The advice making the rounds among people writing checks into AI music is tidy enough to fit on a napkin: buy the growth, the lawsuits are noise. Suno pulls in serious recurring revenue, users generate songs faster than any label could clear samples, and the litigation from the majors will end the way these things usually end — a settlement, a licensing handshake, a footnote. Price the risk, take the position, move on.
It is a reasonable-sounding rule. It is also, on inspection, about sixty percent right, which is the most dangerous kind of right. So it is worth walking through where the napkin holds and where it tears, because the gap between those two is exactly the spread you are being asked to underwrite when you value an AI music generation company in the billions while it is actively being sued by the industry it draws from.
Where the advice is roughly right
Start with the part that holds. The growth is not a story problem. A platform in this category can post subscriber counts in the millions and annualized recurring revenue in the hundreds of millions inside a few years of launch, and the usage is genuinely dense — people do not sign up for a text-to-song tool and generate one track; they generate dozens, iterate on prompts, re-render, chase a hook. That behavior produces the kind of engagement and retention curves that make a growth investor's math work regardless of what the legal department is doing down the hall.
The settlement logic also has real precedent behind it. When a new distribution or creation technology collides with music rights, the sector's history bends toward licensing rather than annihilation. Radio, cable, satellite, digital streaming — each arrived as a supposed existential threat and each ended up inside a rate structure. The recent movement where at least one major stepped away from the courtroom posture and toward a deal is consistent with that pattern. So when a strategist says "this settles," they are pattern-matching to a century of the industry choosing rent over ruin. That instinct is sound.
If you stopped reading there, you would conclude the napkin is correct. Keep reading.
Where it breaks down
The first crack is that "it settles" and "it becomes a profitable business" are two different claims, and the advice quietly collapses them into one.
A settlement resolves past liability. It does not, by itself, define the ongoing cost of doing business. The question that actually determines whether the valuation is sane is not will there be a license but what does the license cost per unit of output, forever. If model training and generation require an ongoing royalty to rights holders — a per-stream, per-generation, or revenue-share obligation baked into the cost of goods — then the company is not a software business with software margins. It is a licensing intermediary with a compute bill on top. Those are very different multiples. The napkin prices the first company and buys the second.
The second crack is the expanded claims. Copyright litigation in this space has not stayed put at "you trained on our recordings." It has reached toward the mechanics of how outputs get produced and how catalogs may have been acquired to feed the machine. Every expansion of the claim surface changes the settlement's shape — it raises the floor on what rights holders can demand and widens what a license would need to cover. An investor treating the suit as a fixed, knowable liability is mispricing a moving one.
The third crack is the one people skip because it complicates the trade: the plaintiffs are not a bloc.
The majors are not moving as one
The convenient version of this story has "the labels" on one side and the AI companies on the other, resolving in a single grand bargain. Reality is messier and more consequential. When one major pursues a settlement-and-license path while others press the litigation and expand it, you do not get one outcome — you get a fractured rate environment where the company must cut separate deals on separate terms with separate leverage.
That fragmentation matters to the valuation in a specific way. A single industry-wide license is a knowable cost you can model. A patchwork of bilateral deals, each negotiated against a different threat posture, is a cost structure that can drift upward every renewal cycle, because each rights holder prices its catalog against what the others got. The advice that says "the lawsuits will resolve" is technically true and analytically useless, because resolve can mean a clean rate card or a permanent negotiation treadmill, and those imply valuations that are multiples apart.
What the money is actually pricing
So what is a large late-stage round at a billions-dollar valuation actually a bet on? Not that the lawsuits vanish. The smarter capital in this space knows they do not. It is a bet that licensing becomes a line item the business can carry — an operating expense the revenue absorbs while still leaving a defensible margin — rather than a toll that eats the model alive.
That is a genuine, investable thesis. It is also a narrower one than the napkin admits. It requires three things to hold at once: that per-output license costs land low enough to preserve margin, that the fragmented deal landscape converges toward something predictable instead of ratcheting, and that user willingness to pay survives whatever price increases the licensing burden eventually forces downstream. Miss on any one and the growth curve stays impressive while the equity story does not.
There is an upside case where this works cleanly, and it is not fantasy. A tool that lets a game studio generate an adaptive loop at 92 BPM in a minor key with proper stems, or a video editor clear a Friday-deadline cue without a stock-library subscription, is solving a real and paid-for problem. Demand is not the risk here. The risk is entirely in the cost of the inputs, and the inputs are owned by the people suing.
The honest version of the rule
So keep the growth. Respect it — it is real, and it is why this is a conversation about valuation instead of a wake. But strip the napkin of its comforting shorthand. "The lawsuits will settle" is not the thesis. The thesis is that the settlement's terms will leave a business standing that looks more like software than like a rights clearinghouse, and that claim rests on numbers nobody has yet — the per-generation license cost, the convergence of a split label front, the price elasticity of users who have never had to pay the true input cost.
Underwrite that, not the headline. Here is the honest version to use tonight: bet on whether the license becomes an affordable line item, because the lawsuit ending was never the question — the price of it ending is.
Not sure which tool to use?
Compare the top AI music and sound tools side by side — honest reviews, real pricing, no sponsorships.
