AI-Generated Images on Adobe Stock Earn ¥500K a Year: 12 Months of Real Data from a Salaryman in His 40s Working Weekends Only, Averaging ¥42K/Month
Rossin, a company employee in his 40s, sells Midjourney-generated images on Adobe Stock and disclosed his 2025 annual revenue of ¥501,834 (¥41,819/month average) with actual month-by-month figures. His best month was December at ¥56K, showing how seasonal assets drive sales. Three years of activity on weekends only.
This is a translation of the Japanese original. The Japanese version and cited primary sources are authoritative for all figures.
The Published Numbers (2025, Monthly)
| Item | Figure |
|---|---|
| Annual total | ¥501,834 |
| Monthly average | ¥41,819 (lowest: February ¥28K — highest: December ¥56K) |
| Production method | AI image generation with Midjourney and similar tools |
| Setup | Company employee in his 40s; weekends-only side project (started 2022) |
What the Business Is
Rossin creates assets with AI image generation tools and lists them for sale on Adobe Stock. Demand for stock assets tracks the calendar of practical design work, so materials for the New Year holidays and seasonal events create the revenue peaks. From the 12 months of actual figures, you can see that stock-based income moves with “accumulated volume of uploads × seasonal demand.”
Our Editorial Take
AI image generation has shifted the barrier to entry in stock photography from “photography skill” to “demand-picking skill.” Now that camera gear and photo trips are unnecessary, the only thing separating earners from non-earners is the eye for which demand (season, use case) to supply assets to. Placed next to the contrast data of a traditional stock photographer earning a cumulative ¥185K over 4 years, the change in production cost structure translates directly into the revenue gap.
¥40K a month is a realistic destination for “weekends-only compounding.” Not a one-off hit, but a structure where three years of accumulated uploads keep selling month after month. The biggest risk — platform policy changes on AI-generated content — is the same structural risk as Kindle publishing.
Reading the Monthly Data: Cracking Seasonal Demand
The published monthly figures range from ¥28K in February to ¥56K in December — exactly a 2x spread. Asset demand tracks the busy periods of design work (New Year’s cards, year-end retail, the new school/fiscal year), and that makes seasonality a variable you can play. Upload seasonal assets 2-3 months ahead of demand — assets that sell in December should be created in September-October. This time-shifted operation changes revenue even at the same production volume.
The practical side of AI generation is also maturing. Prompt iteration keeps driving down the per-image generation cost, and patterns for handling rejections (Adobe Stock’s quality review) accumulate. The value of three years of continuity lies not just in the inventory of assets but in the accumulated learning of “what passes review and what sells.”
Conditions for Reproducing This Case, and Its Limits
The conditions for reproduction are clear: (1) being able to keep paying the subscription fees for Midjourney and similar tools (a few thousand yen per month), (2) sustaining a few hours every weekend for three years, and (3) generating images by working backward from what sells (use case, season). With zero costs for camera gear, models, or locations, the initial investment is a tiny fraction of traditional stock photography.
On the other hand, the biggest risk cannot be eliminated by personal effort. Platform policies on AI assets (whether they can be listed, disclosure requirements, commission rates) can change, and price erosion from an oversupply of AI assets is ongoing. Treat “today’s ¥40K a month” as a number with no guarantee of permanence — spreading risk across multiple platforms or combining with traditional stock photography is the realistic way to operate.
Related Cases
Sources
This article summarizes and analyzes the public sources above. Please refer to the primary sources for details.