The numbers hit like a flash crash. 90,000 AI-generated tracks uploaded to Deezer every day. Not total. Per day. That is roughly one song every second. For context, the entire Spotify catalog in 2023 was around 100 million tracks. At this rate, Deezer would ingest that volume in just over three years. But the real story is not the volume—it is the signal it sends about content authenticity, creator economics, and the structural collapse of trust in digital assets.
I first understood this problem in late 2017 while auditing early ERC-20 implementations. The same signature replay vulnerability that could drain funds across chains is now being mirrored in content duplication. A single AI model can generate infinite variations of a song, each with a different hash, yet identical in spirit to copyrighted works. The code is law, but only if you can verify the code. With AI music, you cannot. The ledger tells you nothing about provenance.
History repeats, but the signature changes. The spam wars of the early internet—comment sections flooded by bots, ad networks gamed by fake impressions—are now playing out on streaming platforms. The difference is scale. A bot can generate 10,000 comments in a day. An AI music model can generate 90,000 distinct audio files in the same period, each unique enough to pass basic duplicate detection. The signature of the attack is no longer text or numbers; it is frequency, timbre, and arrangement.
Let us quantify this. Assume the average AI-generated track is three minutes long and encoded at 320 kbps. Each track requires roughly 7 MB of storage. Ninety thousand tracks per day equals 630 GB of new content daily. That is 230 terabytes per year. For a single platform. This is not a storage problem—bandwidth is cheap. It is a curation problem. The cost of hosting and delivering this content is negligible compared to the cost of verifying its legitimacy. And verification, as Deezer is discovering, has no scalable solution.
Context: The Current State of Content Governance
Deezer is not alone. Spotify, Apple Music, and Tidal all face the same flood. But Deezer’s public disclosure is a calculated move. As a smaller player—market share around 2% globally—they are using this data to position themselves as the responsible steward. “We are the ones who count,” they are signaling to regulators and artists. Meanwhile, the three major labels (Universal, Sony, Warner) are watching silently, waiting for the first lawsuit to set precedent. The legal framework remains trapped in 1976 copyright logic, where “authorship” implies human creativity. AI challenges that distinction.
The technology powering this flood is mature. Models like AudioCraft, MusicLM, Suno, and Udio have been public for over a year. They can generate instrumental tracks, vocals, even mimic specific artist styles. The barrier to entry is zero. A user types a prompt, and the model outputs a WAV file. The user uploads it to a distributor, which submits to streaming platforms. No human creation involved. The 90,000 tracks are not all high-quality. Many are noise. But some are good enough to earn royalties, diluting the pool for human artists.
From a trader’s perspective, this is a classic market manipulation scenario. Fake volume. Wash trading. The platform becomes a venue for counterfeit assets, and the spread between genuine and synthetic collapses. In crypto, we call this “impermanent loss” when liquidity pools are imbalanced. Here, the liquidity of listener attention is being drained by infinite supply. Impermanent is a promise, not a guarantee. The promise was that streaming would democratize music. The guarantee is now broken.
Core: Order Flow Analysis—Quantifying the Damage
Let me apply the same forensic methodology I used after the Terra Luna collapse. I reverse-engineered the UST mechanism using on-chain data. Here, I will use published music industry data to calculate the financial impact.
Assume the average per-stream payout on Deezer is $0.004. (Deezer pays slightly higher than Spotify for some markets, but let us use a conservative estimate.) If each of those 90,000 AI-generated tracks garners just 1,000 streams per month—a reasonable baseline for a promoted track—that is 90 million streams per month from AI content. At $0.004 per stream, that is $360,000 of monthly royalty outflow to AI-generated music, presuming the royalties go to whoever registered the track. That is $4.32 million per year. For a platform like Deezer, whose annual revenue is around $500 million (pre-2024), this is less than 1% of revenue. But the damage is not the direct cost; it is the opportunity cost. Those 90 million streams could have gone to human artists, who would then have a reason to stay on the platform.
Now consider the multiplier effect. If the 90,000 tracks are just the tip of the iceberg—if the detection algorithm catches only a fraction—the real number could be 900,000 daily. At that scale, the revenue leakage becomes material. And the detection algorithm arms race begins. Deezer is investing in AI to detect AI. This is recursion. The same technology used to generate content is deployed to verify it. The winner is the one with better models, not better intentions.
Pattern recognition precedes profit realization. In trading, I look for divergence between price and volume. Here, the divergence is between platform growth (more tracks, more users) and content quality (declining listener satisfaction). The smart money is already rotating out of pure streaming plays into platforms that offer verifiable provenance. That is why Crypto Briefing is covering this story. The narrative is not about Deezer. It is about the infrastructure gap.
Contrarian: The Retail Mind Trap vs. Smart Money Flow
Retail narrative: “AI music is inevitable. Just accept it and enjoy the creativity.”
Smart money reality: The real value is not in generating new music. It is in verifying the origin of existing music. The market is flooded with counterfeit goods. The biggest opportunity is the ledger that tracks ownership from creation to consumption. This is where blockchain enters the stage.
Consider the contrarian angle: AI music is actually accelerating the adoption of digital rights management on immutable ledgers. When you cannot trust the file, you must trust the chain of custody. A song uploaded with a verifiable signature—signed by a private key linked to a human identity or a registered artist—becomes a provably authentic asset. Everything else is noise. The infrastructure of content verification (NFTs for music, blockchain-based royalty registries, smart contracts for automated licensing) suddenly has a burning use case.
Verify the code, trust the ledger. This is not just a slogan. It is the only logical defense against the infinite supply of synthetic content. If a streaming platform requires all uploads to be accompanied by an on-chain proof of human authorship (e.g., a zero-knowledge proof that the creator solved a CAPTCHA-like challenge without revealing identity), the 90,000 daily count would collapse to a fraction. The cost of generating the proof becomes the real barrier. And that proof can be economically efficient—one hash per track, stored on a cheap L2.
But here is the catch: the platforms have no incentive to implement this unless they are forced. Deezer’s 90,000 figure is a confession that they cannot stop it. They are passing the hot potato to regulators and tech vendors. The smart money is betting on two things: first, that regulation will mandate content provenance within 18 months (EU AI Act, US Copyright Office), and second, that the infrastructure to provide that provenance will be built on decentralized networks because centralized databases are too easy to manipulate.
Silence before the volatility spike. Do not mistake the current regulatory calm for inaction. The publishing of this data is a signal that the spike is coming. When the first class-action lawsuit against a streaming platform for distributing unlicensed AI-generated replicas of a famous artist’s voice hits the docket, the market will reprice every music stock. The token projects that already have working provenance solutions (e.g., Audius, Royal, even some NFT music platforms) will see a capital inflow.
Takeaway: Actionable Signals for the Battle Trader
So where does this leave us? The 90,000 daily upload figure is a macro indicator. It tells us that content inflation is accelerating exponentially. The platforms that survive will be those that integrate cryptographic verification. The protocols that profit will be those that provide the rails.
Watch for three signals:
- Deposits into music-related DeFi protocols. If Audius’s AUDIO token sees a spike in TVL, it signals that capital is anticipating a need for decentralized content storage.
- Partnerships between streaming platforms and identity solutions. Deezer or Spotify announcing a collaboration with a blockchain-based identity provider (e.g., ENS or a KYC oracle) would be a catalyst.
- Legal rulings on AI consent. A court decision requiring explicit consent for training data would crush the current generation of AI music models, but vault the provenance rails.
Logic survives the emotional wash. Right now, the emotional response is fear and anger. The logical response is to identify the infrastructure gap and fill it. I am not saying buy tokens. I am saying watch the ledger. The market whispers, the blockchain shouts.
The final question is rhetorical: If you cannot tell the difference between a human-composed symphony and a machine-generated loop, does the distinction matter? Only if the royalty check depends on it. And it does. That is why ownership must be verifiable. The 90,000 tracks are not a crisis. They are a clarion call for a new layer of digital truth.