The Deal Matters Less Than the Word "Opt-In"

On August 12, AI music platform Suno and BMG announced a global partnership — in their own words, "a global alliance that establishes a strategic framework covering BMG's recorded and music publishing repertoire."

The headline ran as "Suno lands its second major label deal." Warner Music Group signed about nine months earlier, so this is the next one in sequence. But the thing worth reading in this deal isn't the partner's name. It's the mechanism.

BMG artists and songwriters have to opt in individually. That applies both to their music being used as training data (inputs) and to it feeding what Suno users generate (outputs). If they don't opt in, their work isn't licensed to Suno at all. This is not a label handing over its catalog wholesale and settling up later.

And there's one more clause you can't skim past. The agreement settles Suno's prior use of BMG recordings and publishing works. Given where the conflict between AI music companies and labels started, that single line sets the weight of the entire deal.

Where Suno and BMG Each Stand

Suno turns a text prompt into a finished song — lyrics, melody, vocals, arrangement, all at once. In June 2026 it raised over $400 million at a $5.4 billion valuation, and it has passed 100 million users. Co-founder and CEO is Mikey Shulman.

The company has spent two years standing on legally uncomfortable ground. From the music industry's point of view, Suno probably trained on their catalogs, and lawsuits followed in the US. The awkward part is that whether Suno won or lost those suits, staying in business eventually required real licenses.

BMG is Bertelsmann's music company and sits on a different tier from the big three of Universal, Sony, and Warner. But its position in this deal is about structure more than scale. BMG holds both recorded music and publishing under one roof. AI training needs both the recording rights and the composition/lyric rights, and there aren't many counterparties who can negotiate both at once.

The executive quotes point the same direction. Shulman said that "together with BMG, we can develop new experiences that give artists and songwriters real choice, generate new revenue," and BMG EVP of Global Marketing & Streaming Celine Joshua said the agreement "establishes strong protections, clear economics, and new creative possibilities for artists and their fans." Both led with choice and economics.

Unpacking the Terms

Item Detail Note
Announced August 12, 2026
Scope BMG recorded + publishing repertoire Global
Participation Individual opt-in by artists and songwriters Not licensed otherwise
Coverage Both inputs (training) and outputs
Compensation Paid to opted-in rightsholders Rates undisclosed
Prior use Settled by this agreement
Sequence Second major deal after Warner Warner was ~9 months earlier
Unresolved Universal and Sony still litigating in the US
Suno valuation $5.4B June 2026, $400M+ raised
Users 100 million-plus

The most important row is "both inputs and outputs." The AI music licensing argument has fixated on training data for years. But what artists actually fear isn't training — it's the outputs: an endless supply of songs that sound like their voice and their style. That this deal explicitly covers the output side signals the argument has moved.

The second row to sit with is prior-use settlement. That's effectively a settlement clause, and undisclosed money almost certainly changed hands. For Suno, it closes the legal exposure on BMG's catalog.

Third: no rates were disclosed. "Clear economics" appears in the announcement; numbers don't. Whether an opted-in artist earns per generated track, per usage volume, or on some other basis is entirely unstated. Evaluating this deal ultimately requires that number, and it doesn't exist publicly.

What Each Side Gets

Suno gets legitimacy. The biggest business risk for an AI music company was never the technology — it was litigation. Each major deal weakens the "service built on unlicensed training" frame and makes advertisers and platform partners easier to sign. And for a company preparing a fully licensed new model, contracted catalog is product raw material.

BMG gets control and a precedent. Winning the opt-in structure means the label held the stronger hand. Instead of grinding through years of litigation, it chose conditional participation — and if those conditions become an industry standard, the next negotiation is easier.

BMG's artists get choice, and that's the substantively new thing here. Individual artists have had no voice in the AI training argument: the label signs, the catalog moves. What matters under an opt-in regime is whether declining carries any penalty, and that part isn't in the announcement.

Universal and Sony are holding different cards. Both are still suing Suno in the US. Maintaining the suits is probably about negotiating position more than damages. BMG and Warner signing first establishes a market price, and if that price isn't good enough, litigating is a rational way to wait for a better one.

Rival AI music services get squeezed. As Suno accumulates licenses, "we trained on properly licensed data" becomes a differentiator. Everyone else either lands the same deals or lives with being the unlicensed option.

Two Endings for Content Industries Meeting New Technology

Music has been here before, twice, with opposite outcomes.

The first was Napster. The industry sued and shut it down. Legally a total victory, and the result was bad anyway. Demand didn't disappear; it moved to unlicensed downloads. Apple walked into the gap and rebuilt distribution around iTunes. The industry won the lawsuit and lost control of distribution.

The second was Spotify. This time the labels took equity and contract terms and participated. Streaming restored industry revenue, but complaints about what reaches individual artists have run for more than a decade. The industry survived, and the trust between creators and the business cracked in the process.

The opt-in structure reads as the lesson from both. Don't only litigate (Napster), and don't hand over the whole catalog (Spotify). Leave the decision with the individual creator.

There's room for it to fail, though. The obvious failure mode is opt-in that works only on paper. If a label uses contract renewals or promotional allocation as leverage, "voluntary consent" becomes coercion with extra steps. The inverse failure is an opt-in rate so low that the deal is meaningless, which would give Suno grounds to demand opt-out structures in the next negotiation. The success or failure of this deal will be visible not in the press release but in the opt-in rate six months from now.

Spotify Moved the Same Week

Put this deal next to Spotify's August 11 announcement and the picture sharpens. Spotify said it will attach an "AI Persona" label to AI-generated artist identities and exclude labeled profiles from recommendations by default.

The two landing in the same week isn't coincidence — they're complementary. Suno's deal governs the input side: which music can serve as material, decided by rightsholder consent. Spotify's policy governs the output side: making sure what gets generated can't pass as human, and pulling it out of recommendations.

Competitively, this favors Suno. If distribution platforms move toward labeling and managing AI music rather than banning it outright, licensed AI music gets a legitimate place to exist. Unlicensed services get blocked at both ends instead.

Competitor counterplays run three ways. Sign the same label deals. Or lead with models trained exclusively on pre-licensed data from the start. Or avoid commercial music entirely and go where rights are simple — background music, games, video soundtracks. The third is safest and the smallest market.

The big remaining variable is Universal and Sony. Whether they eventually convert to deals or litigate to the end will determine the legal standing of AI music. And that decision turns on how good the terms BMG and Warner extracted end up looking.

So What Actually Changes

For BMG artists, there's a real decision to make: opt in or not. The material for that decision is incomplete. Compensation rates aren't public, and how an opted-in song actually feeds outputs isn't specified. The concrete step available now is asking your label in writing for the rate structure and the withdrawal terms.

For music creators generally, there's a precedent. Opt-in can be a contract term rather than an aspiration. That's one more thing you can ask for when negotiating with any label or distributor.

For people using AI music tools, the commercial usability of outputs should get clearer over the medium term, since a model trained on licensed catalog has cleaner rights downstream. That doesn't retroactively clean up outputs from models already shipped.

For content industry people generally, the architecture is the takeaway: govern inputs with rightsholder consent and outputs with platform labeling. Music is the first test, and the same frame is likely coming for video, images, and text.

For music businesses outside the US, the timing matters. Global labels are setting the price and terms for AI training licenses right now, and once those harden, whoever negotiates later inherits the terms. Whether opt-in becomes the standard isn't a thing to watch — it's a thing to be in the room for.

🥄 Three Things You're Probably Wondering

— So Suno is copyright-clean now? Only for opted-in works within the BMG and Warner catalogs. Universal and Sony are still litigating, and nothing here resolves the full question of what Suno trained on historically. If you're using outputs commercially, read the service's output-rights terms directly.

— How much do opted-in artists actually get? Undisclosed. Both sides said "clear economics" and neither published a number, nor whether it's per track or per usage. Judging whether this is a good deal is premature without it.

— Will Universal and Sony sign too? Too early to call. But their reason for litigating is probably not just damages. BMG and Warner are setting a market price, and both appear to be holding out for better. As long as leverage remains, the lawsuit is the stronger card.

References

Numbers and criteria are as of announcement and may change.