How we could design Kanye's "Y Combinator for Music"
And more importantly, how not to.
This week, Tech Twitter was abuzz with Kanye West’s thoughts about record contracts—including his idea to create a Y Combinator for musicians.
I have been thinking about this problem for a long time, so today we’ll go through what it would take to build such a thing.
Subscribe to Click Track via email.
What is Y Combinator?
Y Combinator (YC) is a startup accelerator. Startup founders apply for $125,000 in funding and a three-month long mentorship program in the San Francisco Bay Area, ending in a Demo Day where startups pitch investors. So far, they have funded over 2,000 companies, and a small handful of them are now worth billions of dollars each.
Kanye West apparently got the idea of creating a YC for musicians in a conversation with the venture capitalist Katie Jacobs Stanton—who sits on the Board of Directors of Vivendi, the French conglomerate that owns Universal Music Group.
YC is known for being particularly founder-friendly, and for having created simplified fundraising documents like the Simple Agreement for Future Equity (SAFE) that have become an industry standard for early-stage startup fundraising. Culturally, YC’s business dealings are a stark contrast to the legally-complicated ways of major labels and publishers.
Are the economics of a music accelerator feasible?
The Y Combinator model is successful because it only takes a few home run investments to pay for all of the strikeouts. YC’s most valuable portfolio company is the $36 billion payments company Stripe. When the wins are that big, it almost doesn’t matter how many $125,000 investments that YC makes into startups that will fail.
But if you and I are starting an accelerator for musicians, we have a big problem. No individual musician is ever going to have a $36 billion catalog. The Beatles are the best-selling musical act of all time, and their songwriting catalog has been rumored to be worth around $1 billion. In fact, the entirety of Universal Music Group—the world’s largest music company—has been valued around $33 billion.
The most lucrative investments for a music accelerator are likely to be 1/100 or 1/1,000 the value of a startup accelerator’s best investments, and there are no obvious solutions for decreasing our losses by a similar proportion. Ultimately, this why major labels, publishers, and investment funds like Hipgnosis prefer investing in the reliable cash flow of already-proven catalogs.
A musician accelerator could try increasing the value of their portfolio by negotiating for a greater share of (successful) portfolio artists’ royalties. But the endless media circus around bad record deals has made artists wary of any deal that doesn’t have a sky-high valuation. Small deals feel cheap to many early-career artists, even if there is no other music business offering them better terms.
There is only one way left to make the economics work: to be sure that a high percentage of artists will at least break-even with the investment.
How would we design a Y Combinator For Musicians?
Landing a traditional recording or publishing deal typically requires commercial success, industry connections, or both. But if we’re starting an accelerator, we are presumably reviewing thousands of applications from artists that don’t have commercial success or connections yet. We have to come up with an admission process that can quickly and accurately filter through all of them.
If we are concerned about losing money on commercially unsuccessful artists, the obvious solution is to only accept artists who already have profitable catalogs. Various companies like Stem, TuneCore, The Music Fund are offering advances to artists based on their catalog’s historical royalty data.
But the margins on these financial products are low compared to what a successful accelerator should earn, and if we truly want to build an accelerator as opposed to a bank, then we have to figure out how to reliably bet on artists before they have made any money. Similarly, many of Y Combinator’s best investments had insignificant or no revenue when they first applied.
In order for us to start a musician accelerator that can scale investments without being overwhelmed with losses, we’ll have to consider some crazy ways to vet potential artists:
For one, we could try applying machine learning directly on the audio of artists’ catalogs or demo tapes, looking for songs that could become high earners if correctly marketed. Hit song science is the application of machine learning models to predict which undiscovered songs will become popular. This field has hardly been solved, but we don’t need our model’s predictions to be perfect—they just need to point our accelerator’s admissions team in the general direction towards high-potential artists such that our successful investments cover our losses.
Alternatively, there may be ways to make the application process interactive to get a better sense of which artists to invest in. For example, Pioneer is a cross between a startup accelerator and an online multiplayer video game. Contestants (called “Pioneers”) work on ambitious startups, earn virtual points, and rank each other’s progress. The rankings are then used to generate a global leaderboard. The top Pioneers can level up to being offered larger and larger investments in their company.
What could an accelerator offer musicians?
Y Combinator is far from a passive investor. Their office hours with partners, dinners, the Demo Day, and various other programs do an incredible job of increasing the value of each batch of startups. Companies often seek help from Y Combinator partners or alumni even years after their Demo Day.
If you and I are creating an accelerator for musicians, ideally we offer services that increase the financial value of our catalog without being too labor- or capital-intensive—such as playlist pitching, brand deals, or sync licensing placements. But the downside for our hypothetical accelerator is that we’d be offering these services in a much more crowded market than Y Combinator faced when they started in 2005, nor may our services be sufficient to boost the trajectory of an early-career artist.
Who is already running accelerators in the music industry?
There are a handful of coworking spaces that boldly market themselves as accelerators for musicians. But since they are in the business of charging rent as opposed to investing in artists, they aren’t real accelerators.
That said, there are some real ones out there:
TechStars Music and gener8tor’s gBETA Musictech are accelerators, but for music-tech startups—not musicians.
gener8tor also runs two accelerators for musicians—Backline and Motown Music Accelerator. Both programs are twelve weeks long and offer a $20,000 grant without asking for any royalties or revenue share in exchange.
Zoo Labs in Oakland, California is a nonprofit accelerator for musicians. They run a four-month program for musical groups to develop their business skills.
Given that many of these accelerators are either nonprofits and/or don’t actually invest in musicians, it seems that nobody running a musician accelerator has figured out deal terms that are acceptable to the musicians and profitable for the accelerator.
Which leads me to ask…
Are there better models for artists than accelerators?
As we have seen, it is difficult to map the economics of a startup accelerator to the music industry. It might be better to serve artists by redesigning the record label, music distributor, or the talent agency.
Here are a handful of companies trying to capitalize on early-stage artists in other ways:
Indify is a marketplace between emerging artists and potential investors. Founded in 2016, they originally pitched a proprietary algorithm that could theoretically predict the future commercial success of an early-career artist based on popularity metrics for artists collected from social networks and streaming sites.
On Tuesday, Indify CEO Keshav Garg replied to Kanye with a video to explain how they’re fulfilling Kanye’s vision for a Y Combinator for musicians.
Similar to Indify’s original pitch, Instrumental is a platform that uses data science to identify quickly-rising artists. They sell a software platform designed to help record labels and brands scout for talent. Instrumental also runs frtyfve, an in-house record label and music publisher.
Conrad Withey, who was previously President of Warner Music Group, founded Instrumental in 2015 as a rebrand and pivot from his wife Abi Hanna’s startup PopShack—which was a network of YouTube channels and an artist development label for YouTube talent.
UnitedMasters is a company that distributes independent artists’ music to streaming sites like Spotify. They use this data to help get brand deals for their top artists, or for entire playlists of songs—like they did with Apple Music, ESPN, and the National Basketball Association. I recently wrote about their deal to offer distribution to artists on TikTok.
Much like UnitedMasters, Amuse offers music distribution to streaming sites like Spotify. The catch is that Amuse offers this service to artists for free to collect data about their commercial success. If an artist on their platform begins to find significant commercial success, Amuse may try to sign them to their in-house record label. Amuse also offers a program called Fast Forward, which offers artists advances based on machine learning models to predict their revenue.
Snafu Records is a record label that claims to find unsigned artists using artificial intelligence. They brand themselves as a label, but they’re signing the sort of early-career artists that would join an accelerator for musicians. Snafu claims to look at 150,000 tracks every week, looking for music that is roughly 70% similar to Spotify’s Top 200 chart.
Heavy Sound Labs
Heavy Sound Labs is music executive Jason Geter’s attempt at reinventing the record label. Geter is best known as the founder of the Atlanta-based label and management firm Grand Hustle, where he worked with T.I., B.o.B, Travis Scott, and others. Heavy Sound Labs itself is part of the Los Angeles-based startup studio Science Inc.
Well, if you happen to be Kanye West, reply to this email and we’ll start an accelerator.
But ultimately, the most important part of building a successful for-profit musician accelerator is to find a way to invest in a large number of musicians without requiring them to already have lucrative catalogs. Otherwise, it is hard for the winning bets to outsize the losing bets while still being differentiated from later-stage labels and publishing companies.
September 19, 2020: Added the section about Heavy Sound Labs.
January 11, 2021: Minor formatting changes.
January 31, 2022: Minor formatting changes to support new features in the Substack editor.
February 13, 2022: Minor formatting changes.
Technically, he said “‘Y Combinator’ for the music industry” but I’ll take a leap and assume that he meant an accelerator for musicians rather than music businesses.
The COVID-19 pandemic has pushed Y Combinator’s main program to be fully online. Y Combinator also runs a free, fully-online class called Startup School.
Of course, Y may invest additional money via pro rata or YC Continuity, but this doesn’t change the fundamental win/loss calculation.