If everyone can make the film, nobody can sell it

The sequence out of China is worth studying because it has already happened there, in compressed time.

AI short films and comics took off because the video models became genuinely good. The first people to master them earned well: a small team, or one person, could ship in a week what used to take a studio a month. That advantage lasted roughly a few months. Once the workflow was understood, it could be written down — literally, in a markdown file — and then everybody had it. Output that had been scarce became ordinary.

When everyone is more productive, nobody has an advantage. The work did not disappear; the margin did. Creators report being busier than before, for less money, and the only durable winners are the platforms that own demand and distribution. In the accounts coming out of those communities, advertising buys reach up to 80% of a creator's operating cost, while AI tokens stay under 5%.

80%

of a creator’s operating cost going to ad spend, in the accounts reported from those communities

or less going to AI tokens — the thing everyone thought was the product

markdown file needed to transfer the skill that used to take years to learn

The moat was never the content

If a human is genuinely good at something, and other humans who can do it are hard to find, that scarcity is the business. It is why a good illustrator, a good editor, a good structural engineer can charge.

AI breaks that specific mechanism. If the capability is in a model, it is available to everyone with a subscription, at the same quality, on the same afternoon. There is no scarcity to rent out. That is not a moral failing of the technology; it is simply what "infinitely replicable" means for a price.

Ad spend80%

AI tokens5%

Where a creator’s money actually goes. The model was never the cost centre.

The benefit was real. Somebody else banked it

More creators made more things people enjoyed. That is a genuine good, and it should be said plainly before the complaint: audiences got more films, more comics, more of what they liked, faster.

But the economic benefit of all that extra output did not stay with the people who made it. It pooled at the layer that owns the audience — the feed, the app, the recommendation engine. The poem is a thousand years old and describes the same structure:

A skill is scarce
Few people can do it, so the price reflects the scarcity. The moat is the person.
The workflow is documented
Once it fits in a markdown file, the skill stops being scarce — it becomes a procedure.
Output floods
Everyone produces at once. Quality rises, differentiation falls, attention becomes the bottleneck.
The platform charges for attention
Whoever owns the audience sells it back to the producers. That is where the margin ends up.

遍身罗绮者，不是养蚕人 — those dressed in silk are not the people who raise the silkworms.

What to do about it

The lesson is not "avoid AI". It is that AI compresses production and leaves distribution untouched, so anything you build on production alone is a treadmill with a subscription fee.

Three things survive the compression, and none of them are content:

Proprietary data. Not scraped, not generated — gathered from a relationship nobody else has, and therefore not replicable in a markdown file.
Owned distribution. A list, a contract, a channel you control. Renting attention from a platform is what put those creators at 80%.
Systems that carry liability. The metering, the audit trail, the compliance workflow. Boring, defensible, and impossible to vibe-code once a regulator asks.

That is the work we do: data platforms, distribution plumbing, and the machinery that makes a claim checkable. Not because content is worthless, but because content is now abundant, and abundance is exactly when the plumbing starts to matter.

If our society does not find a better way to distribute the gains, the likely outcome is unpleasant and easy to state: AI helps billionaires become trillionaires while everyone else gets more exhausted than before.

Source: an account of the AI short-film and comics market in China circulating in November 2026; the 80% / 5% cost split and the timing are as reported in those creator accounts, not independently measured here. The poem 遍身罗绮者，不是养蚕人 is from Zhang Yu (張俞), Song dynasty.

Contact: [email protected]