One Stream, Six Rights Holders, a Fraction of a Penny

Press play. In the time it takes a song to load—a beat, maybe two—a streaming service has just done something legally astonishing. It has reproduced a musical composition onto a server, transmitted a public performance of that composition to a listener, reproduced and performed a sound recording embodying that composition, and triggered payment obligations to as many as four collective rights organizations across potentially dozens of countries, all to be reconciled against a catalog of roughly 100 million songs. The listener experiences three minutes of music. The system underneath experiences a clearance event implicating at least six distinct categories of rights holders, settled in fractions of a cent.

The money is not a fraction of a cent. The global recorded-music industry generated $28.6 billion in 2023 by the IFPI's count, with streaming roughly two-thirds of it, and music publishing—the business of song copyrights, distinct from recordings—added something like $15 billion more. Yet for most of the last decade the industry could not answer a question a child would think simple: when a song plays, who, exactly, gets paid, and how do you find them? The answer turned out to require an act of Congress, a new nonprofit handling more than a billion dollars, a half-billion-dollar pile of money owed to people nobody could identify, and a federal lawsuit over whether an audiobook is enough of a "product" to cut a songwriter's check in half. This article is about that machinery—how it broke, how the Music Modernization Act rebuilt it, and where it still leaks.

To keep the abstractions honest, follow two characters through the system. Áine is an independent singer-songwriter. She writes her songs, records them in a spare bedroom, owns both the songs and the recordings outright, and pushes them to streaming services through a digital distributor. She has signed nothing away. Cascade is a fictional interactive streaming service—think of a smaller Spotify—that wants to make Áine's catalog, and everyone else's, available on demand to its subscribers. Almost every difficulty in music licensing lives in the gap between Áine pressing "upload" and a Cascade subscriber pressing "play." This article walks that gap end to end: what rights exist, who controls them, what each license costs, who collects it, and—crucially for an independent like Áine—what it takes to actually receive money you are unquestionably owed. (For the broader question of how platforms manage their content-licensing and liability exposure, see our analysis of Section 230 reform and platform liability.)

The Anatomy of a Song: Two Copyrights, Not One

Start with the distinction that trips up musicians, executives, and more than a few lawyers: every piece of recorded music contains two separate copyrights, owned by different people, licensed through different systems, generating different royalties collected by different organizations on different schedules. Miss this, and nothing downstream makes sense.

The first copyright protects the musical composition—the song itself: melody, harmony, lyrics, structure. It is born the moment the songwriter fixes it (humming a tune into a phone counts, once written down or recorded), and it belongs initially to the songwriter or, for a typical modern pop song with five co-writers, the songwriters jointly. In the ordinary case the writer assigns or licenses administration of that copyright to a music publisher. The second copyright protects the sound recording—a specific, fixed performance of that composition, the actual audio file. It belongs initially to the performing artist (and often the producer) and is, in the conventional industry deal, owned or controlled by a record label. Industry slang calls the recording "the master."

The cleanest way to feel the difference is to watch what happens with a cover. When an artist records a new version of someone else's song, the cover creates a brand-new sound-recording copyright in the new performance, while the composition copyright stays put with the original songwriter, who must be paid a mechanical royalty for the cover. That is why licensing an existing recording for a film requires clearing both copyrights—a sync license for the composition from the publisher and a master-use license for the recording from the label—but a filmmaker who commissions a fresh, original recording of the same song needs to clear only the composition. Two copyrights, two licenses, two checks.

For Áine, this is not doctrine; it is half her income and most of her paperwork. Because she wrote and recorded her songs, she owns both copyrights—a happy position, but a doubly demanding one, because the two halves are administered through entirely separate plumbing. Register on one side and neglect the other, and half her money silently goes uncollected. The conventional division of labor looks like this:

Copyright Initial owner Typical administrator Royalties it generates
Musical composition Songwriter(s) Music publisher Mechanical + public performance (composition)
Sound recording ("master") Performing artist / producer Record label Master royalties (interactive streaming, sync); SoundExchange (non-interactive digital)

Both copyrights are themselves bundles of the exclusive rights granted by 17 U.S.C. § 106: the right to reproduce, to prepare derivative works, to distribute copies, to perform publicly, and to display publicly. (For sound recordings specifically, § 106(6) grants only a digital public-performance right—a quirk we will return to, and one that explains why terrestrial AM/FM radio pays songwriters but not the artists whose records it spins.) Different rights in the bundle get exercised by different uses: a service storing a song on its servers exercises the reproduction right; a transmission to a listener implicates the public-performance right; a download implicates distribution; a sample, remix, or arrangement implicates the derivative-work right. Much of the field's baroque complexity exists because separate collective organizations grew up, often a century ago, to administer separate slices of these bundles, and the law has frozen those historical accidents into place. For the foundational mechanics of how these rights arise and why registration matters, see our comprehensive guide to copyright registration; for the song-specific version aimed squarely at musicians, see understanding copyright registration for a song.

A Taxonomy of Music Licenses

Before tracing the streaming money, it helps to lay out the full menu of licenses, because "music licensing" is not one thing—it is at least seven, and a single use can require several at once. Practitioners sort them by which copyright they touch.

On the composition side: a mechanical license authorizes reproduction and distribution of the composition in "phonorecords"—physical discs, downloads, and (the streaming-era twist) the server copies and stream-delivery that interactive streaming entails. A public-performance license authorizes performing the composition publicly, whether by a streaming transmission, a radio broadcast, a concert, or background music in a restaurant. A synchronization ("sync") license authorizes using the composition in timed relation to visual images—a film, TV show, commercial, video game, or YouTube video—and, because syncing inherently reproduces the song and arguably creates a derivative work, it draws on both the reproduction and derivative-work rights. A print license covers sheet music and lyric reproduction. A composition sample license covers lifting a snippet of one song into another. (Two specialized cousins fall outside this article's scope but are worth naming so you know they exist: grand rights, for using songs in a staged dramatic-musical production, and karaoke licenses.)

On the sound-recording side: a master-use license is the recording's analog to a sync license—it authorizes using a specific recording in or as part of another work, most often an audiovisual one. A sound-recording sample license covers sampling the actual recorded audio (which is why clearing a sample usually means two licenses: one from the publisher for the composition and one from the label for the master). And the digital public-performance license under § 114, administered by SoundExchange, covers non-interactive digital transmissions of the recording.

The reason this taxonomy matters is that streaming does not use one license; it stacks several. Interactive on-demand streaming of a single track requires, on the composition side, both a mechanical license and a public-performance license, and on the recording side a license from whoever controls the master. That is three or four separate authorizations, from three or four separate licensors, for one play. Hold that thought—it is the engine of every rate fight in the industry.

The Mechanical Right and the §115 Compulsory License

The word "mechanical" is a fossil. It comes from the player-piano age, when reproducing a song in a "mechanical" device—a piano roll, a music box, then a phonograph record—required the composition owner's permission. Music publishers of the early 1900s feared that the dominant maker of piano rolls would lock up exclusive deals and monopolize the reproduction of popular songs. Congress's answer in the Copyright Act of 1909 was radical and durable: a compulsory license. Once a composition has been recorded and distributed to the public with the owner's consent, anyone else may make and distribute their own recording of it by following statutory procedures and paying a government-set rate. The owner cannot say no. That regime, modernized many times, now lives at 17 U.S.C. § 115.

So a mechanical license is, in copyright terms, simply a license to reproduce and distribute a composition in a phonorecord—and when Cascade ingests Áine's track onto its servers and streams copies to subscribers, it is exercising the reproduction right and needs a mechanical license to do it. (That interactive streaming implicates the mechanical right at all was once genuinely contested; the settled consensus, reflected in § 115 and the rate regulations, is that interactive streaming requires both a mechanical and a performance license for the composition.)

Two features of § 115 deserve underlining. First, the statutory rate is not whatever the market bears; it is fixed by the Copyright Royalty Board (CRB). For physical records and permanent downloads, the headline mechanical rate has long been 9.1 cents per copy, or 1.75 cents per minute of playing time, whichever is greater (37 C.F.R. §§ 385.10–385.11)—so a seven-minute album cut earns more than the flat penny rate. Ringtones, oddly, carry their own statutory rate of 24 cents apiece. For interactive streaming and limited downloads, there is no flat penny figure at all; the rate is computed by a revenue-and-usage formula (37 C.F.R. §§ 385.20–385.22) that we unpack below. Second, although § 115 is compulsory, sophisticated licensees historically preferred to negotiate voluntary mechanical licenses on essentially the same economic terms, because the statutory license's notice, accounting, and audit formalities were a genuine nuisance to comply with. The largest mechanical-rights collecting society, the Harry Fox Agency, built a business issuing voluntary mechanicals on behalf of tens of thousands of publishers precisely to spare licensees that paperwork—though Harry Fox's role has receded sharply since the MLC took over blanket digital mechanicals in 2021.

Before the Music Modernization Act, obtaining streaming mechanicals at scale through § 115 was a slow-motion catastrophe. The compulsory license theoretically let a service get mechanical rights by serving a "Notice of Intention" on each copyright owner and paying the statutory rate—but that machinery assumed a world in which licensees knew who owned what. Streaming demolished that assumption. A service launching with millions of tracks faced ownership data that was incomplete, conflicting, and frequently unknown even to the publishers themselves, who often could not say which songs they controlled. The majors papered over the problem by cutting direct deals with major publishers, but the long tail—independent publishers, and self-published bedroom writers exactly like Áine—frequently got nothing, because the service literally could not find them to serve a notice or cut a check.

The result was unpaid royalties and litigation on an industrial scale. Spotify alone faced class actions alleging it had reproduced compositions without proper mechanical licenses; one such case settled for $112.55 million (with Spotify admitting no wrongdoing), and a separate suit by Wixen Music Publishing demanded damages in the billions before settling confidentially. The Copyright Office, in its landmark 2015 study, described the licensing system as one no rational person would design. For a writer like Áine, the abstraction had a concrete shape: her songs streamed, the mechanical royalties accrued, and the money piled up unmatched—stranded in an industry "black box" she had no practical way to claim. The mechanical right she held by law produced nothing she could deposit at the bank.

The Music Modernization Act and the Birth of the MLC

The fix took an act of Congress, and it came—rare for copyright—with near-unanimous, bipartisan support. The Music Modernization Act (MMA), signed October 11, 2018 (Pub. L. 115-264), was the most consequential overhaul of U.S. music copyright in forty years. It bundled three titles. Title I (the Musical Works Modernization Act) created a blanket mechanical license for digital services and established a new collective to run it. Title II (the Classics Protection and Access Act) finally extended federal protection to pre-1972 sound recordings, which had floated in a patchwork of state law—an issue litigated hard by Flo & Eddie of the Turtles, whose suits against SiriusXM helped force the question. Title III (the Allocation for Music Producers Act) codified the right of producers and engineers to a share of SoundExchange royalties via the longstanding "letter of direction" practice.

Title I was the earthquake. Under the new blanket-license framework (17 U.S.C. § 115(d)), a digital service like Cascade no longer clears songs one at a time. Instead it obtains a single license covering every musical work available for licensing, by (1) submitting a notice of license, (2) paying the CRB-set statutory rates, (3) delivering detailed monthly usage reports identifying every track streamed, and (4) engaging in good-faith efforts to match the recordings it played to the underlying compositions it owes on. Critically, on January 1, 2021, the blanket license was automatically substituted for the compulsory licenses digital providers had previously cobbled together (§ 115(d)(9)). The blanket license covers the mechanical right only—Cascade still needs performance licenses from the PROs and master licenses from labels—but by eliminating song-by-song mechanical clearance, it solved the scale problem that had made the old regime unworkable. Just as significant, the MMA gave services that comply with the new system a limitation on liability for past unmatched uses, retiring the wave of class actions. The deal, in effect: services got peace and a workable license; songwriters got a collective built to find and pay them.

That collective is the Mechanical Licensing Collective (MLC), the nonprofit the Copyright Office designated to administer the blanket license. It opened for business on January 1, 2021 and is now one of the largest music-rights organizations in the country. By statute its board seats both major publishers (Universal, Sony, Warner Chappell) and songwriters—major-affiliated and independent alike—a mandated dual representation meant to keep the collective from serving only the largest catalogs. Its core jobs are five: administer the blanket license; build and maintain the most comprehensive musical-works database ever assembled; match recordings (identified by their ISRC codes) to compositions (identified by their ISWC codes); distribute royalties to the right owners; and manage the historical pool of unmatched money collected before it existed. The scale is real—the MLC has now distributed over $1 billion in mechanical royalties since launch. For Áine, the MLC is the institution that, for the first time, lets an independent writer be paid streaming mechanicals at the same scale as a major—provided she registers her compositions accurately, with complete writer and publisher shares and a correct ISWC, because a database can only match what it can identify.

The Matching Problem and the Half-Billion-Dollar Black Box

The matching problem and the unmatched-royalty pool deserve a close look, because together they reveal precisely what the MMA fixed and what it structurally could not. The mismatch is conceptual: when a service reports a stream, it knows the recording it played—it has the ISRC. But the mechanical royalty is owed on the underlying composition—identified, if at all, by an ISWC—and the two are linked only if a human being somewhere supplied the connection. When the link is missing (the composition was never registered, or was registered with conflicting splits, or two publishers each claim 100%), the royalty cannot be paid to anyone, and it accrues as unmatched.

The MMA confronted the historical backlog of such money head-on. When the blanket license took effect, digital services transferred their accumulated unmatched royalties to the MLC—a combined $424.4 million in legacy "black box" money handed over by twenty providers, a figure that has since climbed past $1 billion in cumulative receipts as monthly accruals continue and a sizable unmatched balance persists. The statute directs that, after a diligent matching effort and a holding period, remaining unmatched money may be distributed to publishers by market share—each publisher receiving a slice proportional to its overall blanket-license usage.

This was a pragmatic answer to a genuinely intractable problem, but it carries a built-in tilt, and the tilt runs against the very people the MMA was sold to protect. Distributing unmatched money by market share sends it disproportionately to the largest publishers, who already hold the most matched works—while the independent writer whose unmatched royalty helped fill the pool may receive nothing from it, because she is too small to register on the market-share formula. The largest publishers, in other words, collect a share of money earned by catalogs they do not own. That is why independent-writer advocates watch the MLC's distribution policies with a wary eye, and why accurate, timely registration is not a clerical nicety for someone like Áine but the difference between income and quiet, permanent loss. Every composition she fails to register correctly is not merely a delayed payment; it is a royalty at risk of being swept, eventually, into a market-share distribution that flows to the majors instead of to her. The MLC's design rewards diligence and punishes neglect—not with a penalty, but with silence.

Performance Rights: The PROs and the 1941 Consent Decrees

The mechanical right is only half the composition's streaming income. The other half is the public-performance right, and it is administered by an entirely different set of organizations: the performing-rights organizations, or PROs. This bifurcation—one composition, two collectives, one for "mechanicals" and one for "performances"—is the single most confusing structural feature of the system, and it exists for no better reason than history.

In the United States, four PROs license performance rights in compositions. ASCAP (the American Society of Composers, Authors and Publishers, founded 1914) and BMI (Broadcast Music, Inc., founded 1939) are the giants, together controlling the performance rights to the large majority of songs. SESAC (now owned by the same group as the Harry Fox Agency) and GMR (Global Music Rights, the upstart founded by Irving Azoff to represent a small roster of marquee writers) fill out the field. Each PRO does the same basic work: it licenses performance rights on behalf of its affiliated writers and publishers, collects from anyone who performs music publicly—streaming services, broadcasters, venues—and distributes royalties to its members. They issue blanket licenses covering their entire catalogs, which is why a service or broadcaster that plays a wide range of music typically needs licenses from all four: no single PRO can promise that every song will be covered.

Here antitrust law enters, and it never quite leaves. ASCAP and BMI have operated since 1941 under federal antitrust consent decrees, the legacy of Justice Department concern that two organizations controlling performance rights to nearly all popular music could extract supracompetitive rates. The decrees impose three signature constraints: ASCAP and BMI must license any applicant who asks (no refusals, no discrimination); they must make their catalogs available on non-discriminatory terms; and when they cannot agree on a rate with a licensee, the dispute goes to a federal "rate court" (the Southern District of New York) that sets a reasonable rate by court order. SESAC and GMR, as for-profit entities never subjected to the decrees, operate free of these constraints—they can refuse to deal and can hold out for whatever the market will bear, which is exactly why GMR's licensing standoffs with the radio industry have repeatedly threatened to pull marquee songs off the air. For a service like Cascade, the upshot is that performance licensing stacks atop mechanical licensing: it maintains separate licenses from ASCAP, BMI, SESAC, and GMR, each with its own rates, reporting cadence, and timing.

Organization Type Approx. U.S. share Defining feature
ASCAP Nonprofit ~40% 1941 consent decree; SDNY rate-court supervision
BMI For-profit (since 2024) ~45% 1941 consent decree; SDNY rate-court supervision
SESAC For-profit ~5–8% No decree; can refuse to license
GMR For-profit ~3–5% Newest; selective marquee roster

The performance royalty is wholly separate from, and additional to, the mechanical royalty. When Cascade pays the MLC for mechanicals, it also pays the PROs for performances—and then pays the labels for the masters on top of that. These three costs stack on the same dollar of subscription revenue, which is the structural reason rate fights are perpetual: mechanical, performance, and master rights holders are each laying claim to the same revenue stream, the service argues the total leaves no margin, and every participant believes someone else is overpaid. For Áine, the practical lesson is blunt: writing and registering with the MLC captures her mechanicals, but she must separately affiliate with a PRO and register every song there, because her performance royalties are collected on a different track entirely and will never reach her otherwise. ASCAP and BMI accept all comers; SESAC and GMR are invitation-only. Choose one, register accurately, and check the work.

Following One Stream: Where the Money Actually Goes

Abstractions dissolve the instant you ask who gets paid for a single play, so trace one. A Cascade subscriber streams one of Áine's songs once. Because Áine wrote and recorded it and owns both copyrights, that one stream sends money down three separate channels, with Áine at the end of all three—but only if she has personally connected to each.

Channel one—the mechanical, on the composition. Cascade pays the MLC under its blanket license. The MLC matches the recording (ISRC) to Áine's registered composition (ISWC) and, if her registration is accurate, pays her the mechanical share computed under the Phonorecords IV formula. Channel two—the performance, on the composition. Cascade pays Áine's PRO, which matches the performance to her registered song and pays her the writer's share and—because she self-publishes—the publisher's share of the performance royalty too. Channel three—the master, on the sound recording. Because Cascade is interactive, it pays the label share under its deal with Áine's distributor; since Áine owns her master, that share flows to her through the distributor's accounting.

Three channels, three organizations, three registrations. A gap anywhere means earned money that is never received. If Áine never registered the composition with the MLC, the mechanical sits unmatched in the black box. If she never affiliated with a PRO, the performance royalty is collected from Cascade but cannot be routed to her. If her distributor mishandles the metadata, the largest of the three shares—the master—can go astray. The single stream teaches the field's central practical truth, the one that separates a major-label artist from a bedroom songwriter: being owed money and receiving it are different things, divided by the unglamorous administrative work of connecting to each collecting body. A major label or publisher has whole departments that do this connecting automatically, as breathing. An independent like Áine must do it herself, song by song, registration by registration—and the difference between a careful registration and a careless one is the difference between getting paid and watching her own streams enrich a black box that pays the majors.

The Master Side: Direct Deals, §114, and SoundExchange

The sound-recording copyright—the master—is licensed through machinery entirely separate from the composition, and the two streaming worlds diverge sharply depending on whether the listener gets to choose.

For interactive streaming, where users pick what to play (Spotify, Apple Music, Cascade), there is no compulsory license for sound recordings. Services must negotiate direct licenses with labels. The three majors—Universal, Sony, Warner—command the bulk of streaming catalog and negotiate comprehensive, confidential catalog-wide deals; independents negotiate their own, frequently through aggregators or distributors. These agreements typically pay labels a percentage of the service's revenue (historically in the neighborhood of 50–55% of total content costs, shared across all the labels by stream-share), often with per-stream or per-subscriber floors. The label then pays the artist under the individual recording contract—where, after the label recoups its advances and recording costs and takes its customary share, a conventionally signed artist may see something in the range of 15–20% of the master royalty, and may see nothing until recoupment is complete. The negotiating dynamic is a standoff of mutual dependence: the majors control so much essential music that no service can launch without them, and the major services command so many subscribers that no label can afford to be absent. The result is a set of stable, fiercely debated deals that neither side loves and neither can escape.

For non-interactive streaming—internet radio and similar services where the user cannot summon a specific song on demand (Pandora's radio mode, SiriusXM, Music Choice, webcasters)—an entirely different and far more artist-friendly system governs. Section 114 creates a compulsory license for the digital public performance of sound recordings (recall that § 106(6) grants a digital-performance right but no general one, which is why old-fashioned AM/FM radio pays the songwriter through the PROs but pays the recording artist nothing). That § 114 compulsory license is administered by SoundExchange, the nonprofit Congress's framework spun up for the purpose. SoundExchange collects CRB-set statutory rates from qualifying non-interactive services and distributes them in a fixed statutory split: 50% to the sound-recording owner (usually the label), 45% to the featured artist, and 5% to non-featured musicians and vocalists through union-administered funds.

The crucial, almost subversive feature is that SoundExchange pays the artist's 45% directly to the artist, regardless of the recording contract and regardless of whether the artist is recouped. This is a deliberate departure from the traditional model in which the label collected everything and the artist saw money only after recoupment. The statutory split cannot be contracted away. For Áine, who owns her masters, SoundExchange is found money she must simply claim by registering: the artist's 45% is hers by federal law, but only if she signs up to receive it.

The asymmetry between the two streaming models is worth dwelling on, because it dictates how much control an artist retains. On the interactive side—where most streaming revenue is generated—there is no statutory backstop: the master royalty is whatever the label negotiated with the service, and the artist receives whatever the contract provides, which after recoupment can be a sliver of what the stream earned. A conventionally signed artist has, in effect, sold most of that royalty in advance for the label's investment. On the non-interactive side, by contrast, Congress hard-wired the split and routed the artist's 45% around the label entirely, so even a hopelessly unrecouped artist still collects the SoundExchange share. The lesson for Áine, who signed nothing away, is that she captures the full master royalty on the interactive side (through her distributor) and the full statutory shares on the non-interactive side (through SoundExchange)—but, once again, only if she has registered with both her distributor's accounting system and SoundExchange. Owning your masters is worth a great deal, but only to an artist who has connected to the bodies that pay them.

Sync, Master-Use, and Controlled Composition Clauses: The Negotiated World

Not all music money flows through compulsory licenses and collectives. A large, lucrative, and entirely negotiated corner of the business handles the use of music in audiovisual works—film, television, advertising, video games, and the river of online video—and it operates by old-fashioned deal-making, song by song, with no statutory rate to fall back on.

A synchronization (sync) license authorizes using a composition in timed relation to visual images. There is no compulsory sync license—§ 115's compulsory mechanical is, by its terms, audio-only and does not cover music videos or audiovisual works—so a producer who wants a song in a movie must negotiate directly with the music publisher, and the fee is whatever the parties agree (a placement in a Super Bowl ad and a placement in a student film are not priced alike). And here the two-copyright structure reasserts itself with a vengeance: a producer who wants a specific existing recording in a scene needs two negotiated licenses—a sync license for the composition from the publisher and a master-use license for the recording from the label. The master-use license is the recording's analog to a sync; its fee turns heavily on the fame of the artist and the recording, which is often the single biggest cost driver. A music supervisor who can't afford the original Aretha Franklin master sometimes licenses only the composition (the cheaper of the two) and commissions a fresh cover recording—legal, common, and the reason you so often hear an unfamiliar voice singing a familiar song in a commercial.

Two further wrinkles matter to creators. First, even where a writer has assigned administration to a publisher, music-publishing agreements frequently reserve to the writer a personal approval right over sync uses—a songwriter may contractually be able to block her song's use in, say, a political ad or a fast-food commercial, even though she no longer "controls" the copyright. Second, recording contracts routinely contain a controlled composition clause: when the recording artist is also the songwriter, the label's contract typically grants the label a mechanical (and often sync) license to the artist's own compositions at a discount to the statutory rate—classically capped at 75% of the statutory mechanical, and limited to a set number of songs per album. This is one of the most consequential and least-understood terms in a record deal: a songwriter-artist who signs a standard controlled composition clause is agreeing, in advance, to be paid below the statutory rate on mechanicals for her own songs sold through her label. Áine, who never signed a label deal, avoids the clause entirely and collects the full statutory mechanical—another quiet advantage of independence that only becomes visible once you know the clause exists. (For the contractual drafting dimension of IP licensing generally, our guide to drafting software license agreements covers analogous negotiation dynamics in a different medium.)

Direct Licensing and the Spotify Bundling War

Here is where the post-MMA system, for all its elegance, starts to fray—and where 2024–2026 produced the defining dispute of the streaming era.

Despite the MLC's blanket system, major publishers increasingly pursue direct mechanical deals with streaming services outside the collective framework. Such deals must be disclosed to the MLC (the MMA requires it, so the collective keeps visibility into the whole landscape), but they can carry terms diverging from the statutory defaults. Direct licensing appeals to a major publisher for higher effective rates—achieved through advances, minimum guarantees, and creative structures the statutory formula doesn't permit—plus faster payment, richer usage data, and alignment with broader strategic partnerships. It appeals to a service for reduced MLC administrative fees, closer relationships with the suppliers it cannot do without, and customized terms. The result is a two-tier landscape: major publishers cut direct deals with major services, while smaller publishers and independent writers like Áine remain under the MLC blanket. Critics say this lets the majors extract value that might otherwise flow through the collective to the long tail; defenders call it the free market working as designed, with the statutory license as a safety net underneath.

There is a subtler consequence that bears directly on independent writers. Recall that the MLC's unmatched-royalty distribution runs on market share. When major publishers pull catalogs out of the blanket system into direct deals, they reshape the composition of what remains in the collective pool. The interaction is debated, but the concern independent advocates raise is clean: a system in which the largest, best-documented catalogs increasingly transact directly, while the collective is left holding the harder-to-match long tail of independent works, risks concentrating the matching difficulties in precisely the pool least equipped to absorb them. Defenders respond, fairly, that direct deals do not reduce what any properly registered independent writer is owed on her own streams—the blanket license still pays Áine for Áine's matched plays—and that the safety net stays intact for those who use it. Both can be true at once: a diligent independent is paid what her streams earn, while the structural advantages of scale, documentation, and negotiating leverage keep accruing to the majors. The takeaway for Áine is neither alarm nor complacency but vigilance—the system pays her correctly for what she registers correctly, and the burden of capturing that value sits squarely on her, because no other party in the two-tier structure has a strong incentive to chase the royalties of a writer too small to negotiate her own deal.

The Bundle That Halved the Rate

This tension detonated into open litigation, and understanding it is essential to understanding where the law stands today. Begin with the rate formula. Under the Phonorecords IV determination—the CRB ruling governing mechanical rates from 2023 through 2027—interactive streaming mechanicals are computed by an "all-in" formula. The headline rate climbs from 15.1% of a service's revenue in 2023 to 15.2% in 2024, then by half a point per year to 15.35% in 2027 (the highest U.S. streaming mechanical rate ever set), subject to alternative caps—for standalone subscriptions, the lesser of 26.2% of the service's total content costs or roughly $1.10 per subscriber—from which the performance royalties already paid to the PROs are subtracted to leave the mechanical residue. It is, in a word, a formula, and formulas have categories. And the Phonorecords IV regulations permit "bundled" multimedia services—offerings that package music streaming with other products of more-than-token value—to pay a lower mechanical rate than standalone music subscriptions, on the theory that not all of the subscription revenue is attributable to music.

In March 2024, Spotify added 15 hours of monthly audiobook access to its Premium tier and reclassified Premium as a "bundle." The reclassification let Spotify pay the lower bundled mechanical rate, cutting its U.S. mechanical payments to publishers and songwriters substantially—by the MLC's estimate, by as much as 50%. The MLC sued in the Southern District of New York in May 2024, alleging Spotify had "unilaterally and unlawfully" slashed royalties by miscategorizing a music service as a bundle.

On January 29, 2025, U.S. District Judge Analisa Torres dismissed the suit with prejudice. Reading the Phonorecords IV regulations as "unambiguous," she held that "audiobook streaming is a product or service that is distinct from music streaming and has more than token value," so Premium is "properly categorized as a Bundle." She did not find that Spotify exploited a loophole in bad faith; she found that the regulations, read naturally, permit exactly what Spotify did. (Commentators were blunter than the court—one trade outlet headlined that the judge had let Spotify "keep cutting songwriter royalties through its 'cynical trick.'") Spotify disclosed in a securities filing that, were the MLC ultimately to prevail, the additional royalties owed over a roughly fifteen-month window could reach around €256 million (about $290 million), plus possible penalties and interest.

The story did not end at dismissal—a point the input's account predated. Judge Torres granted the MLC "at least one opportunity to amend," and the MLC filed an amended complaint by the October 2, 2025 deadline, retooling its theory to align with the court's own reasoning and refocusing on whether Spotify's audiobook-inclusive product is genuinely a qualifying bundle or whether the music component still triggers the higher obligation. Spotify answered in October 2025, fact discovery proceeded into 2026 (with a March 2026 fact-discovery cutoff), and—because the amended suit faces a long road—the MLC simultaneously pressed for interlocutory review by the Second Circuit, arguing the dismissal turned on a "novel question of regulatory interpretation" that the appellate court should resolve now. As of mid-2026 the bundling question remains genuinely open, with hundreds of millions of dollars and the future architecture of streaming bundles riding on it. A parallel MLC suit against Pandora, filed in February 2024 over how Pandora computes mechanicals for its ad-supported personalized-radio tier, remains pending on the same compulsory-license framework.

What the Bundling Fight Reveals

Two consequences ripple outward. First, the major publishers did not sit on the litigation. They negotiated direct "around-the-bundle" agreements with Spotify—all three majors have inked direct publishing deals (Sony's, for instance, included a new U.S. direct-licensing arrangement) engineered to move past the CRB bundle mechanics and restore songwriter value outside the statutory formula. Independent writers and publishers, lacking that leverage, are left under the statutory framework, watching. Second, the dispute trains attention on the next rate proceeding: because Phonorecords IV runs through December 31, 2027, the successor proceeding—Phonorecords V—takes effect January 1, 2028, so the bundle rules that enabled the Spotify maneuver remain in force in the meantime, and the questions they raise will dominate that negotiation.

The deeper lesson is about the limits of the compulsory framework. The MMA solved the administrative problem—how to clear and pay mechanicals at scale—brilliantly. But it left the rate problem to the CRB, and the Spotify ruling shows that the rate a songwriter actually receives can hinge on a service's product-packaging decision that nobody contemplated when the rate was set. Judge Torres's ruling is, fundamentally, a ruling about regulatory text, and its quiet message is that the protection songwriters thought the bundle rules afforded them was narrower than they assumed. The majors' response—exit to private deals rather than litigate—proves the point: where the statutory rate can be undercut by product design, the parties with leverage leave the statutory system and negotiate around it, leaving the compulsory framework to those without leverage. For an independent writer, that is the structural inequity of the moment in one sentence: the same mechanism meant to guarantee her a rate can be reshaped by decisions made far above her head, and she has neither a seat at the bundling negotiation nor the catalog to strike a direct deal of her own.

International Licensing: The Same Lesson, Multiplied by 180

Music licensing is inherently territorial: copyright laws, collecting societies, and rights representatives all vary country by country. A song owned by a U.S. publisher is typically administered abroad by a web of sub-publishers—one in the U.K., another in Germany, others in Japan and Brazil—each collecting locally and remitting back. For a global service, this fragmentation is daunting: Spotify operates in more than 180 countries, each with its own societies, rates, and rules, and a single song may need licenses from dozens of organizations worldwide. Services cope through multi-territory deals (especially within the EU, where the 2014 Collective Rights Management Directive, 2014/26/EU, eased cross-border licensing), hub arrangements with large reciprocal-agreement societies, direct global deals with major publishers, and local compliance teams. The EU has been especially active: the CRM Directive prompted major publishers to withdraw their Anglo-American digital rights from traditional national societies in favor of pan-European licensing hubs, and the Copyright in the Digital Single Market Directive (2019/790/EU)—its much-debated Article 17—shifted platform-licensing-or-liability dynamics across Europe.

For Áine, the international layer repeats the domestic lesson, just larger and easier to forget. Streaming is global by default: the moment her music goes live on a service operating in 180 countries, her songs can be streamed anywhere, and each territory's collecting society may be sitting on royalties accruing in her name. But those societies pay only the representatives registered with them. Without a sub-publishing or administration arrangement reaching into each territory, the foreign royalties accumulate and, after a holding period, are typically redistributed locally—often, again, by market share to the largest local rights holders, exactly as the domestic black box redistributes unmatched money. An independent artist with genuine international listenership can therefore forfeit a substantial fraction of her total streaming income simply because she registered domestically and assumed that was enough. The fix is to engage a publishing-administration service with international reach early, before the foreign royalties age out, rather than discovering years later that money was collected abroad in her name and given to strangers. The territorial structure of music licensing thus echoes, at the global level, the same truth the domestic system teaches: the money exists, the law says it is hers, and whether she receives it depends entirely on whether she connected to the body that collects it.

Practical Guidance Across the System

Seeing every participant's obligations together clarifies the whole machine. A digital service like Cascade must file its MLC notice of license and report usage monthly in the prescribed format; maintain PRO licenses from ASCAP, BMI, SESAC, and GMR, each with its own reporting rules; negotiate direct master licenses with labels and aggregators (and comply with the § 114 statutory license through SoundExchange for any non-interactive component); and obtain licenses from the relevant society in each foreign territory. A music publisher must register every controlled work in the MLC database with complete, accurate splits and identifiers (unregistered or mis-registered works simply cannot be matched); decide whether direct deals make sense for its catalog; register works with the right PRO with correct splits; build sub-publishing relationships abroad; and invest in metadata quality, because bad metadata is the leading cause of lost royalties industry-wide. A record label negotiates its service deals (directly or through aggregators), runs the artist-accounting machinery that tracks streams, applies contractual splits, recoups advances, issues statements—and separately registers any publishing interest it owns with the MLC and PROs.

For an independent artist like Áine, with no institutional infrastructure, the checklist is the difference between being paid and not:

  • Choose a distributor that also handles publishing. Some distributors merely deliver recordings; others additionally register your compositions with the MLC and PROs. Know which you have, and fill any gap yourself.
  • Verify your MLC registrations directly. The MLC permits writers to register and to check, and self-checking catches the distributor errors and split conflicts that silently strand royalties.
  • Affiliate with a PRO and register every song, with accurate writer/publisher splits and identifiers.
  • Register with SoundExchange to claim the statutory 45% featured-artist share of non-interactive royalties—it is yours by law regardless of any other deal, but only if you sign up.
  • Document co-writer splits in writing before release (a "split sheet" signed in the room), because a disputed or undocumented split is unmatchable and unpayable.
  • Arrange international publishing administration once you have meaningful overseas listenership, before foreign royalties age out.

Each of these is a separate pipe through which money flows. Money in any pipe Áine fails to connect simply never arrives.

The Unresolved Fights

Beneath the machinery sit several disputes that will shape the next decade, and naming them is worth the space because they determine what a stream is actually worth.

Rate-setting is perennially contentious. Publishers argue streaming mechanicals remain too low to reflect a song's value or keep pace with inflation; services argue rates already consume too much revenue to sustain the business. Phonorecords IV produced rates neither side celebrated, and the bundling controversy it enabled has made the looming Phonorecords V proceeding (effective January 1, 2028) a focal point for everyone who believes the structure shortchanges songwriters.

The consent-decree question persists. ASCAP and BMI have chafed under their 1941 decrees for decades, arguing the decrees disadvantage songwriters relative to record labels, who face no equivalent constraint. The Department of Justice studied the decrees in 2014–2016 and again declined to terminate them; BMI's 2024 conversion to for-profit status and its sale added new pressure to the debate, but the decrees endure, and their fate would reshape the PRO–service negotiating balance.

Songwriter compensation is the most human fight of all. Despite the MLC and the billions flowing through streaming, working songwriters routinely report that their mechanical payments feel disconnected from their songs' popularity—the per-stream mechanical is a fraction of a penny, and after the split between mechanical and performance, the division among co-writers, and the deductions baked into the formula, individual checks can feel like a rounding error. That disconnect fuels ongoing advocacy for higher statutory rates, for songwriter equity in services, and for alternative compensation models, and it is the emotional backdrop against which the bundling fight landed so hard: a maneuver that lowers the mechanical rate strikes hardest at the people who already felt shortchanged.

Data and transparency remain the quiet bottleneck. The industry's data infrastructure, much improved, is still fragmented—different organizations use different identifiers and sometimes cannot reconcile their records, so even a database as good as the MLC's leaves gaps and conflicts that strand royalties. Wider adoption of the ISWC for works and the ISNI for individuals aims at interoperability, but full implementation requires coordination among hundreds of organizations worldwide—slow, unglamorous work that leaves current royalty tracking imperfect and Áine's royalties, like every independent writer's, partly hostage to it.

And looming over all of it is the newest entrant: generative AI, trained on vast troves of recorded music and now capable of producing songs and voices indistinguishable from human ones. The licensing system described in this article assumes a human writes the song and a human records it; it has no native answer for music generated by a model trained on copyrighted catalogs, nor for AI-cloned voices, and the litigation now defining those boundaries will eventually reshape music licensing as profoundly as the MMA did. For where that fight stands across the creative industries, see our analysis of copyright infringement claims against generative AI, and for the parallel question of how digital royalties travel on resale, our discussion of NFT marketplaces and secondary-sale royalties.

Conclusion: The Money Exists; The Question Is Whether It Reaches You

Music licensing in the streaming era is a web of rights, organizations, and obligations that can seem impossibly complex—but the complexity is not gratuitous. It reflects real underlying problems: compensating creators for genuinely global use, managing catalogs of a hundred million songs, and balancing the interests of multinational majors and bedroom producers under the same statute. The Music Modernization Act and the MLC fixed the most acute dysfunction—the sheer impossibility of clearing and paying mechanicals at streaming scale—and that is a real achievement, measured in the billion-plus dollars the MLC has distributed. But significant problems remain. Rate disputes persist, and the Spotify bundling fight shows how contestable the rate structure still is even after the MMA. Data gaps still cause matching failures. International licensing remains fragmented. And songwriter-compensation concerns keep driving advocacy and litigation, with the majors increasingly routing value through direct deals that an independent writer cannot replicate.

For everyone in the system, the discipline is the same: understand which rights are at stake, which organizations administer them, and what each participant must do. A service must comply with multiple overlapping licensing regimes at once. A publisher or label must register accurately and monitor its royalty flows. And an independent artist like Áine must understand how money moves through this system well enough to connect every pipe through which she is owed—because the recurring lesson of the streaming era, domestic and international, statutory and negotiated, is that being owed money and receiving it are different things. The billions flowing through streaming prove the system works in the narrow sense that music is licensed, royalties are collected, and payments are made. Whether it works well enough, fairly enough, and efficiently enough—and how the next rate proceeding, the bundling appeal, and the gathering AI fight come out—will be litigated and legislated for years to come.

For assistance with music-licensing compliance, publishing administration, royalty audits, or rights clearance, contact our intellectual property and technology practice.

Related Articles

Selected Authorities

17 U.S.C. §§ 106, 106(6), 110(5), 114, 115 (including § 115(d) blanket license and § 115(d)(9) automatic substitution); 37 C.F.R. §§ 385.10–385.11 (physical/download mechanical: 9.1¢ or 1.75¢/minute; 24¢ ringtone), §§ 385.20–385.22 (interactive streaming formula). Music Modernization Act of 2018, Pub. L. 115-264 (Title I, Musical Works Modernization Act; Title II, Classics Protection and Access Act; Title III, Allocation for Music Producers Act). Copyright Royalty Board, Determination of Royalty Rates and Terms for Making and Distributing Phonorecords (Phonorecords IV), 87 Fed. Reg. 80,394 (Dec. 30, 2022) (headline rate 15.1% of revenue (2023) escalating to 15.35% (2027); all-in cap of 26.2% of total content costs or ~$1.10/subscriber; rate period 2023–2027). Mechanical Licensing Collective v. Spotify USA Inc., No. 1:24-cv-03809 (S.D.N.Y.) (dismissed with prejudice Jan. 29, 2025; leave to amend granted; amended complaint filed by Oct. 2, 2025; discovery ongoing into 2026; interlocutory Second Circuit review sought). Mechanical Licensing Collective v. Pandora Media (S.D.N.Y., filed Feb. 2024) (pending). United States v. ASCAP and United States v. BMI antitrust consent decrees (S.D.N.Y. 1941, as amended). EU Collective Rights Management Directive (2014/26/EU); EU Copyright in the Digital Single Market Directive (2019/790/EU), art. 17.

This article is for general informational purposes only and does not constitute legal advice, nor does it create an attorney-client relationship. Music-licensing law and the rate proceedings described continue to evolve, and several of the matters discussed (including the MLC's bundling litigation and the pending Phonorecords V proceeding) remain unresolved; consult qualified entertainment and copyright counsel about your specific circumstances.