Some music executives start their day hoping to sign the next great pop star, land a No. 1 on the Billboard Hot 100 or sell out a stadium tour. But Paul Sinclair has a different aspiration: “Every day I wake up and I’m like, ‘Don’t ruin music,’” he says with a laugh.

Sinclair has spent decades working in the industry, primarily for Warner Music Group and Atlantic Records, at the intersection of recorded music and tech. After exiting his post as Atlantic GM/executive vp in 2025, he found himself consulting for a new artificial intelligence music startup, Suno, which had quickly become the most controversial company in music since Spotify’s launch 15 years ago. It’s arguably even more controversial than that.

Many have compared Suno, and the rise of generative AI music in general, to Napster’s launch at the turn of the century, which upended album sales and led to the darkest financial decade in recorded-music history. Suno could disrupt music consumption, like Napster and Spotify once did. But it’s also doing something different: disrupting the sacred act of creation itself.

One of Suno CEO/co-founder Mikey Shulman’s interviews for this story has been released as an episode of On the Record, Billboard‘s music business podcast. It is available now on YouTube, Spotify and Apple Podcasts.

Sinclair became Suno’s chief music officer last July and sees his skeptical eye as a selling point. “Truly, every single day I’m conflicted,” he says. “This s–t is complicated … I want to make sure there’s whole future generations of the beauty of art and music and the ability to build careers around it.” He mentions his daughter, who aspires to enter the music business in the not-so-distant future: “I really want her to be able to do that.”

Research from French streaming service Deezer suggests that 97% of people cannot differentiate between AI and human-generated music, and of the now-60,000 fully AI-generated tracks delivered to the service daily, a Deezer representative says the “vast majority” are flagged as being created with Suno specifically. According to an unpublished company pitch deck Billboard obtained in November, Suno generates 7 million songs a day, which equates to an entire Spotify catalog’s worth of music every two weeks — and the songs are sounding uncannily similar to those made the old-fashioned way. With that deck, Suno recently raised $250 million in Series C funding, a figure nearly unheard of in music tech that further cemented it as the clear front-runner in AI music.

Naturally, many musicians see Suno and other generative AI music companies as an existential threat — and for some of them, it could be. Some uses for the technology, beyond silly novelty songs or co-writing, could include creating vocals, producing demos and replacing stock or production music, all tasks that have previously required paying working-class musicians. In 2024, for a Billboard story about how AI could threaten the production music business, I spoke to Henry Phipps, an aspiring film composer and production music-maker who had also briefly worked in AI.

“Very few people aspire to be production library composers long term,” he explained, but “it is a way into [the music business] to survive, eat and pay rent and work toward projects that are more creatively fulfilling.” He said then that he believed AI music could augment, but not entirely replace, the compositions of blockbuster film scorers — but it might “cut off the bottom rungs of the ladder” for young upstarts like him.

The following year, Phipps started working at Suno. Now, he is one of dozens of musically inclined employees the AI company has hired, and he acts as a product manager and liaison with the growing class of professional musicians who use the platform. Phipps has become a favorite among those pros, in part because he’s also a musician and because he monitors the company’s VIP program, a private group of thousands of top music-makers who get early access to new products and offer the company feedback. Employees at Suno tell me this exclusive group includes some of the biggest talents in the world — but add that they can’t tell me exactly who.

Over the course of a month, I met with executives at Suno several times to better understand the company — and to decide for myself whether it is deserving of the boogeyman reputation it’s taken on to some in the music business. As part of this reporting, I was invited to join the generative AI music company at one of their songwriting camps during Grammy week. As with the VIP program, I wasn’t allowed to say who was there or where it was. In an email laying out the ground rules, the company’s publicist suggested I “just refer to it as a studio in Hollywood.”

I found a few producers at the camp who were willing to share their names, though, including Timbaland, Gino the Ghost and Om’Mas Keith — the lattermost of whom spearheaded the camp with Suno. Keith is a Grammy-winning record producer who has worked with Frank Ocean, Erykah Badu and Jay-Z, and he welcomed me into the studio to see what he had been working on with his colleagues. “This is a music creation camp. I wanted to bring over every great musician I know for this,” he said, gesturing to the dozen or so musicians in the crowded room, where everyone was chattering excitedly. As a few Suno employees schmoozed with engineers at the massive mixing board, Keith pointed out what each musician at the session excels at: One was a sought-after session drummer, another a platinum-selling record producer, another a top violinist.

“This is all about human involvement; even the prompting is democratic. We do it together,” Keith said. I joked about whether I could get a courtesy credit for being in the room. “No, you didn’t contribute to the prompt. It starts there. You got to be careful who you prompt around,” Keith replied in a serious tone. “These are some of the best musicians in the world.”

While I sat on the studio’s couch with Keith, I witnessed a song come to life. In 30 minutes, what he said started with a prompt and some lyrics became a master recording just as good, if not better, than the scores of pitch songs I’ve heard in my years as a music publishing reporter. But interestingly, by the end, what I hear is approximately 90% human-recorded. The only Suno piece left was the vocals, which I was told would also eventually be rerecorded too.

As several other professional musicians who use Suno tell me, this camp’s approach to AI is a common one: prompting the model dozens of times with original lyrics to reach a song idea that sparks their interest. Then, Keith and his co-writers re-­create the Suno song stem by stem, adding in their own flourishes and subtracting bits that they don’t like. This is partly due to, Keith whispered to me, the fact that while Suno-created music may sound good, it still isn’t perfect, studio-quality audio. It’s also because these music pros simply enjoy molding the platform’s initial output into their own. Optimistically, this approach makes Suno a writing companion that encourages and accelerates human creativity. Cynically, it’s a way to use AI-generated ideas without disclosing the use of an AI model. Sitting there, I wondered how many songs released in the last few years were made this exact way without anyone knowing.

Around this time, Mikey Shulman, the 39-year-old co-founder and CEO of Suno, walked into the room from outside, where he’d been shaking hands with a mix of record-label executives, Suno employees and venture capitalists who, like me, had all come to watch the camp. With his white T-shirt, unbuttoned oxford, athleisure pants and nondescript sneakers, Shulman looks like your typical tech founder. Arms crossed over his chest, he listened to the song, bobbed his head and surveyed the room contentedly. Nearly everyone turned to say hello to him with a big smile. I couldn’t help but wonder how he felt at that moment: Like a god, surveying his own creation? Like Dr. Frankenstein beholding his monster? Or was this just like the buzz that comes from writing a great song?

The dawn of music’s AI age didn’t arrive with a big bang — it stumbled into being, by accident, at a kitchen table in Cambridge, Mass. “It wasn’t exactly, you know, the proverbial three guys in a garage thing,” Shulman jokes.

In January 2022, Shulman and his co-founders, Georg Kucsko, Martin Camacho and Keenan Freyberg, had just left their jobs at Kensho, an AI business tool company acquired by S&P Global in 2018, and were pursuing a new startup together that would help navigate and analyze music libraries and catalogs using AI. “We didn’t even think it was possible to generate music like [we can now],” Shulman explained.

Still, excited by the developments in AI image rendering, the foursome — all hobbyist musicians as well as Ivy League-educated technologists — started staying up late to experiment with generative AI after finishing their other work. One day, the group legend goes, they created some “vaguely passable sounds.” “The first things that came out you would be very generous to call music,” Shulman admits. The pet project quickly grew into an obsession for the group, leading them to abandon their original company idea and start what is now known as Suno (Hindi for “listen”) instead. “Why are we not doing the thing that we can’t put down?” Shulman recalls thinking.

Suno was not the first company to generate music with AI, but it was likely the first to generate singing voices alongside instrumentals with a traditional verse-chorus structure — or simply put, a full song. “Nobody else could make songs, and there’s something really special about a song. A song is a story, it’s not background music,” Shulman says. The team first launched the product on the messaging platform Discord in July 2023 — the same way that popular AI image model Midjourney was first launched — which allowed them to disseminate the model quicker and to Discord’s built-in audience.

Immediately, Suno found a fan base. The product wasn’t anywhere near perfect, but it had promise. “I remember at the beginning we didn’t even have control over the genre, so the models would basically infer the genre from the lyrics,” Shulman says. “We found people hacking their lyrics, adding in words like ‘yo, yo, yo’ at the beginning because they demanded that it came out as rap.”

As Suno developed, the co-founders took on different roles at the company. “Mikey’s the face; Martin’s the soul, overseeing engineering and brand; and Georg is the mad scientist, the machine learning genius, the man behind the curtain,” Sinclair tells me. The final founder, Freyberg, was the company’s COO, but left months before Sinclair arrived in July 2025. “He wanted to do different things,” Sinclair explains with a shrug.

Suno also picked up early evangelists, like Timbaland, who preached that the company was a “new frontier” for music and joined as an ambassador for the company in 2024. Rolling Stone added to the hype that year, calling it “ChatGPT for music.” But around the same time, music business executives around the world began to worry about how Suno’s model got so good so fast. Like ChatGPT, Suno was training on copyrighted material — their companies’ songs — without consent or compensation. Suno was not yet talking about its training data, but one investor told Rolling Stone, “Honestly, if we had deals with labels when this company got started, I probably wouldn’t have invested in it.” It felt suspicious.

In Denmark, Koda, the local collection society for songwriters, composers and music publishers, launched an investigation. “To me, it was obvious what they were doing,” Koda CEO Gorm Arildsen says. “That music had to come from somewhere, and we expected some of it to be ours.” He instructed his full-time tech scout to try prompting Suno for Danish works in Koda’s catalog. Suno has protections so that users cannot prompt songs to sound like specific artists, but Arildsen says “to be honest, it was not that difficult to prompt one of our songs,” which include hits by Aqua, MØ and contemporary Danish phenom Christopher; in the United Kingdom, Ed Newton-Rex, a musician and founder of nonprofit Fairly Trained who formerly worked as Stability AI’s vp of audio, was performing his own tests, which he’d ultimately publish to fanfare in Music Business Worldwide; in Germany, local collection society GEMA, which represents songs recorded by Lou Bega, Milli Vanilli and Alphaville, was doing the same.

“We tried to quantify this impact, and then came to a stunning number,” GEMA CEO Tobias Holzmüller tells me. “Up to 950 million euros [about $1.1 billion] in Germany and France could be endangered in 2028 alone due to generative AI if we let it all happen without taking action.” Its study also found that 27% of music creators’ revenue will be at risk for the same reason by 2028. Holzmüller and Arildsen say they reached out to Suno to talk but neither received a reply. Eventually, both CEOs say, this prompted them to launch separate lawsuits against Suno. GEMA’s was filed in January 2025 and Koda’s in November 2025. Arildsen called the move “the largest theft in music history.”

But Suno’s biggest challenge so far has come from North America, where the major-label groups — Universal Music Group (UMG), Sony Music and Warner Music Group (WMG) — which also found ways to trick Suno into regenerating songs that sounded very much like those in their catalogs, specifically James Brown, Chuck Berry and Michael Bublé. In a rare sign of camaraderie, the rivals teamed to launch a blockbuster $500 million copyright infringement lawsuit against Suno, as well as a near-identical suit against Udio, another significant AI music model on the market, in June 2024. The labels said that using their sound recordings to train Suno’s model was infringement “at an almost unimaginable scale” and that Suno’s resulting songs could “saturate the market with machine-­generated content that will directly compete with” what the labels call “genuine sound recordings.”

Suno held the position taken by other AI companies before them: that it’s “fair use” to train on copyrighted works. Shulman responded to the filing by saying “[the majors have] reverted to their old lawyer-led playbook,” likely a reference to the majors’ past treatment of Napster and file-­sharing, which many now view as a failure. Back then, Shulman spoke more like a provocative, move-fast-and-break-things tech founder. (One quote from a podcast interview during this period follows Shulman to this day: “I don’t think the majority of people enjoy the majority of time it takes to make music.” He now says of that quote: “I really wish I had chosen different words.”)

These days, Shulman is much more subdued. “I do have a lot of respect for music,” he says. “The major labels are very important, but what we did is legal and so that’s what we did.” While he maintains his position about fair use, Suno did agree to a licensing deal and settlement with WMG in November, ending just WMG’s portion of the major labels’ lawsuit. As part of the deal, Suno agreed to retire its current model and launch a new one, sometime this year, that is only trained on licensed copyrights in which the owners opt-in to be part of its data set. Suno also agreed to limit the number of AI songs a user can download in a given month.

It’s a sign that, despite several high-profile law­suits that remain in its path, the stigma around Suno in the music business establishment could be fading. “I don’t think of what we did as a settlement,” Shulman says. “I think about this much more like a partnership … It is much more long term.”

“I’m looking for the Kanye of AI music designers,” says Neil Jacobson, his eyes wide with excitement.

“Obviously, I hope he doesn’t hate Jewish people… but what I mean is he’s going to be brilliant and he won’t apologize,” the music executive continues. “He’ll be like, ‘Yeah, I created this new world, this whole new way of listening to music.’ ” He pictures a day when his AI-driven talents pioneer virtual reality-native entertainment while his human artists continue to dominate in more traditional spaces.

Jacobson, founder and CEO of management, label and publishing company Hallwood Media, is one of Suno’s most outspoken evangelists and is the first music executive to publicly invest in it, through his fund Hallwood Ventures. In September, Jacobson made headlines for signing what he calls “AI music designers” — or the people who use generative AI to make music — to Hallwood, a first in the business, and he even scouted his new AI signee, imoliver, from Suno’s streaming service.

Since then, Jacobson has doubled down on AI-powered talent, signing Telisha “Nikki” Jones, who created AI gospel persona Xania Monet, to what Billboard previously reported was a multimillion-dollar deal. He understands that his bullish AI views will likely lose him some clients (“I have major issues with some of my artists who are really upset about my involvement in AI music, some of which I believe will leave me over [it] — it’s heartbreaking”), but he feels that the negative discourse over AI in music will eventually subside. More specifically, he calls this “artistic fascism” — a colorful way of saying that detractors are policing what art is and how art is made.

Now, Monet’s manager, Romel Murphy, is joining in with a company of his own called dai + drm (pronounced “daydream”), a label joint venture through Create Music Group that will sign AI-powered talent, including Solomon Ray. Timbaland has also been building his own AI entertainment company, Stage Zero, which generates artists from scratch, including his first project, Tata, who was created with the help of Suno and other AI tools.

These so-called “AI artists” or “AI music designers” — everyone in the space seems to have a different name for them — some but not all of whom use Suno, have made an impact on the Spotify, Billboard and TikTok charts, and the broader discourse. The Velvet Sundown, for example, which is touted in Suno’s investor pitch deck with the title “Suno songs go viral off platform,” ignited a debate around the future of music last summer after becoming an early viral example of a seemingly nonexistent band posting AI music on streaming services.

Despite the commotion it caused, which was started by a concerned poster on Reddit, none of The Velvet Sundown’s Suno-generated songs — faithful descendants of classic rock à la Creedence Clearwater Revival and Buffalo Springfield — have more than 5 million plays on Spotify. Monet, who was also created using Suno, climbed to the top of Billboard’s Hot Gospel Songs chart, yet is far from cracking the Hot 100. Ray, a clean-cut AI persona constructed by MAGA rapper Christopher “Topher” Jermain Townsend, hit the top of the Gospel Digital Song Sales chart with the song “Find Your Rest,” but given downloads are an increasingly unpopular form of music consumption, the stat remains a niche measure of success. These milestones last year proved that a breakthrough from an AI-generated act is clearly possible — but to date, it is still yet to truly come.

Some Suno supporters are not concerned with the inevitable rise of AI-driven competition. Timbaland says that people who use Suno to make songs in seconds don’t threaten producers like himself who have “done the over 10,000 hours” of mastery in his craft. Keith agrees. “As a lot of ‘AI slop’ enters the ecosystem, those with true artistry will be differentiated.”

Gino the Ghost, a writer-producer for Sabrina Carpenter, The Chainsmokers and Saweetie who was also at the songwriting camp, adds that he uses Suno “like a more intuitive Splice,” referring to the popular sample and sound library, and that he’s “not worried” about its implications for his career. “This isn’t me being naive,” he assures me. “It will never be good enough to get to that top, top, top level of writing and producing on its own.” Still, he is concerned about the undeniable rise of quick-made slop: “I’m torn as someone who’s a big proponent of songwriter rights and us being paid fairly. I’m worried the royalty pool on Spotify is going to be diluted.”

There’s also a growing concern that the fan base for AI songs isn’t even real. According to research from Deezer, up to 85% of streams on fully AI-­generated songs on its site are flagged as artificial or fraudulent. In late February, several artist rights groups, including the Music Artists Coalition, sounded the alarm on this issue in an open letter titled “Say No to Suno,” raising concerns that “Suno has, in effect, become a fraud-fodder factory on an industrial scale.”

Speaking to those at Suno or those who use it, I’m often met with a mix of total enthusiasm, a little fear and some dark humor. Keith jokes about how one day we’ll all be “in a pod with fluid and hooked up to devices,” and Jacobson sees a vision of the future like the popular dystopian novel Ready Player One.

But regardless of AI music’s current popularity and the discourse around it, the number of AI-­powered songs on streaming services is growing fast and sparking fears about increased competition for human-made songs, which take much longer to create. To stop the flood, UMG has now taken a hard-line stance when negotiating with AI companies, saying that AI partners must instate a “walled garden” to make it impossible for users to download AI music and take it elsewhere on the internet. When Billboard asked if that was why UMG had not yet settled with Suno, UMG chief digital officer/executive vp Michael Nash said, “If I were to treat your question as a rhetorical one, then yes. It’s kind of a hat-hanger in this discussion.”

Shulman says that this “somewhat black-and-white view of things lack[s] nuance. [It can] also really prevent a lot of innovation.” As for whether Suno would ever become a closed environment, Shulman says, “Completely? I just think that’s way smaller than it needs to be.”

The same day that I talk to Jacobson to get his take on the future of AI music, I also call Shulman to ask about his first-ever Grammy week, the famously exhausting run of nonstop parties and networking in the music business ahead of the actual awards ceremony.

“Last week was kind of an eye-opening moment,” he says with a wide grin. “I really felt a change in how excited and curious and optimistic people were in a way I hadn’t felt before. I think there was a lot more acceptance, and a lot more public acceptance of it, which is a really new thing.” He had been invited to WMG’s Grammy party, among other events, for the first time — which felt like evidence that the establishment was ready to talk.

Suno’s stated vision of the future looks a little different from what it has inspired from its biggest fans. Suno is already working to create a “verticalized” service, as noted in the company pitch deck, incorporating a TikTok-like social media feed (Hooks, announced in December), a streaming service and a number of different music tools, to target, as Shulman puts it, anyone from “Grammy winners to grandmas.”

Is it really possible to get everyone to make music? Shulman thinks so. “Our usage is showing that a hell of a lot of people, way more than anybody expected, want this,” he says. The hope is to expand music-making as something more like a pastime, competing for the same hours as one spends scrolling on TikTok, playing video games or watching a movie. “I think we’ve reached peak scrolling,” Shulman explains. “People want to do something else.”

It’s a bold vision of the future, and if Shulman is right, Suno projects itself to reach $1 billion in revenue by 2028. “When we talk about ‘verticalizing’ inside the company, it’s not like we want to smush TikTok and Spotify together,” Shulman tells me. “Those two things already exist, and that is not going to reap a lot of benefits … I’m thinking, ‘How do I make discovery way better than it is now?’ Because we are able to do something no one else can do. [The point of Hooks] is to get you off of the feed, playing with content and remixing it. That’s the kind of discovery that doesn’t exist right now.”

And, he continues, Suno doesn’t have a “fixed pie mentality,” like he thinks much of the music industry does, where “if one thing wins, another has to lose.” As Shulman explains, “We want to grow the pie and make the music industry even bigger.” This thinking is, in part, what attracted Mike Mignano, partner at venture capital firm Lightspeed Ventures, to the platform. He says his team “definitely views [Suno’s] technology as a market expander” for music, even saying he sees Suno becoming “the most important company in music” one day.

Armed with its impressive Series C funding, licensing momentum and the growth of its music off-platform — not to mention Suno’s recent recruitment of longtime Merlin CEO Jeremy Sirota as its new chief commercial officer — Suno’s total takeover of the music business can feel inevitable. But the most common question remains: Does everyone really want to make music?

Suno also faces the challenge that, due to differing views on “walled gardens,” UMG, the world’s biggest music company, may not reach a settlement with Suno before its deadline to retire its current model and to launch the new, fully licensed version later this year. Given musical copyrights are often split between a number of different songwriters and artists across multiple labels and publishers — and all rights holders must opt-in to the use for it to work — it’s possible that Suno’s agreement with WMG has put the platform in a tough position with a limited catalog of training data. When pressed about how many songs he needs to train the new model to produce quality results, Shulman is noncommittal: “It’s really hard to say — more is always better. It’s really hard to give definitive answers here. Whenever you change anything, some things get better and some things worse.”

There’s also, of course, the various lawsuits still active against Suno from GEMA, Koda and the remaining majors — as well as a couple of class action lawsuits from independent songwriters. And while the path to becoming the de facto AI music model has seemed increasingly clear for Suno (especially after its competitor Udio pivoted its offerings in November), now Google — a goliath in general AI development with nearly unlimited resources to burn — seems increasingly interested in entering the AI music race, acquiring ProducerAI and launching its latest version of its own AI music model, Lyria, as part of Gemini in February.

Plus, Suno’s users are primarily 25- to 34-year-old men. Young people, Shulman admits, have “general apprehension around AI everywhere” — a troubling trend, given that teens have historically been the music industry’s greatest cultural drivers and consumers.

“I get so many questions from people in music … and the questions are always like, ‘Is AI going to end the world?’ ” Shulman says. “I happen to think it’s not going to, but certainly Suno is not going to turn everybody into paper clips! That’s not the domain we play, so I say, ‘Why don’t you try it? Most people like it when they try it.’ ”

This story appears in the March 7, 2026, issue of Billboard.