Jan 10, 2025 · 1h 0m · news
Mikey Shulman, CEO @Suno: The Future of Music, What is Gonna Happen? | E1244 · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of 20VC, host Harry Stebbings interviews Mikey Shulman, CEO of Suno, about the democratization of music creation, the technical and legal challenges of generative AI, and why future music experiences should be as active and interactive as video games.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 23.1% of the talking time here. How this is scored →
Harry as informed peer 5.0 Guest teaching 4.6 Guest disagreement 2.3 Harry pushing back 4.2
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 39:07 Mikey explicitly rejects host premise on prompt engineering
Mikey directly rejects Harry's prompt framing, asserting that relying on prompt guides or measuring candidate quality by prompt proficiency represents a fundamental product failure.
Hardest push from Harry ▶ 11:39 Harry interrupts to challenge customer support market sizeHarry aggressively cuts in to reject Mikey's claim that customer support is larger than music, using Zendesk's $10B valuation cap as explicit counter-evidence.
Biggest teaching moment ▶ 23:30 Mikey reframes music creation through survivor selection biasWhen Harry compares the difficulty of music creation to running, Mikey reframes the argument by demonstrating that current runners and musicians represent a heavily biased survivor population.
Harry holds his own ▶ 37:25 Harry pushes back on Mikey's product regret regarding DiscordHarry challenges Mikey's self-critique regarding delayed web development, offering a sharp strategic argument that staying on Discord preserved a 400,000-person active community.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Welcome and What is Suno | 3 | 5 | 1 | 2 | Harry asks about Suno's origin story based on insider background from Mignano and probes scaling laws in audio. Mikey educates Harry on why scaling laws operate differently in music compared to objective text benchmarks due to subjective human taste. | |
| Why Music Should Resemble a Video Game | 4 | 3 | 1 | 2 | Harry offers his own perspective as a power user and investor on model pricing cycles, arguing users won't pay extra for raw model access in five years. Mikey strongly agrees and explains Suno's architectural focus on audio tokenization rather than novel transformer design. | |
| Infinite Supply and the Value of Music | 6 | 5 | 4 | 6 | Harry challenges Mikey with supply-and-demand economic principles, arguing infinite music supply reduces overall price. Harry also pushes back on Mikey's career advice against quantum computing by defending long venture capital time horizons. | |
| Recruiting Talent: Physicists and Economists | 3 | 4 | 1 | 1 | Harry asks structured questions about talent acquisition and model alignment to subjective taste. Mikey explains how economists and physicists excel at empirical research in data-poor environments. | |
| The Threat of Big Tech Entering the App Layer | 6 | 4 | 2 | 7 | Harry directly interrupts and rejects Mikey's claim that customer support is a larger market than music, citing Zendesk's ten billion dollar buyout ceiling. Mikey maintains that automated BPOs expand the customer support TAM far beyond traditional software. | |
| Pricing Strategy and Charging from Day One | 4 | 4 | 1 | 3 | Harry offers user experience critique on Suno's annual pricing presentation being jarring versus monthly billing. Mikey explains why charging from day one was critical for collecting high-intent user data. | |
| GPU Burn, Scalability, and Team Structure | 5 | 5 | 3 | 5 | Harry confronts Mikey directly about the June 2024 RIAA copyright lawsuit. Mikey acknowledges copyrighted works in training data while criticizing the litigious, fixed-pie mindset of traditional record labels. | |
| Silicon Valley Mentality vs. Cambridge Culture | 5 | 4 | 3 | 5 | Harry probes the game theory of legal settlements and challenges Mikey on the existential risks if Suno loses in court. Mikey rejects the aggressive Silicon Valley mindset in favor of collaborative industry growth. | |
| Redefining the Creative Process for One Billion People | 6 | 5 | 4 | 6 | Harry challenges Mikey's assertion that making music isn't currently enjoyable by comparing it to the painful gratification of distance running. Mikey reframes running and traditional music creation as suffering from severe survivor selection bias. | |
| Merging AI Creation with Traditional DSPs | 5 | 4 | 1 | 3 | Harry asks strategic product questions about integrating Spotify directly and creating personalized AI models for major stars like Ariana Grande. Mikey details legal name-and-likeness constraints versus original generation capabilities. | |
| Reworking StreamShare and Music Business Models | 6 | 5 | 1 | 5 | Harry challenges the fairness of music royalties by comparing Spotify's algorithm radio payouts with prompt-based attribution. Mikey outlines alternative monetization structures beyond traditional capped StreamShare models. | |
| The Timbaland Partnership and Music Industry Cover | 4 | 5 | 1 | 4 | Harry bluntly asks if Suno paid Timbaland cash or equity for his endorsement. Mikey explains Timbaland's equity advisory role and breaks down why digital production software has made mainstream pop music structurally homogenous. | |
| The Future of Music Discovery: Spotify vs. TikTok | 6 | 4 | 1 | 4 | Harry analyzes competitive discovery dynamics between TikTok, Spotify, and YouTube. Mikey explains why Suno refuses to sell background music APIs to AI video startups to protect music value. | |
| Suno's Product Regret: Moving From Discord to Web App | 6 | 4 | 3 | 6 | Harry pushes back against Mikey's product regret regarding delayed web app development, arguing that remaining on Discord preserved 400,000 active community members. Mikey maintains that Discord was an inefficient UI for music production. | |
| The Fallacy of Prompting and the Future of AI UX | 5 | 6 | 6 | 5 | Mikey forcefully rejects Harry's premise about evaluating talent by prompt engineering skill, calling heavy prompting a UX failure. Harry challenges Mikey's stance by quoting top tech executives who view prompt engineering as a core competency. | |
| Fundraising Strategy: Capital as a Weapon and VC Alignment | 6 | 5 | 5 | 4 | Harry discusses venture alignment and capital deployment strategy. Mikey delivers a candid critique of VC pro-rata reserves as deceptive marketing, which Harry acknowledges and validates. | |
| Suno as a Social Network: Beyond the SoundCloud Model | 5 | 5 | 2 | 4 | Harry shares personal anxiety regarding the rapid pace of technological change across society. Mikey distinguishes passive acceptance of AI inevitability from active human stewardship of technological outcomes. | |
| Dystopian AI Futures: Unauthorized Clones and Hyper-Personalization | 4 | 5 | 1 | 2 | Mikey outlines two dystopian futures for AI music: jurisdictional IP theft and anti-social hyper-personalization. Harry brings a personal anecdote about generating a song for his girlfriend to test personalization boundary limits. | |
| Art vs. Craft: Taste as the New Creative Currency | 6 | 5 | 3 | 5 | Harry cites Ben Affleck on art versus craft and questions whether an AI winter is approaching. Mikey argues that curation and taste are replacing technical mechanical skill as the core currency of artistic creation. |
Statements from this episode (55)
Shulman: Empty text box UI is wrong for most AI products
“They did every AI company a huge disservice because everybody thinks that just like empty text box is now the right interface, and it is for ChatGPT, and it is incorrect for basically everything else.”
Shulman: AI companies will soon stop releasing standalone raw models
“At some point, I don't know if it's version four or version five, there will be a last model release that is released as a model, and everything else is just product releases.”
Mikey Shulman: Suno aims to make musicians, not just music
“You know, I like to think we're not making music, we're making musicians.”
Shulman: Audio AI technology lags significantly behind text AI
“Audio is really far behind the world of text.”
Shulman: Suno abandoned sense-making tool as generative audio advanced unexpectedly fast
“Our backgrounds are all NLP, and we thought it would actually be a lot harder to do good generative stuff, and so we thought the first product would be more sense-making. You know, try to scratch your head and think back to, like, GPT-II. No one was really mak…”
Shulman: Scaling laws will not solve AI music generation
“For music, It's very different from text, and I think people will very sloppily look at the world of OpenAI and Anthropic and the hyperscalers and say audio is just a couple years behind, which it is, but that scale is gonna solve all these things. But unlike …”
Shulman: Interactive music will drive consumer spending like video games
“I think if you accomplish all those things and if you make music interactive and you make music engaging people will pay for it like they pay for video games.”
Stebbings: Consumers won't pay extra for newer AI models in five years
“I don't, I tweeted this yesterday actually because of this, sorry but I don't think in the future we will pay more for access to newer models. I think that is just a stage of the development cycle that we're in. In five years time we won't pay more for a bette…”
Shulman: Suno's edge is audio representation, not model architecture
“Our competitive edge here has not been to innovate on the architecture, and has been to innovate on the, Audio representation.”
Shulman: Infinite AI supply lowers individual track value but increases music's overall worth
“I think it reduces the average value of any given piece of content, but I think it greatly increases the value of music to society or to any given person.”
Shulman: Quantum computing startups face fundamental physics challenges, not engineering ones
“The challenges are not they're still physics challenges. They're still like basic research challenges that have to happen here, and so it's not something where you can kind of just turn the crank and make companies out of it, and to my knowledge, all of the co…”
Shulman: VCs will not get timely returns on quantum computing investments
“I don't think you're gonna get your money back in a reasonable time.”
Mikey Shulman: Suno pays employees less than OpenAI does
“Let's be honest, I will pay you less than OpenAI will pay you, and I need to find a reason to convince you to come join us.”
Mikey Shulman: The team that runs high-quality AI experiments fastest will win
“Experimental physicists just get good at running high-quality experiments really quickly, and AI is an empirical discipline, and so whoever can run more high-quality experiments quickly will win.”
Shulman: Suno Is the Only Place Aligning AI to Human Taste
“If you are interested in a different AI problem, in the problem of how do I align Models, not to objective truth, but to human taste. I think there's nowhere else to do it.”
Stebbings: Zendesk Is a $10B Company Overbid by Private Equity
“Come on, Zendesk is like a ten billion dollar company at best, overbid by PE providers.”
Shulman: Users Choose Suno for Product Quality, Not Model Size
“At the end of the day people will use Suno because it's a better product, not because there's a bigger, better, bigger model under the hood.”
Shulman: Suno Will Not Reach Sub-Second Generation Speed Anytime Soon
“And so I don't see us getting to one second anytime soon, but I think the crossover that you're talking about, we're still kind of far away from.”
Shulman: Suno's primary user success metric is hitting the day-one paywall
“The most salient one is actually just like, did you hit the paywall your first day? Because even if you didn't go through it for whatever reason, if you hit the paywall, you just enjoyed the last 10 or 12 minutes of your life, and I know I did something good t…”
Shulman: Suno chose to charge users from day one rather than stay free
“We made a decision to start charging from day one, which is like a little bit against the traditional wisdom of Silicon Valley of just like give the product away for free, scale, scale, scale.”
Shulman: Suno's subscription revenues offset a lot of GPU burn
“Even, even without the revenues, which are fantastic to have, and they offset, you know, they offset a lot of GPU burn knowing what gets people to subscribe is such valuable data that I don't know, like, how we would do this without it.”
Shulman: Suno's GPU compute spend is several times its entire payroll
“It's the biggest thing by far. Yeah. It's a few times payroll.”
Shulman: Scaling AI research headcount yields sublinear returns
“The machine learning team at Suno will almost certainly not scale as quickly as the software team, or said differently, I think research is not something that you can solve with scale. You can just throw more bodies and you'll get, you know, comparatively more…”
Shulman: Using copyrighted works in AI training data is industry standard
“Yeah, we know that there are some copyrighted works in our training data. That's not illegal. It's stock standard for the industry. It's what every AI company does.”
Suno CEO argues the music industry suffers from a fixed-pie mindset
“An embedded fixed pie mentality. There's a fixed pie of money out there and we are all just, you know, trying to divide it unfairly for ourselves. And if we were focused on growing that pie, I think like just everything gets easier.”
Shulman: Suno actively works with plenty of music industry incumbents
“And by the way, there are plenty of incumbents in the industry who do work with us.”
Shulman: Music incumbents suing AI startups before negotiating is inefficient
“On paper, I would say it just seems silly to throw a bunch of venture dollars at lawyers instead of sitting down and talking about how you could Work together. You know, like, deciding to sue first and then ask questions later seems to me to be inefficient, bu…”
Shulman: Founders who claim getting sued is a good thing are lying
“Those people, I think, are lying if they tell you, like this is a good thing. You know, it's like, maybe it's a sign of a good thing, but it itself is not a good thing. If nothing else, it's gonna cost money and time.”
Shulman: Suno will survive even if it loses copyright lawsuit
“If this lawsuit goes to trial and we lose, it's obviously not good for us. The company, the company's not dead, but it's obviously not good for us. I don't think that's likely.”
Shulman: Most people do not enjoy the time spent making music
“It's not really enjoyable to make music now. It takes a lot of time. It takes a lot of practice. You need to get really good at an instrument or really good at a piece of production software. I think the majority of people don't enjoy the majority of the time …”
Mikey Shulman: Major AI companies are hitting data walls
“I think the bigger players Are hitting data walls and are more actively working on it, and like everything else, it's really nice to have the big players solve a lot of the really thorny problems for you, and you can kind of focus on your competitive advantage…”
Shulman: Integrating AI Music Creation with Streaming Creates a Larger Market
“That just shows you that there is a lot of friction, and if there were two worlds of music, I think that would A, be friction for the average consumer, but I actually hope it's much more than just consumption. I hope that, like, there shouldn't be, like, a set…”
Shulman: Most artists privately admit to using and loving Suno
“The vast, vast majority of artists that I speak to Behind closed doors admit that they use and love Suno.”
Shulman: Suno blocks users from generating music with real artist names
“You've noticed in our product, you can't make Ariana Grande songs. If you put her name in, we're going to wave our finger at you, and you're going to say, like, that's not what the future of music is. That's not original music. Suno is for original music. It's…”
Shulman: AI music fan fiction is more engaging than artist AMAs
“What I can only describe as the equivalent of fan fiction, if she wants that, and that would be an immensely engaging activity, in my mind, way more valuable and way more engaging than like having an AMA with her, if you're a huge fan.”
Shulman: Suno users pay to create music, not to listen to it
“The thing that people don't realize is that right now we've made something where people enjoy making music so much they are paying to create the music. This has nothing to do with listening.”
Suno does not pay Timbaland in cash for partnership
“No, we don't pay him in cash. He's, yeah, he's an advisor to the company.”
Mikey Shulman: Modern pop is shorter and more homogenous due to TikTok
“I think songs have gotten Shorter, yes, but also much more homogenous in melodies, harmonies song structures, and they, and music innovates very much sonically. Like, they're really interesting sounds, and they all fit into the same song structure, and this is…”
Shulman: Hit songs succeed via algorithms and promotion, not quality alone
“I think that they probably reverse the causality. Things break on TikTok, and then they become famous on Spotify, but that is a function of both algorithms and people pushing music, and not necessarily only a function of the quality of the music.”
Shulman: YouTube is More Engaging Than Spotify Based on Ad ARPU
“Like YouTube is more engaging than Spotify as measured by the amount of, you know, the ARPU for ads, for example, right?”
Shulman: Suno refuses API requests from generative AI video companies
“The vast majority of Gen AI video companies have asked us for an API so that they can have, you know, Suno music behind their AI videos, and the answer's always no.”
Shulman: 90% of Suno traffic migrated from Discord to web in five days
“And so we released this app, we released this web app, which is not even the full functionality of the Discord bot in November, and it takes five days for 90% of the traffic to move over to the web. Five days.”
Mikey Shulman: Great AI software UI must enable novel workflows
“The UI should just in no way resemble what it does today. But I think good UI is just like, what did you enable that is like completely not doable in the old pattern?”
Shulman: The term 'prompt' will disappear within six to twelve months
“I hope that, you know, in, in six months from now, or 12 months from now, we're not using the word prompt. There are just far more intuitive ways to interact with the music, and we shouldn't be guiding you we should be listening to you if that makes sense.”
Shulman: Requiring complex prompts is an AI product failure, not user failure
“Well, if it took a really complicated prompt and 1600 iterations, that's a failure of my product, not a failure of the candidate.”
Shulman: Suno was built in-person, only allowing remote work exceptions
“We never, you know, we built the company in person, and we make exceptions for people to be remote.”
Shulman: Judgment and taste are underrated compared to skills
“I think judgment, you can call it taste, you can call it judgment, is like so underrated on how important it is. It's like compared to people's skills.”
Shulman: VC promises of reserve capital are actually not founder-friendly
“So that's, you know, that, that's billed as very founder friendly and it's actually the opposite.”
Shulman: Suno prioritizes user growth twice as much as revenue
“If you, like, you know, hold my feet to the fire, maybe I can give you a quantitative number of, I care about growth twice as much as I care about revenues.”
Shulman: SoundCloud's growth is limited by the difficulty of traditional music creation
“Where SoundCloud goes wrong, I mean, there's two things, like, one is by no fault of SoundCloud, it's like, it's still hard to make music and so, so you're kind of limited there, and it's still kind of, there's the creators, and there's the consumers, and they…”
Shulman: Foreign actors will build worse AI music if US fails to lead
“We can actively build a bigger, better future of music with AI, or we can just wait, and someone else in another country, not bound by US laws, not with the same intentions will build a worse future of music with AI.”
Mikey Shulman: Creating AI songs for loved ones is more social than streaming
“Like, that is not something, that is so much more social than what exists now, where you listen to music from artists, and you love the music, and you see a piece of yourself in the artist, but you don't actually interact with them.”
Mikey Shulman: Human taste will outweigh technical skill in future art
“It's increasingly taste is the only thing that matters in, in, in art, and skill is going to matter a lot less because you're going to be able to make a lot of stuff And the people who are going to be recognized are people who are able to pick from the vast qu…”
Shulman: ChatGPT acts as a median-competent tutor for every student
“The first order effect here is that GPT means that every person in the world has, like, a median competent tutor or sidekick. This is obviously amazing for education, and if you can't recognize that fact, I'm not sure you have such good judgment.”
Shulman: Suno topped generative AI corporate card spending on Ramp
“We were at the top of what people put, the gen AI products that people put on their ramp credit cards, which means that like lots of, and so we were surprised people are using this for Cold outreach for sales, for marketing, for all kinds of stuff.”