I sat down and did the math once.
Servers, domain, CDN, tokens — all told, several thousand yuan went in. What came back was zero revenue. Note: not little. Zero.
For a while I couldn’t make sense of it, and kept suspecting my product just wasn’t good enough. Tweaking the interface at night, tweaking the interactions during the day, then looking again — still nobody using it. Eventually I spread the profit-and-loss sheet out and followed the money line by line: the server money went to the cloud vendor, the domain money went to the registrar, the CDN money went to the network provider, the token money went to the model vendor. They all made a fortune, and the only one losing money on this chain is me.
But think about it again — no users means the product isn’t good, doesn’t it? That’s true, and not entirely true. Good wine fears a deep alley. Even a good product has to be known. I have no money for ads, no channel, and even an article I write gets read by no one but me. Between “I can build it” and “I can sell it” there’s more than a line of code.
After AI showed up, apps sprung up like bamboo shoots after rain. There’s a phrasing I find far more accurate than “bamboo shoots”: more like sandblasting — a whole pile sprayed out at once. That’s how the supply and demand relationship changed. More people are building, and there aren’t enough users to go around.
Later someone summarized this for me in a harsher way, and I think it’s clearer than what I came up with myself: the collapse of the production barrier on the supply side doesn’t bring “opportunity for everyone” — it brings several things at once. The number of people building tools explodes, and a single idea can spawn dozens of versions in a day; user attention becomes the scarcest thing there is, scarcer than compute; general capabilities get commoditized fast, and tokens, models and cloud resources will all look more and more like utilities — available on tap, and nobody pays a premium for utilities. What’s actually valuable is no longer “I can build an AI app”, but the scenario, the distribution, the trust, the stickiness of being embedded in a workflow so deep it can’t be pulled out, and pricing power. I have none of those.
So this industry feels like it depreciates the more you work. Every cent I put in is rent paid to someone else; the thing I build is worth less and less.
A few years ago, being able to write a page was enough to get work. Now everyone can call an API, models stack version on top of version, and whatever I stayed up learning becomes a default setting half a year later. It’s not that I’m not trying — the craft itself is sliding downhill.
The sheet is still spread out on the table. The server bill is due again next month.
评论
0 条评论