For most of my career, the most powerful person in music was whoever decided what you got to hear. A radio programmer. An MTV scheduler. A magazine editor. Later, an algorithm.
The history of dance music, of all music, is the history of that gatekeeper changing hands, and every time it changes, the music changes with it. I mapped the whole sequence, from the night MTV launched to the year AI started generating songs no one wrote, because it’s one of the clearest ways I know to understand why the music sounds the way it does in 2026.
The Pattern, in One Line
The entire history, compressed: Labels → Radio → MTV → MySpace → YouTube → Spotify → TikTok → AI. Each shift took roughly five to eight years to become dominant, and each one broke the monopoly of the last. That’s how power in music keeps getting handed to a new owner.
The Moments That Mattered Most for Dance Music
A few of these shifts landed harder on electronic music than on anything else. The first was pirate radio, which from 1991 kept jungle and drum and bass alive for nearly a decade when the mainstream wouldn’t touch them. No illegal FM transmitters on London rooftops, no UK rave culture.
Who Decides What You Hear Next
Each shift took five to eight years and broke the last monopoly
The pattern in one lineEach new gatekeeper broke the last one, and each shift took five to eight years.
- 11991Pirate radio Keeps jungle and drum and bass alive for a decade
- 22004Beatport A chart position that could set your fee
- 32007SoundCloud Hear it before it is signed
- 42008Mixcloud DJ mixes, licensed
- 52015Discover Weekly The algorithm as A&R
- 62023TikTok Add to Music Eight of ten Billboard No. 1s in 2025 went viral there first
- 72023 onAI generation Deezer receives about 90,000 AI tracks a day
New dataThe share of new uploads to Deezer that were fully AI generated, now about 90,000 tracks a day. Across all services, about 106,000 new tracks arrived every day in 2025, and 120.5 million of 253 million tracks got ten plays or fewer.
Then the music got its own infrastructure. Beatport (2004) gave the electronic industry its own economy: uncompressed WAVs sold straight from a producer’s studio to a DJ on the other side of the world, and a chart position that could set your booking fee. It was the industry’s Bloomberg terminal. SoundCloud (2007) became the underground’s default infrastructure, the grey market mixtape where you heard a track before it got signed. Mixcloud (2008) did what SoundCloud couldn’t: it licensed DJ mixes legally, and gave the set, the art form most DJs practise, a permanent, searchable home.
Then the machine started choosing. Discover Weekly (2015) created a new kind of hit, a song nobody promoted, served by a playlist that reached 40 million listeners within its first year. The algorithm became the most powerful A&R executive alive, and it doesn’t take meetings. TikTok’s Add to Music button, from 2023, turned the platform into an A&R department with a billion users. Between April 2025 and April 2026, the feature saved more than six billion tracks to streaming services, and by TikTok’s own count, eight of the ten Billboard number ones of 2025 had a viral TikTok moment before they topped the chart. The algorithm became the chart.
AI generation, from 2023 on, is the first shift where the gatekeeper makes the music as well as distributing it. Suno reached two million paid subscribers and $300M ARR in early 2026, a few months after Warner Music Group settled its lawsuit against Suno in November 2025 and signed a licensing partnership, a week after settling with Udio. Deezer now receives about 90,000 AI generated tracks a day, more than half of all new uploads, and in a blind test of 9,000 people across eight countries, 97% couldn’t reliably tell the AI tracks from the human ones. Every previous shift handed power to a new middleman. This one is auditioning to replace the artist.
What It Means for the Booth
Every shift in this timeline reduced the gatekeeping power of any single human, including the DJ. When the algorithm is the radio, what’s a DJ for? When AI is generating 90,000 tracks a day, what’s a producer for?
The more the machine handles discovery, and now creation, the more valuable the two things it can’t do become: curation and context. Anyone can be served a perfect song by a feed. Suno can write you a new one in thirty seconds. Almost nobody can build a four hour arc in a real room, reading real people, making a thousand small human decisions that no platform shift has yet automated away.
That gap is getting bigger, not smaller. More than twenty years behind the decks, and the job has never felt more irreplaceable, which is a strange thing to say in the middle of an AI gold rush.
What I Trust Now
From where I stand, social media has changed the game. A lot of people now discover music as the background to an Instagram reel or a TikTok short, just as, back in the day, people discovered it through movies, television soundtracks and OSTs. The screen changed. The habit didn’t.
On AI music, my view is simple. It should be labelled as AI, and people shouldn’t be charged for it. It should be free. Remember whose work it was trained on: the real bands, musicians and producers. Someone types “give me a deadmau5 kind of sound” and the AI can replicate it. So it shouldn’t be in stores. If people want it as background music for a film, a TV programme or their social media, great. But it shouldn’t be in clubs.
Even the collecting societies haven’t settled it. In August 2026 KOMCA (한국음악저작권협회), Korea’s main songwriting society, brought in and then withdrew, within three weeks, a rule for registering AI assisted songs after the culture ministry asked for wider discussion first.
What do I trust? I trust myself, and I trust what I see. You can tell when people are feeling a track and when they’re ignoring it. The same goes for DJs: you can see the wise ones who know how to work a room, from the opening, through peak time, to the close.
Where I Found Music Growing Up
Growing up in Canada, we had MuchMusic, and its dance show Electric Circus was fun to watch. But for me, new music came from the record shops and music stores. That’s where you heard what was coming next, and of course you dug and found things for yourself.
Mix CDs were the other great discovery tool. One good compilation could introduce you to a dozen artists you had never heard of, and send you back to the shop looking for their records.
Frequently Asked Questions
Who controls music discovery today?
Music discovery has passed from labels to radio, MTV, MySpace, YouTube, Spotify, TikTok and now AI, and each shift broke the monopoly of the one before it. Between April 2025 and April 2026, TikTok’s Add to Music feature saved more than six billion tracks to streaming services, and by TikTok’s count eight of the ten Billboard number ones of 2025 had a viral TikTok moment first, so the algorithm has in effect become the chart.
How long does each shift in music discovery take?
Each shift in music discovery took roughly five to eight years to become dominant. Every new platform broke the monopoly of the last, which is how power in music keeps changing hands. The pattern runs from labels and radio through MTV, MySpace, YouTube, Spotify and TikTok to AI.
How is AI changing music discovery?
AI generation is the first shift where the gatekeeper isn’t only controlling distribution but also making the music. Suno reached two million paid subscribers in early 2026, after Warner Music Group settled its lawsuit and signed a licensing deal with it in November 2025, and Deezer receives about 90,000 AI generated tracks every day. In a blind test of 9,000 people across eight countries, 97 percent couldn’t reliably tell the AI tracks from the human ones.
Do DJs still matter if algorithms and AI handle music discovery?
Yes, because the more the machine handles discovery and creation, the more valuable curation and context become. A feed can serve a perfect song, but almost nobody can build a four hour arc in a real room while reading real people. That gap is getting bigger rather than smaller, a theme also explored in how streaming changed the DJ economy.
Further Reading
- How Streaming Changed the DJ Economy
- Beatport vs Bandcamp
- The Press and Reading Library
- Harper’s: The Ghosts in the Machine, by Liz Pelly: the investigation into playlist filler that every DJ should read.
- Mood Machine, by Liz Pelly: the full book on how Spotify reshaped listening and who it served.
- IEEE Spectrum: The Story of Spotify’s Discover Weekly: how a hack week project became the most powerful A&R tool in music.
- DJ Mag on Traxsource labelling music as human made or AI assisted: one store putting AI labelling into practice.
- 경향신문: 음저협, AI 활용 음악 저작권 등록 결국 철회: Korean coverage of KOMCA reversing its AI registration rules in August 2026.
- RA Exchange podcast: long conversations with the DJs, labels and promoters still doing human curation.
- Wikipedia: the history of pirate radio in the United Kingdom
- Wikipedia: Beatport, the download store built for electronic music DJs
- Wikipedia: Mixcloud, the platform that licensed DJ mixes and radio shows
- Wikipedia: SoundCloud and its role as an open platform for new music
- Wikipedia: an overview of artificial intelligence in music creation and discovery
Sources
Music Business Worldwide, TikTok’s Add to Music passes six billion track saves, April 2026
TikTok Newsroom, Year in Music 2025, December 2025
Engadget, Discover Weekly reaches 40 million listeners, May 2016
Music Business Worldwide, Suno passes two million subscribers, 2026
SiliconANGLE, Warner Music settles with Suno, November 2025
Music Business Worldwide, 90,000 AI tracks a day at Deezer, 2026
Deezer Newsroom, AI music exceeds 50 percent of daily uploads, July 2026
Mixcloud Pro, licensed DJ mixes
Kyunghyang Shinmun (경향신문), KOMCA withdraws its AI registration rules, August 2026
Written from Seoul, by someone who has been in the booth for more than twenty years. More on the site, or book a DJ.
Stay in the Loop
New writing on DJ culture, electronic music, and the Seoul underground, delivered when it matters.
More From The DJ Diaries
Play with the history
Four tools built from the same research as this site.
