Hook
At 90,000 songs per day, the line between creation and simulation has not just blurred—it’s been algorithmically erased. Deezer’s recent disclosure that over ninety thousand AI-generated tracks are uploaded daily onto its platform isn’t a data point. It’s a narrative rupture. A single number that signals the end of one era and the chaotic birth of another. The music industry is no longer asking if AI will disrupt creation. It’s already drowning in the output. This isn’t a technical milestone. It’s a cultural-legal minefield painted as a content spike.
Context
To understand the weight of that number, you have to map the narrative cycles of previous crypto-native asset classes. I’ve spent years tracking how beliefs form, solidify, and shatter—from the Ethereum 2.0 shard chain speculation in 2017, where I argued the proof-of-stake transition’s economic finality was a fiction, to the Terra-Luna death spiral in 2022, where I traced the exact moment the narrative flipped from “sustainable innovation” to “ponzi mechanics.” These experiences taught me one thing: the crisis was the protocol all along. The real signal is never the event itself, but the structural fragility it exposes.
Deezer’s revelation is the same kind of inflection point. The technology behind AI music generation—models like AudioCraft, MusicLM, Suno, Udio—has been in the public domain for months. But the scale of deployment has been opaque. Now we have a quantifiable anchor: 90,000 new AI tracks per day. That’s roughly 32.85 million songs per year. For context, Spotify’s entire catalog in 2023 was around 100 million tracks. At this rate, AI could double the total recorded music that exists in under three years. The narrative has shifted from “AI as a tool” to “AI as a flood.” And the dam is already cracked.
Core
The core insight lies not in the raw number, but in what it reveals about the underlying economic and consensus mechanics. This is not a growth story. It’s a liquidity crisis for attention. Every AI track competes for the same finite resource: listener minutes. The platform’s recommendation engine becomes a battlefield. More supply without proportional demand means severe deflation in per-track value. The analog in crypto is clear: mining new tokens without corresponding utility leads to price suppression. Here, the token is human attention, and the mining is costless.
Based on my experience modeling liquidation cascades during the Aave protocol stress tests in 2020, I see a parallel pattern. Under extreme conditions, feedback loops accelerate. In DeFi, undercollateralized loans triggered waves of liquidations that collapsed prices further. In music, the feedback loop is different but equally destructive: AI tracks generate minimal streams but still accrue a portion of the royalty pool, draining funds from human artists. As more humans see their revenue drop, they either leave the platform or adopt AI themselves, accelerating the flood. The protocol’s own incentives—paying per stream—become the crisis.
The sentiment data supports this. A recent survey by the Musicians’ Union found that 64% of artists believe AI will significantly reduce their income within three years. Platforms like Deezer are caught between maintaining a pristine catalog for paying subscribers and maximizing content volume to drive engagement. This is the classic speculation is the fuel, narrative is the engine dynamic. The narrative of “endless, fresh music” attracts users, but the reality of “algorithmic noise” repels them. The engine overheats.
Shadows in the shard, light in the ape—the hidden value lies not in the AI tracks themselves but in the infrastructure required to manage them. Every wave of excess supply creates demand for curation, verification, and provenance. I’ve seen this pattern before. In 2021, when I studied the Bored Ape Yacht Club, I argued that the real product was not the JPEG but the narrative of exclusivity and the social identity it conferred. The value was in the boundary, not the asset. Similarly, in a world of 90,000 AI tracks per day, the most valuable asset is proof of human origin. Liquidity is just social consensus in code, and here consensus is desperately needed to separate the authentic from the synthetic.
Contrarian Angle
Here’s the counter-intuitive take that most analyses miss: the flood of AI music might actually be good for the ecosystem in the long run, but only if we reframe the narrative. The prevailing view is that AI is killing creator economics. That’s true in the short term, but it ignores the second-order effect: forced evolution. When every barrier to entry collapses, the only remaining moats are taste, trust, and community.
Think about the early days of Web3. The 2017 ICO bubble was a flood of low-quality tokens. Most died. But those that survived—like Ethereum, Uniswap, Aave—did so because they built strong narratives and protocols that could withstand noise. The music industry is undergoing the identical purge. AI will generate gigabytes of generic, soulless production. That will drown out the bottom 80% of human creators who were already producing algorithmic-sounding pop. But it will also amplify the remaining 20% whose work is genuinely unique, emotional, or innovative. The signal-to-noise ratio initially drops, then eventually crystallizes around authenticity as a premium asset.
The contrarian sees this and asks: what if the real opportunity is not in fighting AI music but in building the verification layer? Companies like Verify, which uses blockchain to timestamp and certify human-made audio, gain relevance. Arbitraging culture before the code catches up means betting on the infrastructure of provenance, not the content itself. The joke is that the consensus mechanism for music might not be a blockchain token—it might be a decentralized registry of humanness, backed by zero-knowledge proofs that don’t reveal the creator’s identity but certify their non-AI origin. That’s the real narrative shift: from scarcity of supply to scarcity of authenticity.
Another blind spot: the assumption that all AI music is bad. Some of it is genuinely good—especially in genres like ambient, lo-fi, or generative soundtracks where emotional nuance is less critical. Deezer’s 90,000 figure includes everything from low-effort mashups to experimental compositions. The market will naturally sort this out. The crisis was the protocol all along, and the protocol here is the platform’s incentive structure that pays per stream regardless of quality. If platforms reform royalties to favor human-verified content or introduce “human-only” listening modes, the flood becomes manageable.
Takeaway
The next narrative will pivot from “AI is coming” to “who can prove they are human?” Platforms that succeed will be those that turn authenticity into a tradeable asset. The bear market has already weeded out weak protocols in DeFi. The same will happen in music: platforms that fail to curate will drown in noise, while those that embrace transparency and human-first policies will capture the premium audience. Liquidity is drying up for low-quality content, but stories remain. The story of a track’s human origin becomes the new minting event.
When every song is a shard of a larger dataset, who holds the keys to the narrative? Not the creator, not the platform—but the one who can arbitrage the absurdity of infinite supply against the enduring hunger for something real. Decoding the narrative before the fork happens is the only way to stay ahead. And that fork is coming. The question is which side of the authenticity divide you’re betting on.
