A techno DJ just raised $1 million to build the 'artists get paid' AI music platform. The idea is older than the money.

Updated:

Techno producer Reinier Zonneveld has raised a $1 million pre-seed round for GasLab — "Generative AI Solutions Lab" — a generative AI music platform that, per details shared with Digital Music News, is built to "recognize and pay musicians for their work," trained on fully licensed data, with attribution back to the original musicians on every generated song (per DMN). The round includes Eric Demuth, co-founder and executive chairman of Bitpanda, and Bas Beerens, founder of WeTransfer.

The pitch lands at a moment when the music-AI sector has split into two camps: platforms that settled with the majors and folded rightsholders into the product, and everything else. GasLab is proposing a third path — not just licensed training data, but a per-output revenue split traceable to the specific musicians whose work shaped each generation.

What's actually on offer

The core mechanism is what GasLab calls the Digital Notary: a record of which contributing artists' music shaped each generated output, which then serves as the basis for sharing subscription and release revenue back to those contributors. Zonneveld's framing is blunt: "The current state of generative AI has a major flaw: it takes from artists but gives nothing back. This is a problem that has to be solved if we want there to be any semblance of a creative industry in the future."

The founder is not a tourist. Classically trained from age three, self-taught in electronic production, founder of the Filth on Acid label in 2017, holder of the world record for the longest electronic music live set, and top-30 in the DJ rankings for five years running, Zonneveld has been working with machine learning for eight years with his business partner Cas Luuring. That work produced R2, an AI model trained on his own catalog that he plays back-to-back against in live sets — a human and a model trained on the human, taking turns. The third co-founder, Thijs Emondts, is a former VC partner and Uber general manager.

The company is pre-launch: its site currently reads "Opening soon. Generative music, by artists, for artists," with a guest-list signup for private testing invites. The public debut comes with a physical-world flourish — at Amsterdam Dance Event, where Zonneveld speaks October 22-23 and plays an eight-hour set on October 25, a poster campaign across the city will let visitors generate a track from a prompt and watch the participating artists get attributed to it.

Why the timing is awkward — and pointed

The interesting thing about GasLab's raise is what it's raising against. A week before this announcement, the RIAA's new Notorious Markets filing put AI voice-clone platforms on the official piracy-infrastructure list. A year of Character.ai's bankruptcy has demonstrated that AI unit economics are brutal even before anyone gets paid. And the biggest platforms have chosen consolidation over attribution: Suno's v6 shipped "in partnership with the music industry," developed with Warner Music Group, BMG and Believe — a licensing-and-structure answer to the consent problem, not a per-output payment one.

GasLab's bet is that attribution is the product, not the cost. If the Digital Notary works — if every generation genuinely credits and pays the humans whose music shaped it — the company says the engine "could also serve other genAI platforms," which is the more ambitious half of the plan: selling the provenance layer to competitors who don't want to build one.

That ambition collides with three hard realities.

First, the money is small. One million dollars is a pre-seed; Suno's partnerships, Spotify's audiobooks empire and the majors' litigation budgets operate on different orders of magnitude. GasLab's edge, if it has one, is that it doesn't need to win the generation-quality race — it needs to win the trust race, which is cheaper.

Second, the attribution claim is unverifiable for now. The platform is pre-launch, the training corpus and the Digital Notary's mechanics are described only by the company, and the funding details come through a single outlet (DMN, which notes the figures were "shared with" it). Until independent parties can generate tracks and audit where the money goes, "fully licensed data with complete attribution" is a promise, not a feature. The ADE poster campaign will be the first public test — an attribution demo the company itself controls.

Third, per-output attribution at scale is a genuinely hard technical problem, and the industry's biggest players have conspicuously not attempted it. When Spotify wanted AI voices for audiobooks, it shipped a tool built with ElevenLabs — consent-based, but with no revenue-share machinery per output. The majors' settlements with the AI platforms bought catalog access, not per-song micropayments to working musicians. If attribution-at-scale were easy, someone with more than $1 million would probably already be doing it.

The counter-case for optimism

The steelman for GasLab is that "someone should" has finally become "someone with standing is." The loudest critics of AI's take-from-artists economics have been rightsholders and outside commentators; Zonneveld is an inside-the-scene producer with a festival-scale audience, a label, and eight years of his own ML work — someone whose peers will judge the product on whether it actually pays them. An attribution engine born inside the techno scene, demoed at ADE, carries a credibility that a corporate launch cannot.

And the sector has a hole exactly this shape. The authorized platforms have answered "may we train on your music?" but not "does an individual musician see anything when the model generates?" If GasLab's Digital Notary survives contact with real releases, the answer switches from "no, but the label got a settlement" to "yes, and here's the ledger." That is a genuine innovation worth a million-dollar bet — and worth more skepticism than a million dollars buys.

Sources