k, which are included here.
Why did it work?
I think it mainly benefited from three things.
First, we started early. In March, the entire relay station industry wasn't that competitive yet.
Second, traffic. My partner and I both had a certain number of followers on Twitter, so we quickly gathered our first batch of users. At that time, the relay station industry was chaotic, with many people cheating or diluting quality. Because we had trust built through self-media, users easily gathered around us. We reached 100 daily active users in just over two weeks.
This was the first time I truly felt the dividend brought by traffic.
Third, besides traffic, our models themselves could withstand scrutiny. We prioritized authenticity and stability above all else. Prices might be slightly higher, but we guaranteed the models were real and sufficiently stable.
Things I learned:
First, gather people first, then sell things.
You can look back at our previous posts. The very first thing we did wasn't selling the relay station, but posting a discussion thread about Claude account bans. We created five WeChat groups of 200 people each, gathering a large number of people with needs. Then we ran lotteries and gave away credits in the groups, letting everyone experience our service first.
Second, create differentiation visible to users.
For example, the landing page. At the time, many relay stations barely had landing pages or just used open-source templates directly. I thought this lacked credibility, so I specifically hired a UI designer to create one. Many users feedback that our landing page looked professional. Now many relay stations are also redesigning their landing pages. (Of course, ours hasn't been updated in a long time and feels a bit outdated.)
Third, don't hard-sell, tell stories.
When we promoted, we never did aggressive advertising. Instead, we constantly told our startup story and the story of building the relay station. It was more about promoting ourselves than spamming DragonCode ads relentlessly.
Current Status of the Relay Station
August was the most profitable month for the relay station, my peak, after which it dropped significantly.
In August, a major B2B user trusted me deeply, likely a fan as well. That month, they had a big development project, and this single user contributed nearly 200k in consumption alone.
Now, our main users are also B2B. Looking at the data, 7 B2B users contribute 70% of the platform's revenue.
This might seem counterintuitive: Promoting on X should theoretically result in mostly C2C (consumer) users. I initially thought so too. Later, I gradually discovered that many people, after using it themselves, were willing to recommend it to their companies because they could get reimbursed.
Additionally, the quality of followers on X is genuinely high. Many users are company leaders or bosses themselves. They fear buying fake or diluted tokens. With a foundation of trust, they are willing to cooperate with us continuously.
Another point is service. I handle after-sales support personally. For every B2B user, we create a dedicated service group. Most of the time, I can respond immediately.
Do problems occur? Definitely, we often have issues. But I never neglect customers first. When problems arise, I admit mistakes first, explain the root cause clearly, and compensate if necessary, even if it means earning less. The most important thing is making users feel: someone is always responding.
I've talked a lot about the relay station because the gains there were significant. I'll share more details in the future if possible.
4. Other Small Incomes: ~30k RMB
1. Self-Media Community: ~12k RMB.
Wuyun, Rachel @Zesee, and I organized a self-media community priced at 99 RMB per person. The group is still active. My share is roughly 12k.
2. Lectures/Training: 18,800 RMB.
This also came from self-media. People asked me to give lectures or conduct training. The number is precise because I only took three orders.
5. Failures Too
Above are the successes, but of course, there were failures.
First: SaaS Going Global.
In January and February, I focused almost entirely on launching a SaaS product globally. But the energy investment was too high, and I slowly realized I might not be suited for it. I built a product and earned tens of dollars, but didn't break even. Later, Stripe banned my account. I didn't have the energy to monitor new keywords daily and realized this wasn't the right direction for me.
Second: FDE (Forward Deployed Engineer).
Since I worked on AI transformation at my previous company, I wanted to take on some FDE gigs.
I contacted one client who wanted a workflow for batch-generating AI videos. Negotiating the price took a lot of my time; then creating demos took a long time, with various constraints and difficult explanations. Understanding their business took time, and they haggled aggressively, wanting payment based on results rather than upfront installments.
Eventually, I abandoned this order and gave up on doing FDE work. Delivering on-site for B2B is too exhausting; it feels like you can't even recover costs.
6. Future Direction
After discussing these, I've been thinking about how to move forward.
I had a deep chat with Claude, and it reminded me: The relay station carries high risk, completely dependent on upstream providers. If Claude ever stops being accessible, or GPT imposes restrictions, or domestic models catch up in capability, this business could vanish overnight.
So it suggested I treat the relay station income as exploration funding to buy time; calculate it as if it could disappear at any moment.
Thus, I started considering future directions, still based on self-media:
1. Expand Domestic Self-Media. Create videos, post on Xiaohongshu (Little Red Book), and develop multi-channel presence.
2. Knowledge Monetization. I feel skilled at this and willing to share my experience and knowledge. The first course starts with AI Programming.
Moreover, I don't want to limit myself to just one course. In the future, I hope to build a platform aggregating more good AI courses. Of course, it's just beginning now.
My AI Programming course materials are fully complete. I've recorded part of it and am updating while recording. If you really want to learn AI programming, or have never programmed before, you're welcome to check it out.
This represents everything I've accumulated from years as a programmer plus extensive AI coding experience. Polishing this course alone took me about 3 months.
This course is something I hold high hopes for. It's currently version 2: For version 1, I asked friends on WeChat Moments if anyone wanted to learn, did a trial round, gathered feedback, and re-recorded parts. I will continue updating and release V3 in the future.
If interested, scan the QR code to join the consultation group.
7. Summary
Finally, here is the breakdown of this year's income:
Totaling approximately 1.04 million RMB.
I believe many aspects of this are replicable by ordinary people. Because I am also an ordinary person, an ordinary programmer, with a modest salary and no Big Tech background.
I believe you can achieve what I have done. Now is the best era for AI, worth seizing to do something meaningful.
