Picture the scene: November 2025, and Suno—a four-year-old startup run by Harvard physicists out of Cambridge, Massachusetts—is pitching investors on a $250 million funding round while simultaneously sitting across from Universal Music Group's lawyers in federal court. The charge? Mass copyright infringement. Training AI models on millions of songs without asking first.
Suno didn't exactly dispute it. The company's August 2024 court filing essentially shrugged: yes, we trained on copyrighted music scraped from the internet. We just think it's fair use.
Six days after the Series C closed at a $2.45 billion valuation, Suno did something unexpected. It announced a settlement and partnership with one of its plaintiffs—Warner Music Group. First deal of its kind. The first time a major label publicly bet that working with an AI music generator might be smarter than fighting one.
Whether that gambit works is the billion-dollar question.
When Physicists Build Music Apps
Mikey Shulman transitioned from a career in physics and AI to leading a music tech company. The Suno CEO holds a PhD in physics from Harvard and spent years leading machine learning at Kensho before S&P Global acquired it. His three co-founders—Georg Kucsko, Martin Camacho, and Keenan Freyberg—all emerged from similarly technical corners of academia. Camacho, the company's president, graduated from Harvard at eighteen with degrees in computer science and math.
They founded Suno in 2022 with a pitch that sounded almost too simple: type words into a box, get a fully produced song back. Not instrumental snippets or background loops. Actual songs—verses, choruses, vocals, production polish, the works.
It launched quietly. The growth didn't stay that way.
By late 2025, Suno claimed nearly 100 million users had cumulatively created music on the platform. According to investor documents that Billboard got hold of that November, the platform was churning out roughly seven million songs daily. Do the math: that's Spotify's entire catalog recreated every two weeks.
The Freemium Frenzy
The business model followed the playbook: free tier for hobbyists, paid subscriptions for serious users. Pro ran about $10 monthly; Premier hovered around $30. By February 2026, CEO Shulman announced two million paying subscribers and $300 million in annual recurring revenue—a 50% jump from the $200 million figure investors had seen just three months prior during the Series C raise.
Product releases came fast. Version 3, in March 2024, extended songs to two minutes on the free tier. Version 4 dropped that November. October 2025 brought Suno Studio, a browser-based generative workstation with multitrack editing, stem separation, loop recording—the kinds of tools you'd expect in professional DAWs, not AI toy apps.
Then came v5.5 in March 2026. New features: Voices, which let users record and clone their own singing voice. Custom Models for style fine-tuning. A personalization engine called My Taste.
To muscle into professional workflows, Suno acquired WavTool—another AI-powered browser DAW—in June 2025. That same November, as part of the Warner settlement, Suno picked up Songkick, the concert discovery platform Warner had owned. Strange bedfellows, perhaps. Or maybe strategic ones.
The Lawsuits Nobody Really Denied
The Recording Industry Association of America sued first, in June 2024, on behalf of Universal, Sony, and Warner. Federal court, Massachusetts. The allegation: Suno had trained its AI on copyrighted recordings en masse, without licenses or permission.
Suno filed its response a month later. The company's defense was bracing in its candor. It didn't claim the training data was clean. Instead, it argued the entire practice qualified as transformative fair use.
"We need to train on the sounds of instruments, the sound of the human voice," Shulman told Axios that August, shortly after the filing landed. "We couldn't do what we're doing without training on existing music."
That argument didn't make the lawsuits disappear. According to court dockets, two more arrived in 2025: Justice et al. v. Suno in June, filed by independent artists, and Woulard et al. v. Suno in October, filed in Illinois state court with claims spanning copyright, DMCA violations, and Illinois biometric privacy law. By early 2026, depositions were underway in the Massachusetts case. The docket showed discovery protocols established on March 25.
This wasn't going away.
Warner Blinks First
Then came the November 2025 Warner deal, which scrambled the calculus.
No public dollar figure attached to the settlement. But the terms told a story. Warner and Suno agreed to build licensed AI models together—artists could opt in, allowing their names, images, voices, and compositions to be used in AI experiences. Crucially, Suno agreed to impose new download limits: monthly caps for paid subscribers using the text-to-music generator, and free-tier downloads eliminated entirely. (Suno Studio remained untouched.)
In his blog post announcing the partnership, Shulman struck a tone somewhere between conciliatory and triumphant. The deal, he wrote, was about creating "a future where everyone can share in the joy of creating new music" while ensuring artists "are empowered and rewarded."
Not everyone was buying it.
Late February 2026, a coalition of artist representatives launched "Say No to Suno," an open letter campaign accusing the company of running a "smash and grab" operation. The letter argued that AI-generated music—what critics derisively called "AI slop"—was diluting royalty pools for real artists "from whose music this slop is derived." Music Business Worldwide, Forbes, the AP—they all covered it.
The backlash had its own momentum now.
Hiring for Détente
Meanwhile, Suno was staffing up as if preparing for coexistence, not war.
July 2025: Paul Sinclair, a major-label veteran from Atlantic and Warner, joined as Chief Music Officer. February 2026: Jeremy Sirota, former CEO of Merlin (the licensing collective for independent labels), came aboard as Chief Commercial Officer. These weren't the hires of a company planning to burn bridges. These were the hires of a company planning to build them.
Suno was expanding geographically, too. Axios reported in April 2026 that the company had opened a new office in San Francisco's SoMa district. The company's website listed operations spanning "three offices," though it declined to specify cities beyond Cambridge.
There were success stories to wave around. Telisha "Nikki" Jones, performing as Xania Monet, used Suno to create a viral R&B track and reportedly signed a multi-million-dollar deal with Hallwood Media in September 2025. Another creator, imoliver, signed with Hallwood that July. Hallwood Media, incidentally, was also an investor in Suno's Series C.
Convenient? Sure. But also indicative of the ecosystem Suno was trying to build.
The Unresolved Question
Here's what hasn't changed: the lawsuits with Universal and Sony are still active. The Warner settlement peeled off one plaintiff, but the core legal question—whether training generative AI on copyrighted works without permission constitutes fair use—remains unresolved. And unlikely to be resolved anytime soon.
Court rulings take time. Legislative action takes longer. The artist backlash shows no sign of fading.
What Suno does have is scale, momentum, and a blueprint. The Warner deal sketches out a future where AI music platforms pay for training data and artists opt into AI applications of their work—rather than discovering, after the fact, that their voices and styles have been synthesized without consent.
Whether that future actually arrives depends on factors Suno can't control: judges, lawmakers, and whether Universal and Sony decide Warner had the right idea.
In the meantime, seven million songs a day. Two million paying subscribers. Product releases every few months. Acquisitions. Executive hires from the belly of the beast.
Suno is building as if the permission will eventually come. As if the lawsuits are noise, not signals. As if a $2.45 billion valuation is validation enough.
It's a bet—an enormous one—that the music industry's center of gravity has already shifted. Even if the industry itself hasn't quite realized it yet.
