36,500 subscribers, but two years of total revenue in the high hundreds-of-thousands of yen. An AI-explainer YouTube channel's hourly wage: ¥416

On May 25, 2026, the YouTube channel “AI de Sōsaku Kōritsuka to Manetaizu by Aaa Frog”, which posts explainer videos about AI tools, marked its second anniversary. That same month, the channel’s operator, Aaa Frog, published a note article summarizing two years’ worth of numbers. 36,500 subscribers. Two years of total revenue in the high hundreds of thousands of yen. Recently, finally, a few tens of thousands of yen a month. Aaa Frog calls this “my only income source (strictly speaking there are others, but they’re zero to a few hundred yen)” and describes himself as “an unemployed man.”

Looking at the subscriber count of 30,000-plus alone, this case would be filed under “success.” But as Aaa Frog anticipates right at the top of the article, there’s a considerable gap between the reader’s assumption (“with 30,000 subscribers you must be making way more than that, stop lying!”) and the actual figures. The value of this case lies in that gap being disclosed, complete with dates and amounts.

What happened over two years

112 standard videos have been produced, of which 80 are currently public. Divided across roughly 25 months, that’s a production pace of 4.5 videos/month, or 3.2/month on a published basis. Aaa Frog writes: “If one video takes 30 hours to make, that comes out to about ¥416 an hour.”

The numbers jumped right after monetization

Breaking the subscriber growth rate down into intervals reveals a clear point where the trajectory changed.

• April 2024 → July 2024: 1,000 in 3 months, about 333/month
• July 2024 → Oct 13, 2024: 1,000 in about 3.4 months, about 294/month
• Oct 13, 2024 → Feb 11, 2025: 8,000 in about 4 months, about 2,000/month
• Feb 11, 2025 → Feb 26, 2026: 20,000 in about 12.5 months, about 1,600/month
• Feb 26, 2026 → May 25, 2026: 6,500 in about 2.9 months, about 2,240/month

For the first six months after launch, growth ran at about 300/month — a pace that, if sustained, would take 8 years to reach 30,000. Instead, from mid-October 2024 through the following February, that pace jumped roughly 6-7x. After that it stabilized at 1,600-2,200/month, and the most recent three months have actually accelerated further.

Right before this acceleration sit two events: hitting 4,000 watch hours in late September 2024, and monetization on October 4. However, the free portion of the source doesn’t explain why growth picked up at this point. Was it monetization itself? Did the accumulated backlog of videos simply start surfacing in search and recommendations? Did demand for AI-tool explainers itself swell in late 2024? Aaa Frog doesn’t say, so this can’t be stated definitively. What can be said with certainty is only this: extrapolating from the six-month mark (2,000 subscribers) to predict the two-year outcome would have been wrong.

What Aaa Frog calls the turning point isn’t the moment of growth

The more important turning point lies not in the numbers but in a judgment call. Before going fully unemployed, Aaa Frog had set goals in advance as a benchmark for knowing when to pull back. Year one: exceed ¥X0,000/month in ad revenue alone. Year two: keep ¥X0,000+/month while exceeding ¥X0,000 in a single best month (specific amounts are in the paid section). The overarching goal was “at minimum, reach a livable line within 3 years.”

And Aaa Frog explicitly states that this goal has been met. And yet, in the year-two wrap-up, he writes “I’m feeling the ceiling,” and offers the assessment: “Ad revenue has probably hit close to its limit here.”

In other words, the turning point in this case sits at the moment where, right in the middle of things going according to plan, the plan’s own premise got questioned. Where many revenue-disclosure posts read “we missed our target, so we changed course,” this one reads instead: “we hit the target, but there’s no livable income down the road of ad revenue alone.” The evidence behind that judgment is two years of measured reality: subscribers grew 18x, from 2,000 to 36,500, while revenue never reached a livable level.

Checking the ¥416 hourly wage

The ¥416 hourly figure doesn’t quite compute if you take it at face value.

Using 112 videos × 30 hours = 3,360 hours as the denominator, the numerator would need to be about ¥1.4 million for ¥416/hour to hold. Using the 80 published videos × 30 hours = 2,400 hours instead, it comes to about ¥1 million. Both are larger than the “high hundreds of thousands of yen” Aaa Frog states. What exactly goes into the numerator, ad revenue only? membership revenue included? pre-tax?, isn’t specified in the article.

Conversely, if total revenue is assumed to be ¥800,000 and divided the other way, that’s about ¥238/hour at 3,360 hours, or about ¥333/hour at 2,400 hours. Either way, this falls far short of the minimum wage anywhere in Japan, meaning the ¥416 figure Aaa Frog presents may actually be, if anything, on the generous side of the real number.

Averaging the same assumption monthly: ¥800,000 over two years is about ¥33,000/month on average. Since monetization only started in month 7, dividing across the 19 monetized months gives about ¥42,000/month on average. Against 36,500 subscribers, that’s under ¥400,000 a year, roughly ¥11 per subscriber per year. This density is a good measure of how far the intuition “subscriber count = income” can be from reality.

Note that the ¥75 earned on day one of monetization reflects only about 5 hours of ads running (turned on around 7pm that day), so comparing it directly to the recent “a few tens of thousands of yen a month” isn’t quite fair. Still, if subscribers grew 18x while daily revenue moved by roughly the same order of magnitude, the per-subscriber unit value has barely improved over two years.

The 32 videos that never surfaced

Easy to overlook: against 112 videos produced, only 80 are currently public. Of the total, 32 videos, about 29%, have not been released to the world. At the estimated 30 hours per video, that’s 960 hours of production that never saw daylight. The article doesn’t state why they’re unpublished, deleted, restricted, or made obsolete by topic drift are all possible. But the genre of AI-tool explainers is inherently short-lived: tool UIs and specs change within a few months, so content has a short shelf life by nature.

The risk side is stated plainly too. Income is almost entirely concentrated in this one channel’s ad revenue and memberships, with other income “zero to a few hundred yen.” A single platform, a single revenue source, with no buffer to absorb a policy change or algorithm shift. Being unemployed while depending on a single channel is efficient while things are growing, but leaves no fallback the moment growth stops.

How far can this be replicated

What’s replicable, ahead of how time was spent, is the design of clearing conditions. Both 1,000 subscribers and 4,000 watch hours were cleared roughly six months after launch. That speed came from a combination of a 4.5-video/month production pace and choosing a topic, explaining tools whose demand was just emerging. When the topic is “how to use a new tool,” viewers arrive already searching with a purpose, sparing you the work of building initial visibility from zero.

Another replicable element is the stance of deciding numeric targets and an exit line in advance. Aaa Frog achieved the targets, then judged the limit by looking at what achieving them actually meant. Without pre-set targets, that judgment wouldn’t have been possible.

On the other hand, the non-replicable conditions are also clear. Entering during the first half of 2024, when the generative-AI explainer market was just taking off, is a timing that can’t be chosen after the fact. The 30 hours devotable to a single video is the flip side of being unemployed. And as noted in the channel description, Aaa Frog is a multi-disciplinary creator handling novels, screenwriting, illustration, video, 3D, and music, a breadth of skill that lets production be completed without outsourcing. Set the target of “36,500 subscribers in 2 years” without these three conditions in place, and your time estimates will collapse.

And what’s most worth taking away is probably the premise this case disproves. The idea that 10,000 subscribers is enough to make a living didn’t hold up, at least not in the AI-explainer genre, at this scale. What Aaa Frog confirmed over two years isn’t how to grow. It’s what isn’t there once you’ve grown.

Related reading

• Hitode’s blog operations — a case that discloses individual media revenue trends over the long term
• 24 Kindle books’ worth of results — a case breaking down the relationship between production volume and revenue on a per-unit basis