AI that turns a text description into a finished song - melody, arrangement, vocals, all at once. That's why the music industry is suing it and signing deals with it at the same time.
In practice you go into Suno, write a prompt, genre, mood, themes for the lyrics, even an artist's name as a reference, and the platform generates a complete song with a sung melody, an instrumental, and song structure: verse, chorus, bridge. It doesn't assemble anything from ready-made pieces, the way older beat-building programs do. The AI composes from scratch, differently every time, as if you'd commissioned a custom song from an invisible studio crew that works for free and never rests.
It exists because writing a song requires a musician, a lyricist, a studio and a producer, a resource most people don't have. Suno lowers that barrier to a single text field. People with no musical education use it to make a birthday gift with someone's name in the lyrics, YouTubers use it for royalty-free background music, companies use it for ads and jingles, even musicians use it to sketch an idea before going into a real studio.
The problem is where that ability comes from. The company trained its model on millions of existing recordings, and, according to the lawsuits, without permission from the rights holders. Universal Music and Sony are still suing Suno over this, while Warner Music made a U-turn: in 2026 it settled its case and, instead of fighting, struck a partnership, it will license its catalog for new, legal Suno models. The company is valued at $5.4 billion, and the Warner deal gave something back to artists: free accounts can no longer download the songs they make, and paid accounts got a monthly limit.
Here's what it stands on
I've tried it. Technically it sounds surprisingly decent, production, arrangement, even emotion in the voice. But something's missing, and I can't explain it with a word like 'soul', because that sounds pretentious. It's more this: the song didn't come from something lived through, but from a statistical prediction of what sounds lived through. It's AI-generated, and it shows. For background on a video, that's enough. For a song that keeps you awake at 3 in the morning, it isn't, not yet.