What broke after 1999 was not the recording. It was the bundle of pressing, distribution and retail that had made a label necessary.
Photo: Daniil Komov / Pexels
Record labels: the rooms, the people, the catalogues.
The Napster years didn't kill recorded music. They killed the infrastructure that made labels powerful.
What broke after 1999 was not the recording. It was the bundle of pressing, distribution and retail that had made a label necessary.
Photo: Daniil Komov / Pexels
A record label in 1998 was not primarily a talent company. It was a logistics company with taste attached. The bundle it controlled — master recording, pressing plant contract, distribution network, retail relationship — had been assembled across fifty years and was, by the late 1990s, nearly impossible to replicate from the outside. When Atlantic signed an artist, the value it offered was not only an advance or a producer. It was physical access to consumers: a manufactured object, shipped to a chain buyer, stocked on a shelf. The music was inside the bundle. The bundle was the business.
Napster, launched in June 1999, did not attack the music. It attacked the bundle directly, separating the song from every physical and commercial structure that had been required to move it since the pressing plant era. A teenager with a dial-up connection could now acquire a track without touching a distributor, a retailer, a lacquer or a stamper. The label's five-decade infrastructure became, in the course of roughly eighteen months, optional.
| 1999: US recorded music revenues approximately $14.6 billion (RIAA, retail value) | the peak of the CD era |
| 2003: approximately $11.8 billion | the year iTunes launched, after three to four years of peer-to-peer erosion |
| 2007: approximately $10 billion | decline continuing well into the legal download era |
| 2018 | streaming revenues overtook physical and download revenues combined (IFPI global data) |
The RIAA's historical shipment figures make the slope visible and unambiguous. American recorded music revenues peaked in 1999 at approximately $14.6 billion (at retail value), a figure that represented the combined ceiling of the CD era's long bull run. By 2003 that number had dropped to around $11.8 billion. By 2007 it had fallen further still, to roughly $10 billion. The decade-long slide that followed was not a correction. It was structural.
What is significant about the RIAA data is not only the size of the decline but its timing relative to format. CD unit shipments, the metric that had underpinned the catalogue reissue boom of the late 1980s and the blockbuster era of the 1990s, began contracting before legal downloading had matured into a commercially meaningful replacement. The file moved faster than any licensed alternative could be built. Apple's iTunes Store launched in April 2003; by then, three to four years of uncompensated file-sharing had already restructured listener habit, price expectation and — most damagingly — the perceived value of a recorded track.
The shop counter as market research: what sold on Saturday decided what got pressed on Monday.
Photo: Ian / Pexels
An LP sleeve under a lamp. The cover was the only part of a record a buyer could examine before paying for it.
The insight the Napster moment forced into view was that the label had never really sold music. It had sold convenience, scarcity and packaging. Convenience, because a manufactured single or LP was easier to obtain than any alternative. Scarcity, because duplication at scale required industrial equipment. Packaging, because the sleeve — whether Reid Miles's typographic compositions for Blue Note or Peter Saville's Factory sleeves — gave the object a reason to exist beyond the audio it contained.
Digital files eroded all three simultaneously. Peer-to-peer networks were more convenient than a trip to a record shop. Duplication was now costless and perfect. And a compressed audio file carried no sleeve, no liner note, no catalogue number, no physical weight. The bundle that had concentrated power in a small number of major companies — by 1998, the market was dominated by the five majors that would later consolidate into three — simply ceased to be the only path between a recording and an ear.
The labels' response was initially legal rather than architectural. The Recording Industry Association of America pursued litigation against individual file-sharers from 2003 onward, a strategy that generated significant press coverage and almost no commercial recovery. The structural problem was not criminality. It was that the infrastructure the labels owned had stopped being necessary.
What survived the collapse of the bundle was the master recording itself — the underlying intellectual property rather than any physical or logistical expression of it. Labels that had spent decades accumulating back catalogues found that streaming, once it scaled in the 2010s, monetised those catalogues in a new and durable way: not through scarcity but through licensed access, with IFPI global figures showing streaming revenues overtaking physical and download revenues combined by 2018. The business rebuilt around the asset that the file had never been able to copy: ownership.