Mikey Shulman on Building Suno into a $2.5B AI Music Platform

Suno bets AI music becomes an everyday creative act

Suno is betting that generating music becomes an everyday creative act rather than a novelty. Audio had been neglected in machine learning, and a research breakthrough let the company build a model good enough that 100m people now make songs, a quarter of them remixes of each other's work. The next fight is the business model, since recorded music's economics have always followed its format.

Takeaways

• Suno's research bet unlocked AI music. Audio was a backwater in machine learning even though audio is everywhere. Suno assumed it needed 100 times more compute, then concluded a couple of months later it could be done. That breakthrough seeded the company in 2022.
• Suno reached 100m users in three years. Founded in 2022 and released publicly in 2023, Suno now generates 7m tracks a day. Revenue went from zero to more than $200m, and the company raised a $250m Series C at a valuation near $2.5bn.
• Remixing pulls users into making music. A quarter of the songs made on Suno are remixes of other songs on the platform. Schulman calls remixing a gateway drug into making your own music, and says music has become non-social for many people.
• Format has always shaped music's economics. Recorded music's business model has followed its format, from vinyl and CDs to streaming. Suno charges $10 or $30 a month for subscriptions alongside a free tier, and Schulman predicts interactive music will force a new model that grows the pie.
• AI acts are reaching the charts. Xania Monet, the act built around the poet Nikki, became the first AI-supported song act to debut on a Billboard radio chart, with more than 100m global streams since July. Schulman compares it to laptop production and EDM.
• Music is a lower-stakes AI domain. Suno runs copyright detection and mild content moderation, and Schulman says music cannot be weaponized, so the company can build with more speed. He doubts agentic stars matter much because music needs a person behind it.
• Pace comes from people enjoying work. Schulman criticises AI products that start as thin wrappers around a model. He says a company lives or dies by how many experiments it ships to users, and flipped Suno's culture so users drive the product.

From quantum computing to Suno

1. Schulman grew up in New York and studied physics at Columbia. A professor's last-minute decision to stay sent him into a quantum computing lab instead of a Goldman Sachs internship.
2. He spent six years on an experimental quantum computing PhD at Harvard, 2009 to 2015.
3. He joined Kensho as a software engineer when it had about 12 people, led machine learning there, and admits he interviewed badly. S&P Global acquired Kensho in 2018.