Suno.

Somewhere right now, a 14-year-old who cannot read sheet music is finishing a three-minute pop song. She typed two lines. Suno did the rest. Multiply that by seven million a day.

Suno is the loudest argument in tech that you do not need to play an instrument to make music. The company, based in Cambridge, Massachusetts, takes a few words of text - a mood, a genre, a lyric - and returns a finished track with vocals, instruments, and structure in roughly the time it takes to read this sentence. As of mid-2026 it has tens of millions of users, more than two million of whom pay, and a valuation of $5.4 billion. It is, depending on who you ask, the most exciting consumer product in artificial intelligence or the most expensive lawsuit the music industry has ever filed. It is comfortably both.

The product looks deceptively simple, which is the point. A text box. A button. A song. Behind it sits a stack of generative models trained to do something machines were not supposed to be good at: carry a tune, hold a rhythm, and sound like they meant it. The skeptic's reflex - "but is it any good?" - has slowly become "wait, that's not real?" That migration, from novelty to uncanny, is the whole business.

01 / The problem they sawThe 99% who never get to play

Here is the uncomfortable math of music. Almost everyone loves it. Almost no one makes it. The gap between humming a melody in the shower and producing something you can actually share is enormous - years of lessons, expensive software, a vocabulary of keys and time signatures, and the patience of a saint. The tools were built by experts, for experts. Everyone else was handed a "Listen" button and told that was their role.

Suno's founders looked at that gap and saw it the way a previous generation saw the gap between professional photographers and the rest of us before phones put a camera in every pocket. The bottleneck was never desire. It was the equipment, and the decade of practice the equipment demanded. Remove the bottleneck and you do not get fewer musicians. You get a billion of them - most of them terrible, a few of them revelatory, all of them participating for the first time.

02 / The founders' betFour physicists who could also hold a tune

Suno was founded by four people - Mikey Shulman, Georg Kucsko, Martin Camacho, and Keenan Freyberg - who met building artificial intelligence for finance at Kensho, the machine-learning company now part of S&P Global. Shulman, the CEO, holds a PhD in physics from Harvard and lectures at the MIT Sloan School on natural language processing. On paper, an unlikely crew to upend the recording industry.

The detail that matters: all four were musicians. They were the people at the lab who actually played. So when they left to start an audio company, they were not chasing a market they had read about in a deck. They were building the tool they themselves wanted and could never quite buy - one that closed the distance between the song in your head and the song in your ears.

Investors, eventually, agreed. The early checks were modest. The later ones were not. By the time the dust settled in 2026, a company founded only a few years earlier was raising hundreds of millions at a multibillion-dollar price - a trajectory that would look reckless if the usage numbers were not so stubbornly real.

03 / The productFrom a text box to a recording studio

Suno's early models were a parlor trick that got progressively harder to dismiss. Between late 2023 and 2025 the company shipped seven distinct versions - v2 through v5 - each one stretching what the thing could do: longer songs, cleaner audio, vocals that stopped sounding like a robot doing karaoke. v3 opened the floodgates to free users in 2024. v4 made the vocals convincing. v5, in late 2025, did something stranger.

It turned the toy into a tool. Alongside v5, Suno launched Suno Studio - a full digital audio workstation that runs in the browser. Suddenly you could pull apart a generated track into its component stems (vocals, bass, drums, harmony, instrumentation), edit individual sections, adjust EQ, warp timing, strip effects, and export up to twelve time-aligned WAV files into Ableton or Logic like any other session. A February 2026 update piled on warp markers, effect removal, alternates, and time-signature control.

7M+

Tracks / day

2M+

Paid subscribers

Top 10

iOS & Android music app

7

Models shipped (v2-v5)

What can you actually do with it? Write a birthday song for someone who will pretend to hate it. Sketch a demo before you book studio time. Score a podcast intro, a wedding montage, a video game level. Turn a bad day into a worse country song. Professionals use it to prototype ideas at the speed of thought; amateurs use it to make the thing they were never allowed to make. The free tier lets you try; the paid Pro and Premier tiers add credits, commercial rights, and Studio.

04 / The proofThe money believes it, even while the lawyers argue

Skeptics are right to ask whether any of this is a real business or just a very expensive demo. The clearest answer is revenue. Suno reportedly reached around $200 million in annual recurring revenue by late 2025 and roughly $300 million by 2026 - growth that came from people quietly paying every month, not from a one-time gold rush.

Then there is the litigation, which is the part the headlines love. In 2024, labels owned by Sony, Universal, and Warner sued Suno (and rival Udio) over the music used to train its models, alleging copyright infringement on a massive scale. Suno is fighting on fair-use grounds in federal court in Massachusetts. It is the kind of fight that usually ends a startup. Instead, the valuation went up.

The twist arrived in 2026, when Warner Music Group became the first major label to settle and announce a partnership on "next-generation licensed AI music." Universal had already struck a deal with Udio. Suno, for its part, says it will roll out new licensed models and deprecate the old ones as it does. Not everyone is thrilled - the American Federation of Musicians has sued the labels, arguing the settlement money is not reaching the actual musicians. The fight, in other words, is just changing shape.

05 / The missionA future where everyone makes, not just listens

Strip away the funding rounds and the courtrooms and Suno's stated mission is almost old-fashioned: make it so anyone can make music, regardless of training. The company frames the last century of recorded music as an accident of distribution - a small number of professionals making songs for a vast number of passive listeners, because that was the only economically viable shape. Suno's wager is that the shape was never natural. It was just expensive.

Whether you find that inspiring or alarming probably depends on whether you make your living from the old shape. Working musicians have real reasons for concern, and the licensing fights are evidence that the industry intends to be paid. But the genie is the kind that does not return to bottles. Tens of millions of people have now made a song. Most of them never will again. Some of them will never stop.

06 / Why it matters tomorrowThe instrument that plays back the room

The interesting question is not whether AI-generated music is "as good as" human music. That framing already sounds dated, the way "is digital photography as good as film" sounds dated. The question is what happens to a culture when the cost of making a song falls to roughly zero and the skill required falls with it. More noise, certainly. Also more voices that were never going to reach a studio.

Back to that 14-year-old finishing her pop song. A few years ago she would have had a daydream and nothing to play it on. Now she has a track, a chorus she keeps humming, and the strange new feeling of having made something instead of merely having liked something. She may never become a professional. That was never the point. The point is that the door, which used to be locked, is now just a text box. Suno built the box, picked an expensive fight over what goes inside it, and bet several billion dollars that once people walk through that door, they do not walk back.