Three Ways People Are Making Money With AI Music — and How to Test Each One

Public AI-music cases now include a catalog operator reporting more than $152,000 in cumulative revenue, a YouTube portfolio reporting approximately $7,700 in 26 monetized days, and an artist operation reporting a $5,300 best month from streams and downloads.

They reached those numbers through three different businesses.

One built artists that people followed across social media and streaming services. One programmed long YouTube sessions for a specific listening situation. One released music under several artist identities and continued the few that found demand.

I found these patterns after screening 130 people and channels that publicly discussed AI-music income. My team and I have also released more than 80 tracks in a catalog that crossed 20 million streams and earned tens of thousands of dollars. I wrote about that catalog separately.

AI music can earn money. The decision is which business you are prepared to run after the song is generated.

The three models at a glance

1. Build an artist people want to follow

This model works when the listener remembers more than one song.

The artist needs a recognizable voice, visual identity, emotional theme, character, or story. Short videos introduce that promise. Full songs and the artist page give the listener somewhere to continue.

The path usually looks like this:

short clip or story
    → artist profile
    → full song
    → save, download, or repeat play
    → another release

A Reddit operator using the name Fresh-Ad-3574 reported a $5,300 best month, about 2.5 million cumulative Spotify and Apple Music streams, and 2,500 iTunes sales.

The operator started new social accounts and posted every day. TikTok created the first response. Facebook later produced a larger jump. Interested viewers could move from a clip to a complete song, streaming page, or paid download.

Xania Monet used a different version of the same model. Telisha Jones turned her own poetry and personal stories into a consistent R&B and gospel artist. Social attention led into streaming charts, radio, television, and later label support.

Test the artist model

Start with one sentence:

This artist makes emotionally direct gospel songs for women rebuilding their lives after a difficult relationship.

The sentence should guide the lyrics, voice, images, short videos, and next releases.

For the first test:

1. Make 10–20 song candidates around the same artist promise.
2. Finish the strongest three.
3. Give them one consistent artist page and visual style.
4. Create 10–15 short clips from different hooks, lyrics, and stories.
5. Send interested viewers to the same full song or artist page.
6. Track saves, downloads, repeat listeners, and plays of another song.

Views show that a clip worked. Movement into the music shows that the artist may work.

2. Build a YouTube channel for a listening situation

Here the main product is a session rather than a three-minute single.

The listener wants music for work, sleep, prayer, study, a café, a romantic evening, or background television viewing. A long video can generate watch time, repeat visits, and advertising revenue.

The path is:

specific listening need
    → long video
    → watch time
    → YouTube recommendations
    → returning viewers and advertising revenue

Música Artificial reported approximately $7,700 during the first 26 monetized days of a six-channel portfolio.

The operator published four 45–60-minute videos a week. Each video contained roughly 13–15 songs and served a narrow audience. The packaging was easy to read on television screens, and the music was designed for long background listening.

Maurice reported $7,400 in 90 days from a jazz-focused channel. One video reached about 660,000 views and generated roughly $2,080. Maurice added a livestream after the video entered recommendations.

Both cases treated YouTube as a programming business. The title, thumbnail, duration, mood, and upload schedule all served the same listener.

Test the YouTube model

Write a sentence that defines the channel:

This channel gives older gospel listeners one hour of calm worship music for the evening.

Then:

1. Make four long videos for that exact use.
2. Keep the music, image, title style, length, and mood consistent.
3. Publish on a predictable schedule.
4. Track click-through rate, average viewing time, returning viewers, and television viewing.
5. Change one element at a time when a video fails.
6. Extend the topic or format that creates repeat viewing.

Do not mix sleep music, romance, prayer, motivation, and study on one new channel. The viewer and YouTube should receive one clear promise.

3. Run several artists as a catalog business

This model treats each artist or narrow music concept as a separate test.

The operator makes several song candidates, rejects the weak ones, releases the strongest, and watches what listeners do. Artists that find demand receive follow-up releases. Weak concepts stop receiving time.

The path is:

one narrow artist idea
    → several song candidates
    → human selection
    → release
    → streams, saves, and royalties
    → more releases for the winners

AI Guerrilla / Jesse Hull has reported more than $152,000 in cumulative catalog revenue and a current-month update in the $16,000–$17,000 range.

The operation releases music under several artist identities through DistroKid, streaming services, and YouTube Music. It uses a high release cadence, but the artists remain separate and the operator follows the performance of each catalog.

The advantage comes from repeated selection and follow-up. Access to the generator is available to everyone.

Test the catalog model

Begin smaller than the public success stories:

1. Choose two narrow artist ideas for different listeners.
2. Make 10 candidates for each artist.
3. Release only the strongest two or three from each group.
4. Keep separate artist pages, images, credits, and release records.
5. Compare saves, repeat listening, royalty reports, and movement into older songs.
6. Continue the stronger artist and pause the weaker one.

This model requires more organization than the other two. A larger catalog creates more opportunities for wrong artist pages, missing credits, rights questions, delayed payments, and weak projects that continue because nobody decided to stop them.

How promotion changes between the models

An artist needs people to remember and follow a character. Short video is useful because it can test hooks, lyrics, stories, and personality quickly.

A YouTube listening channel needs viewers to stay during an activity. The video itself, its title, thumbnail, duration, and mood do much of the promotional work.

A catalog operator needs each release to create another entrance into a coherent artist catalog. New songs can send people to an artist page, album, playlist, or older release.

The same promotion plan will not serve all three models.

Software used in the public cases

Suno appeared most often for music generation. Udio and private tools also appeared.

The rest of the work used familiar software:

• ChatGPT, Claude, and Gemini for ideas, lyrics, titles, and planning;
• FL Studio, Logic, Audacity, and mastering services for audio finishing;
• Midjourney, OpenArt, Canva, and Figma for artist images and thumbnails;
• CapCut, Premiere Pro, and After Effects for video;
• vidIQ, YouTube tools, spreadsheets, and Notion for channel and release planning.

The same software appeared behind very different results. The operators still had to reject weak songs, rewrite hooks, keep the artist consistent, finish the audio, check rights and credits, and decide what deserved another release.

Choose the distributor after choosing the model

DistroKid appeared most often in the public cases. Other operators used UnitedMasters, LANDR, Hallwood, SoundOn, TuneCore, or private label arrangements.

Compare distributors by the needs of the catalog:

• accepted content and release frequency;
• annual fees or revenue share;
• review and appeal process;
• payment timing and possible holds;
• YouTube Music and Official Artist Channel support;
• Content ID rules;
• collaborator and publishing splits;
• what happens when a release is challenged or removed.

Fresh-Ad moved from LANDR to DistroKid after encountering AI-release limits. Xania Monet later worked through a label-services relationship. A single independent artist and a fast multi-artist catalog should not assume they need the same deal.

Which model should you choose?

Choose the artist model if your strength is storytelling, identity, performance, or community.

Choose the YouTube model if you understand a repeat listening situation and can publish long programs consistently.

Choose the catalog model if you can manage several separate artists, reject weak work, maintain clean release records, and follow royalties by artist.

Do not begin with 300 songs. Begin with one listener, one source of income, and one route to reach that listener.

A released catalog can produce recurring income. Reaching that point still requires selection, packaging, promotion, rights records, release checks, and payment tracking.

Pick one model and run the smallest test that can show repeat listening and traceable money.

Need to choose an AI-music model?

I run a 90-minute AI Music Strategy session for founders, artists, labels, and catalog operators making this decision.

We choose the listener, income model, release route, rights process, and smallest useful pilot. You leave with a recommended model and a concrete launch plan.

Sessions start at $800. Message me on Telegram →