How Suno Grew to $300M in ARR by Democratizing Music Creation

On Suno, this is all about having somebody fall in love with the process of making music. And so, having the thing like really understand you is akin to I'm an artist and there's one producer that I always work with because this person gets me. A lot of the problem with like, "Hey Phil, go make a song." is the blank canvas problems. Like, "What do you mean? What should I make a song about?" When somebody shares a song with you, you have a built-in piece of inspiration and you can riff off of it. The top of the pyramid isn't even entertainment. You can get entertainment anywhere. The top of the pyramid is fulfillment. When you feel the warm and fuzzy feelings of being creative, of seeing the fruits of your labor, of having other people recognize the fruits of your labor, these are feelings that you can't really get anywhere else. Welcome to Subversive, a podcast dedicated to sharing stories

from the best consumer subscription apps in the world. We'll bring you lessons for how to grow your consumer subscription business, including insights and inflection points that led to exponential growth from leaders at category-defining companies and innovative startups. Let's get into the show. This episode is brought to you by RevenueCat. The world's best apps use RevenueCat to power purchases, manage customer data, and grow revenue on iOS, Android, and the web. Join more than 100,000 apps that are already using RevenueCat to grow their businesses and track more than $10 billion in spend, including category leaders like OpenAI, Notion, and Ladder. Learn more at revenuecat.com/subversive.

My guest today is Mikey Shulman, the co-founder and CEO of Suno, an AI music generation platform that has grown to 2 million paying subscribers and 300 million in ARR since launching in December 2023. In this episode, I talked to Mikey about Suno's unusual origin story, about the key inflection points that have turned it into one of the fastest-growing consumer subscription apps in the world, and about how Suno is navigating the complex legal and creative dynamics of building an AI platform at the intersection of music and technology.

All right, welcome Mikey to the Subversive podcast. Really excited to have you on today to talk about Suno. This is a very fast-growing AI company in the music space that I've gotten to know well over the last year and excited to hear more about the story today. Before we jump into it, why don't we

just hear a little bit more about your background and what inspired you to found Suno in the first place? >> Sure, yeah. It's great to be here. I'm I'm excited to chat. I guess my background's a little weird. I'm not the the average consumer founder, but um uh grew up in New York, played a lot of music as a kid, played in a lot of bands in high school and in college, but um I'm only okay. And so I left music. I went to grad school. I got a PhD in physics and then I found myself at a different tech company called Kensho, uh which is machine learning for financial services.

That's where I worked most closely with my co-founders. The revisionist history of how why we started the company is that we used to have jam sessions in my co-founder's basement to blow off steam after work. And that's true. That's not why we started the company. Um we actually thought it would be too hard to do what we do when we started the company. We started actually a little bit before the ChatGPT moment. We thought that these technologies for a while would only really be good at really making sense of music. And we started to build toward that and we quickly had some breakthroughs that made us realize, "Oh my goodness, we can make music." And it started out not great music and little short little clips.

And we realized that we just enjoyed using this so much and we would stay up late making music with computers even though it wasn't great. At a certain point we realized like, "What are we doing? We're we're we're lying to ourselves. We are doing our day jobs and then we're doing this other thing at night that's way more fun." And maybe everybody wants to have this much fun. And so we we shut off the old thing. We quickly released a bot on Discord that lets you ask for music to get made and um we haven't looked back since. Well, I love the reference to Discord because I actually just had Francesco Polizzi, who is one of their former product leaders, on the podcast. That was my last episode. So, funny little tie-in there.

But, let's talk a little bit more about what you were originally building, and it sounds like the primary thing that inspired you to transition to Suno and to your passion around building technology for music was twofold. One was you all were just very intrinsically motivated about music and and were passionate about it. But, then two, you found that the technology was working better than you expected. And so, this opportunity felt like it was more attainable than maybe you thought it was when you first got started.

>> Absolutely. The timing was like I I think really fortunate here. If they're rewind like a little bit before the ChatGPT moment, which which nobody remembers anymore. Um but, there was a thing called GPT before ChatGPT that people were using to make sense out of text. And we had realized that um there's so much audio out there. And audio is really a bit of a backwater in the world of AI. And we love it. Um and we thought like for no good reason is is this like a second-class citizen. And so, um there was definitely some inspiration there.

The The fortunate timing here is really that I think let's say 6 months earlier there wasn't enough open-source stuff for us to use from the text community. Um and 6 months later, you know, competition started to creep in. And so, um we really at the right moment were able to leverage a lot of stuff from what was happening in text and really focus our energy on um the things that are different and particular to audio and really get a lot of help from everything happening in in text. Um and then being able to bring a product to market um and launch it and delight people and learn from that before other people could.

>> Yeah, I mean two big lessons there that are sort of truisms at this point. But, one is build for your own use case, and it sounds like you just had the right group of people to do that who are all very passionate about music. And then two is timing is everything. And you guys hit the wave at the perfect time. Before we jump into the Suno story,

how would you say your background in physics and machine learning and then this lifelong interest that you've had in has shaped your approach as the founder and CEO of the company. >> You know, I think in a few ways. Um much of the spark for the company, at least for me, is just reflecting on some of the fondest memories I have, some of the most fun I've ever had is making music with my friends. Not even performing, you know, like I played in a lot of bands, but actually just practice was so fun. And you get very close to people making music with them. That is something I think that it is innate in people and that everybody enjoys being creative with their loved ones. And so it's something that we can bring to everybody and delight them.

Although we have to build a lot of technology to do it. Um on the physics side, like in some sense, I'm not all that special. Like there's a lot of ex-physicists who went into machine learning and and AI. Um one of them is my co-founder and he's better physicist than I was. Um I think um and I think I I actually just realized pretty quickly that where a lot of my, let's say, competitive advantage is is um on the boundaries between things.

Like I was okay at physics. Um but I did something that was like some physics, some quantum information, some low-temperature microwave engineering, and realizing that getting okay at all of those pretty quickly um was uh an opportunity. And I think in many ways my job now is like that, which is really understanding the boundary of what is possible, really getting partially up to speed in a lot of things at once and being able to make kind of the same decisions. Um and then I think the last thing I would say is both physics and machine learning, at least for me, were really team sports and um really learning to not just work well with other people, but actually I love working with other people. And um that's a huge driver for, let's say, my personal satisfaction.

>> Yeah, well, one thing you said there that definitely resonates is I I feel like many of the greatest breakthroughs in human history have come when two fields have sort of collided together. And I feel like AI is making it more and more possible for somebody to get up to speed on a field and be okay at it faster than they used to be able to do. Uh and so hopefully that will lead to more breakthroughs. I think that's a good transition to the early days of Suno. So, what did the first version of

the product look like? And what were you all trying to improve with the early prototypes you were you were building and the early adopters of those prototypes? >> The first version of the product was a really crude Discord bot, very much inspired by Midjourney. It wasn't nearly as full-featured and you would just ask for the song that you wanted. You would describe the song in like a sentence and out it would pop. At the beginning, we could do I think 12 and 1/2 seconds of audio at a time, which is not a full song. The audio quality was very poor.

Um you know, this is this has been described to me as maybe it's like organized sound and not exactly music. It wasn't I think for us at the very beginning like is there something to prove? It was like we really enjoy this. We think a lot of people are going to enjoy it. Let's let them enjoy it. So, Discord provided an amazing way to um get a product out quickly. Like you don't have to build a UI, you don't have to build auth, etc. etc. Um it also provided a great way for us to get feedback. You know, you have your users um using the product in the same place where they can tell you about the product, in the same place where they can tell other people about the product, etc. Um and at the beginning that's just invaluable for really understanding what are people doing, why are they doing it?

The other counterintuitive thing we did there is actually started charging from the beginning. Business school 101 would tell you try to get really big before you start to monetize. Um I have no counterfactual whether or not this was right. It definitely afforded us some advantages. Most importantly, trying to figure out um for who is this really interesting and that they're willing to pay and for who is it just a novelty item. And at the beginning, um we could literally go and talk to every single person who hit the paywall and didn't go through it and understand why didn't you go through it. And that is like invaluable information that we otherwise just wouldn't have been able to get. Um and then it turns out in the first month a lot of people signed up and so we were definitely encouraged.

>> Yeah, that's really valuable obviously to get early traction cuz it just means not only does it build confidence that you're onto something, but it accelerates the continuous learning loop for you and your co-founder. So, what was some of the early feedback like that you were getting from these people who were not only trying the product, but in your case paying for it since you monetize right away? >> Yeah, it's a great question. Like really varied. You know, there there were plenty of the obvious ones which is like you your music is bad. Um and you know, we would get more information about in which way it was bad. Even at the very beginning an element of like it's not just it's it's it's not just about the utility. I'm not just doing it to there were people who were doing it just to get the sounds out that they wanted to go do something else with it.

But like it was actually enjoyable. It was fun for people to interact with a bot that was helping them make music. And like in many video games, I mean Discord is a a large part of the gaming ecosystem. Um people would sit there for hours playing video games. People were sitting there for hours playing music and feeling amazing about themselves. And again, it was um really confirmatory of what we had seen just playing with it ourselves in our homes of like we enjoy doing this. We even know how to make music with other means, but we still enjoy doing this and so perhaps everybody will.

>> Yeah, well what a better alternative for uh those extra hours relative to video games to be to be building music. I I feel like it's a much more rewarding way to use your time. Um so, let's fast forward a little bit to early 2023, you

released Bark, which was this open source text audio model. And that was before Suno as it exists today existed. So, what was your logic behind releasing that open source research first and then what did you learn from the developers community's reaction? >> It's a great question. That was like really the beginning of the breakthrough for realizing how to make passable um audio. And to just go back, um there were available text-to-speech systems out there. So, Bark was something where I say, "I want a voice that's going to say" and then you would say some some big sentence or a paragraph or three words or whatever it is. And um at the time, technology has progressed a lot since then, but at the time, everything was really robotic. It came out sounding right, but it was like really robotic.

It wouldn't make a lot of mistakes. It didn't exactly sound real. Um sometimes it would sound close to real, but it was never like really lifelike. It was never really emotional. And the technologies that we had been developing um Bark made mistakes. It would get words wrong sometimes, but it sounded really good. And um what we saw is that it's not just about the same old metrics that people care about of like what is the word error rate, how quickly could you actually make the audio. Um there's a different thing, which is like in many ways subjective, which is like how good is it? You know, and and um I think that what we saw is people picked this up, and yes, it was also open source, which is great. People picked this up and said like there are lots of new things to do with this.

Um and we suspected as much. Um and at the time, Bark really would not have been terribly commercially viable. It cost a lot to run. It was like really inefficient. There were definitely better technologies, and I think at the time also everybody's brain was more in the like um just read me the damn text on my screen and don't care about um exactly how good it sounds. And not thinking about the more artistic applications for it. And so uh again, we were like really heartened how quick how how quickly it was picked up. Um it got a lot of GitHub stars very quickly. Um there was like a developer community around it. Um and again, it was like the spark for us of like there's something really beautiful about audio. It is messy. It is so much more subjective, so much more gnarly, and so much more beautiful than text is.

>> Well, one thing I'll intervene with here is as a podcast host, I've gotten to talk to a few founders of companies now working specifically on AI for speech applications. So, uh I talked to 11 Labs who is obviously big been big in text to speech and then have talked to Tunai at WhisperFlow that is in a related field and then we've talked to Speak about their application of AI models to help with more immersive language learning applications. And what's interesting is all three of those companies like Suno are ultimately relying on some sort of AI model to produce audio outputs but for very different reasons.

In Speak's case it's about language learning and so it's very important to have the right inflections and to be able to coach you on your pronunciation. In 11 Labs' case it's more about being able to use a variety of different voices for different applications and then in your case it sounds like the the sort of quality of the tone of the voice and making it more sonorous or being able to support it more for an art form like music was really important but those are very different vectors of innovation.

So, what I hear you saying is there was a lot of innovation being done around sort of the accuracy of the text to speech outputs but not as much being done on making sure that the quality of the voice itself lended itself to good music and and so that presented a real opportunity for Suno. Am I hearing that correctly? >> Absolutely. And so the opportunity to us was very obvious in music just given our backgrounds but also those opportunities existed in speech then too and I actually a lot of people told us like go build a speech company. Like it's so much more straightforward than music and and there are so many applications and I think there's a perception that speech is a much bigger market than than music but even within speech I remember like Bart could whisper.

Yeah, and and you could you could have it do things like "Hey, this is a really cool podcast." And it would do that thing to you that um I I don't know if you remember there were these commercials for Five Gum and they would like whisper and and and um no speech system could do that. And again like how big a market is that? Maybe not it's not terribly big, but it's beautiful. It's interesting and it gets people thinking and it gets people creative and so just thinking about things in terms of the old way of doing things and there were text to speech systems that that predated us like you you know Google had an API that would make robotic text to speech and just thinking about things in terms of like how do I do the same application faster?

I think often misses some of the beauty that comes with like doing something that was only literally possible literally today. >> I totally agree.

And so that brings us to December 2023 when you all launched the Suno web app. What happened in that moment and what was the response like and how did you know that you were really on to something at that point? >> You know, I think about this moment a lot because it's something that I just got totally wrong. So we were on Discord for for a little while and I thought we would not be on Discord for a really long time. Like it seemed to be working really really well for Midjourney.

Happily I have smart co-founders who are like, "No, you're being dumb." And we released a not even full featured web app. It didn't have the entire functionality of the Discord bot. And it took literally 5 days for 90% of the traffic to move over to the web. And it's because it should be easy and intuitive and you shouldn't need to go recover your Discord password just to use this beautiful music making machine. It was also around Thanksgiving and I think people are home a lot and it's a great time around family and friends and that makes people want to make music and so it was a pretty big moment for us.

We definitely had the we definitely had our infrastructure crash a few times during during that time which is like both good and bad. It was just every time a moment like this happens we get more um convinced and convicted that everybody loves making music. It's it's not it's not this niche use case like music is a part of everybody's life and this is something that we should lean into very very hard. >> Yeah, you you said in your blog post recently about the future of music that it's just one of those things that has universally binded humans together almost since the beginning of our history as a species.

And so, it's really exciting to be thinking about how AI will continue to evolve that medium over time. I think we're going to talk a little bit more about that later in the podcast. But for now, let's talk a little bit more tactically around the some of the specific growth mechanics that have helped Suno scale so quickly. Um so, I want to start with you're a product that

has grown largely through word of mouth. And so, what is it about Suno that makes people want to share what they create? And And where do they often share their content? >> I think it's a couple of things. Like one is it's it's actually just delightful. And um when you make something that you love, you're going to fall in love with that piece of content. It I think it's like kind of universally true. And you will get more um of those warm and fuzzy feelings. You will feel more fulfilled and gratified when you show it to somebody and they like it, too. And you know, I think there's like a very small version of this. Let's say if you find a great song on Spotify and you love it. And you send that song to someone that you think has great musical taste, and they say, "Hey, that was a great song." You feel amazing. You feel like a tastemaker. And when you've made that song yourself and shared it with them, those feelings are

amplified by, you know, a factor of 10 or a 100. And so, a ton of the um word of mouth is actually just people sharing songs with one another. And it happens in so many different media. It happens over text message or WhatsApp. It happens posting it on social media. Um it happens um you know, downloading it and putting it to a slideshow of uh family pictures. There's just like so many new use cases of how you're going to interact with another person through music. And so, um it's actually been amazing to not be prescriptive with exactly how people should share it. Like let's get them to arrive at the piece of content that they love and have enjoyed the journey there. And um the rest will kind of take care of itself. And so, every time we make um the journey to get to that piece of content easier, better, more enjoyable, um we just see the sharing behaviors explode.

>> Well, and I think that's one of the things that's really powerful about creator platforms and particularly creator platforms that have an atomic unit of content, which in your case is a piece of music, which could take a lot of different forms. But in your case, the use cases are so horizontal that as you said, you've everything from professional musicians who are wanting to share their content to get feedback on it, to you know, like a casual grandmother who wants to put in in a family album and and have music playing with their family. Um

so it sounds like you've had a lot of different channels that have helped to drive virality and word of mouth, but there are are there any particular ones if we drill down a level that have been most important for for accelerating the growth of the company in the early days? And then have you done anything in particular to sort of lean into that momentum and optimize that channel in particular to drive more growth? >> The biggest one is making it obvious.

Sin- Since most um most users are coming in when somebody shares a song with them, is making it obvious from a song page that you two can create. And here there is a I mean here there's some balance and some art of you come to that song page, we definitely don't want to make you log in to listen to the song. Um we want to make that as friction-free as possible, but we also want to make you understand that actually that song was made on this platform, you two are creative, you two can make music. And so there's there's some I don't think we we have figured that out by any means, but that is when we do make improvements there, um we certainly see a payoff in a big way. The ceiling there should be really high.

Like everybody loves music. And so like my intuitively I would think that when a new person who has never seen Suna before gets a song shared with them, um it should be like probably 90% of the time we should be able to get them to make a song. And just as a like a little tidbit here, a lot of the problem with like, "Hey Phil, go make a song." is um the blank canvas problem. It's like, "What do you mean? What should I make a song about?" When somebody shares a song with you, you have a built-in piece of inspiration. And you can riff off of it.

And this was like music was really meant for. It was meant to be adapted and modified and covered and sampled from and all of the things that you now know to do with music. And so, um actually starting with a song and being able to adapt it to your own um needs is the most intuitive way to do that. And so, you'll see us actually work on that a lot in like the next year. >> Yeah, I love that. That's such an interesting insight that music is inherently a medium that lends itself to riffing and remixing. I mean, it's part of why TikTok, which was previously musically, was so successful in the early days cuz you would have people sort of remix each other's music videos. And so, that

blends into the next question I want to ask, which is nearly 50% of first-time Suno users hit the free tier limit. And so, that's a very strong signal of the value you're creating. But, it relates to the point you just made, which is they already have this built-in inspiration if they're coming in from somebody who shared a song with them. So, how has Suno gotten to such high activation rates? And then, how did you and your team decide what you wanted to make free on Suno versus what you're putting behind the paywall?

>> It's It's a great question. I You know, the unsatisfying answer is I actually don't think we've given this terribly much thought. Our paywall um height, the free plan, has been the same actually since the day we launched. You can You can hit create five times a day. Um and I don't think um it's obvious that the thing that we wanted to do uh 3 years ago is the same thing that we wanted to do now. What I will say from like a product building perspective, we have a principle here of like in some sense you're saying we can trade off revenue for usage by raising and lowering the paywall. And the the principle here is like we are so early and um there's so much in front of us that growth should never be trading one thing off for another. And so, um that is to say the way we will increase revenue and usage is by making the product better. And the way we will trade them off from one another is

basically by bilking people into paying. And it's very obvious where all of our efforts should be. It should be in the former. We should be making things way, way better. And so, we have some rough heuristics that I think are again, they're not the most thought out, but it's like, "We'll release a new model. We'll only give that new model to the to the paid tier." >> Mhm. >> And then, we'll sometime later upgrade the free model when we kind of figure out how to make that a little cheaper to serve, etc., etc. Um but I don't think I don't think this is right forever, and I'm sure we'll change it at some point.

>> Yeah, that makes sense. I mean, the arc you typically see companies go through. Well, I I guess I'd divide into two categories. So, I often talk about chess versus checkers, right? You have the companies who they're indie developers, they're solopreneurs, they're building a product that is targeting a smaller TAM, and they can build a great little 1, 5, 10 million-dollar ARR business, but it's never going to be a massive venture-backed outcome. And for those companies, it's much more important to dial in and optimize onboarding flows and paywalls and pricing and packaging right away because you do need to extract enough value to make the business sustainable. In your case, I think what I hear you saying is the TAM is so large and the opportunity is so massive, it's such a blue ocean opportunity given what AI has unlocked for innovation in music, that wasting time on paywall optimization or figuring

exactly out which feature should be before after the paywall, it's it's just not the bottleneck. The bottleneck right now is just building the best possible product because that that is a tide that will carry all boats. And so, that's where most of your efforts are going. >> It's It's actually even more than that. It's not just that it's not the right use of our resources, which it's not. Um it'll also just be wrong too quickly.

Like, the product velocity here, like, the product changes so quickly, the capabilities change so quickly. Um AI is changing not only our product, but actually how we build products. That any optimizations we might make to the free tier, um are going to be stale in like two months. And so, I would hate to be uh as a user, I would hate to be using a product where exactly what you're giving me is changing month to month. And so, for now, it's actually just better to leave it alone and just make the whole thing better, even if yes, we are leaving a little bit of money on the table, or we are being somewhat non-optimal.

>> This episode is brought to you by Paddle. Today's fastest growing apps are multi-channel. They use web monetization to reach new audiences, improve attribution, and boost margins. But selling on the web gets complicated fast. Managing currencies, taxes, fraud, and disputes gets increasingly difficult as companies scale. Paddle solves this. As the merchant of record for thousands of apps, including Runna, GoodNotes, CodeWay, and HubX, they take on the complexity of web monetization, so you can focus on growing your business. Learn more at paddle.com/subversive.

>> Yeah, that is a great point. And to some degree, it's unprecedented. I mean, I think you could probably make the same argument 10 or 20 years ago about the fastest growing tech companies, but AI has accelerated the pace of innovation so rapidly that you now have a much broader swath of companies that I think fall in the same boat, where it's just like, we shouldn't waste any time thinking about the paywall or feature optimization in terms of monetization, because it's going to be irrelevant in 3 months. So, just keep innovating on the core product. Having said that, let's talk a little bit more about you

recently hit 2 million paid subscribers, 300 million in ARR. I think you said in a blog post that close to 100 million people have made music at least once on Suno over the last 2 years. What are some of the specific levers that have driven subscriber conversion and retention for the business, and where do you see opportunities to drive additional growth, I guess above and beyond the obvious that you already stated, which is just continuing to build better models and make the core product experience better?

>> Here, I'm going to try not to give you an unsatisfying answer, but when you when you look at our growth curves, either users or revenue or subscribers, all of the inflections come from making the product better. It's cuz that's most of what we're doing. But let let me let me try to unpack that a little more. Like, the cartoon you can have in your head, and that I have in my head, is that effectively everybody in the world is creative, and effectively everybody can fall in love with making music as a new form of entertainment. And it is basically our job to guide you down that path, to show you that you can have a great time making music. And once we do that, you will you will retain. And so your first one or two sessions are obviously the most important ones to help to figure that out. And then if you you know, if you get through those, if you have that magical moment, you will retain, you'll probably subscribe.

So that's basically the cartoon I have in my head is the probability that that happens is basically a product of did I make a song that I felt like was really mine, that I made and was that song good? And so very roughly, the first is like is the product experience really good? Is the experience of crafting that song really creative and enjoyable? Is it more than just typing six words into a text box? And then the latter problem is was the song good is like largely a model problem. And we often release products and models together. Very often a new product is enabled by a new model. And so every time we do that, we actually increase the both of those. And so the product of those two numbers increases a lot. And so that is what inflects people. That is what inflects growth curves. That's what gets people to hang around.

And that's what kind of gives the tailwinds to the business. As an example, I said, you know, if you just type six words into a text box, it doesn't really feel like yours even if the song came out okay, it'll feel gimmicky. That's kind of the the the meme use case. But if for example, you let people sing into their microphones, like you start to feel like ownership. A really important thing to release, this is like ages ago, is I made that song by typing six words into a text box.

Very quickly what we realized is people look at that and they say, oh, that's okay. I can do better. And they click and they edit lyrics. And now they're starting to write lyrics of their own. Once that happens, we we You're starting to get hooked. You're starting to feel the joys of being creative and iterating on your work and doing it over and over again. So it's it's stuff like that um that ultimately the way I think of this is like ultimately that is the first step towards spending, you know, an hour making music in our kind of uh pro tool product, which is called Studio.

>> Well, I appreciate you giving me a more nuanced answer and what I hear there, I think it's really important, which is the core of everything is you just have this incredible core product experience that's getting better and better with every model release, but the biggest bottleneck is some people don't actually make it to the aha moment of truly understanding everything they can do with Suno and the magic of what they can get out of it. And so, to the extent that you're investing efforts at all outside of just building the next model and building the next set of features, there is effort that's going into making sure first-time users or new users are using the product enough and making enough new content that they really get to that tipping point. And so, I imagine there are a number of things you can do um to invest in that area, but let's go to

this point around how so much of your inflection and your >> subscriber growth has come from just releasing new models. You said you've you've gone through many models, model quality is a huge growth lever, it's getting better and better with each release. You just recently released V5.5, which had a number of new features associated with it. Um how do model improvements or how have they impacted Suno's key business metrics uh over the last couple years?

>> They've been the biggest drivers. There's three things we care about in the model. There's uh um sound quality, does the audio sound like really crisply recorded? There's song quality, um is it a catchy tune that I get stuck in my head that I want to sing over and over again? And then control, it's like how do I actually interact with the model? Um I think you can argue with me that there's some ceiling to sound quality. There's like some human perception ceiling cuz of our our anatomy of our ears.

The other two I think we'll work on forever. Like songs can always get better. Um the way you interact with it can always get better. Um and so, uh as a result, what ends up happening is that models um the model releases and and the products that they enable are the biggest inflection points um for us. And so uh and basically nothing else ends up mattering. Obviously, tons and tons of effort goes into a new model release, not just technical, not just on the machine learning side, but um these end up being really big moments for us.

>> Yeah, let's jump into it. So, you just

really released 5.5 I think over the last month or so. It's your most powerful model yet and it included several new features. So, voices, custom models, my taste. Um what do each of those features do and how do they fit into your longer-term vision for how Suno's product evolves over the next few years? >> Voices is the ability to use your own voice in a song. It is one of the most, if not the most, requested feature by our users.

Um uh took a little while to get right. You have to pull that out of the model in in a really compelling way. But basically, you can think I can sing for about 15 seconds. Maybe I'm too embarrassed to do that in front of you, so I'll go into a room by myself and I'll sing for 15 seconds. And then I can use my voice in my regular Suno process and just say use this voice, but here are the lyrics I want you to use, here's what I want the song to sound like, etc. etc. Obviously, going back to what I said before, having people feel like the song was theirs is a huge part of the joy and satisfaction that you get from making music. When you hear yourself in the song, you can't help but feel like that is your song.

And then even more so, when I send you a song and you recognize me in it. Um that is a really powerful moment. And so, um this is a really big feature for us. It's like a re you know, a really really exciting one. I think we'll we'll keep developing that a lot. In a similar vein, the other two are really about how do I tailor the creation experience to sound like the things that I like. And so, custom models and my taste are both going in that direction. My taste is all about um how do I kind of give the model, if you want to think about it as memory, the same way the LLMs have memory. How do I tell it like all the things that I like, so I don't have to keep telling you over and over again. Um and so, you can think it's just like describe me all of the musical taste that you like. The amazing thing is that it will adapt and it will kind of learn from your behaviors, what you do and don't like, but then you can go and edit

it. And so, it is a an amazing template or canvas for um describing what you like. It is also as people go through their musical journeys and learn more and more how to describe music um the way they like to describe music, but they get deeper into how they like to describe music, it's um you develop an amazing attachment to it because you have learned more about music through that process. And so, um that one's really exciting.

And then the last one is custom models, which is like let's say I have a whole bunch of music of my own that I would like to inspire all of my creations for today. I can actually just upload that and the model knows how to look at that and contextualize it. And you can think this is like um the the million token context in an LLM. It's like here's 100 pages of documents. Given that, do all of these other tasks. Um this is very similar.

>> Got it. And to the extent you can share, how do those features fit into your longer-term product vision or roadmap? And and how do you see it evolving the way that your customers are interacting with Suno over time? >> When things feel personal, you fall in love with them. And it's so much more than just personalizing the content that we all see. Um today people are kind of buried by personalized content from algorithms.

You don't really fall in love with that because the intention there is to really just keep you on platform, whatever platform it happens to be. And so, you don't really fall in love with that content. If you think about the platforms today, you actually fall in love with creators. You fall in love with uh people typically. On Suno, this is all about having somebody fall in love with the process of making music. And so, having the thing like really understand you is akin to I'm an artist and there's one producer that I always work with because this person gets me.

And um how do we give the same feeling to everybody even because everybody can't have a personal producer who helps them make music? >> Yeah, I love that. And it's so refreshing in an era where we're having machines do more and more of the things that we used to value and take pride in as as human beings. This feels like it's almost moving in the other direction. It's using technology to create stuff that feels extremely personal and and that causes people to fall in love, and I imagine that has to be part of why you all have resonated so much with the market and grown so quickly as a result. So, as as a final topic, let's

talk about how AI is reshaping music and how Suno as a market leader is driving that change, but also navigating pushback from industry incumbents, which is nothing new in the music industry. And and maybe to start with, so Suno is obviously democratizing music creation. It's making it cheaper and more accessible for everyday people to make great music, but at the same time, it may end up displacing musicians and songwriters on the margins. So, how do you think about that tension, and what's your vision for how AI and creativity among human beings can coexist in the music industry over the long run?

>> It's a great question. Look, I I have a lot to say here. You know, I think just one thing is that empirically it's actually not replacing anybody. And so, this is certainly a good start. I feel like coexist is actually a really low bar, and it kind of on its face assumes that there's a fixed pie, and we're all fighting over this fixed pie. And if so a lot of new people are participating, then a lot of other people must be doing less well. I actually just think that's wrong.

It's like we're not living in a fixed pie mentality. We're not living in a fixed pie universe here. We don't need to have a fixed pie mentality. And there's lots of ways in which the music business today is structured to that I understand why people have that mentality. It's just like it's empirically not true. Here's how I think about it, though, for us, because I think there's a lot of human agency involved here, and it's like what world do we want to build? And not just like sit back and and be kind of doomers about the whole thing. And when I think about it, the thing that's so exciting is it's not just that we're making it faster and cheaper to make music. Like there's already infinite music out there for you to listen to, right? So, making more of it isn't um isn't necessarily the goal here. It's about doing things that have not been possible before. And when I think about that, that is like

how do we give the end user an entirely new and better creative digital experience with music. And so, very very explicitly, the intent is not let's build an AI-powered Spotify where you're just going to sit back and listen to music that happens to be AI-generated. I don't think that's a compelling product, incidentally. Like you can just go to Spotify and listen to great music. Again, there's there's more than you can listen to in a lifetime.

The thing that was not possible until like 2 years ago is that the end user experience that a billion people who currently stream music can do is one that is creative. And it's really hard to wrap your head around this, but we're so trained to think of like content is meant to be consumed. And so, anything that is meant to produce content, that piece of content is meant to be taken somewhere else and consumed by somebody else. And the new behavior, the new thing that is really exciting is that the thing that everybody is going to do is entertain themselves being creative, not just entertain themselves watching or listening to some piece of content. And when you are creative, you get so much like more involved in the thing.

And in any medium, I mean for us it's music, but when you are creative, it's like it's a very um innate part about being a person. This is like why we like to cook. This is why we like to paint. This is why people like Cloud Code so much. It's because like it is fun to build things and it's fun to use the things that you build. And um that is what is so exciting. And I think we will see this in lots of other domains. So, the words that we use internally are um creative entertainment. It's like >> Mhm.

>> lots of people are going around and just enjoying themselves being creative. And that is an entirely new thing. It is far more engaging. It does not replace anything. It is just um uh the the most warm and fuzzy feeling. And so, maybe maybe to close it out, I would say like we talk about kind of the the pyramid of what people can get at Suno. And at the base of the pyramid is utility. It's the professionals using Suno as a tool.

And then one level higher is entertainment. Um it is actually um somewhat elevated to be entertained and not just to use something as a tool. But the top of the pyramid isn't even entertainment. You can get entertainment anywhere. The top of the pyramid is fulfillment. When you feel the warm and fuzzy feelings of being creative, of seeing the fruits of your labor, of having other people recognize the fruits of your labor. These are feelings that um you can't really get anywhere else.

And um we don't pretend like we're curing cancer here at Suno, but like making people feel amazing is um a fundamental part of humanity. And so, I think it's just something that is entirely entirely overlooked, especially since like today most AI tools are just tools. >> I tend to agree with you. And I I guess the two key things I'm pulling away from what you just said are one, a human will always be in the loop. The goal here isn't to just robotify the music industry and have an AI-generated Spotify. The goal is to make people who are already creating music more empowered to create more and better music. And then hopefully other people will listen to that music. Uh but then two, there's a lot you brought up doomerism and this whole idea that AI is going to take away all our jobs and you know, everybody's just going to be destitute. Well, there's a worst case scenario where a version of that could be true, but

there's also a world where it just empowers us to do more and better work. And it it frees up more time for creative pursuits like creating music, uh which leads to more fulfillment. And so, I think that is the bright version of the future we're all hoping for. The

last topic I want to talk about here before we get to our lightning round is just the music industry has this whole history of resisting technology and resisting change that could infringe upon the way they've always done things. And so, on the one hand, the RIAA filed this copyright lawsuit against Suno in 2024. On the other hand, uh more recently in November 2025, uh Warner Music Group settled its case with you, and and it sounds like you all have this licensing partnership now that enables licensed AI models. So, how have you all navigated the ups and downs of legal challenges from the incumbents in the music industry over the last few years since you launched? And then, how do you see this ultimately playing out over the next few years? And how does that look the same or different versus previous versions of it like Spotify or Napster?

>> Yeah, look a a lot to say here. You know, without getting into specifics because like, you know, there is ongoing litigation. Music and technology don't have the easiest path together. It's checkered. It's not all good. It's not all bad. Um and this is why I believe so much in human agency. It's like there's a lot of ability to shape it. And um as we go through our journey and we see more and more people falling in love with making music, we get more and more sure that there are incredible ways to partner with people to build things that have never existed before and actually be the proverbial tide that lifts all boats. As an example, um the thing that we talk a lot about is how do fans and artists um engage with one another, interact with one another? Um today, it's not amazing. You have to go to a concert. Um concerts are amazing, but you know, the average person I think goes to less than

one concert a year. And there's a lot of talk in the music industry about um super fans. And it's like, how do we um bring super fandom to more and more people? Um both those a way to monetize and also as a way to engage. And um in my mind, the most elevated form of super fandom is one that I can do every day and one that I can do through music. And when I think about being able to play with the music of my favorite artist, interact with that music, I feel like I am co-creating with that artist.

I feel closer to them. I and it is a again, a far more elevated experience than just streaming their music while I'm doing something else um in the background. And so, it's these types of things where um partnering with the industry is actually just um good for everyone. It's good for right holders, good for artists, good for platforms. Um and uh that's just today. Like I imagine that in the future there will be way more of these types of things. I think people would usually expect me to say something like the major labels are cooked. Um I just don't think that's right. I think that's actually pretty stupid. Um they are um culturally some of the most important institutions uh on the planet.

And a world in which culture suffers because of this is like not the world that we are trying to build. Um and so, I'm actually quite excited about this Warner partnership, about building products together. Um and I'm excited for what may come after that. >> Yeah, I am too. And I I'm I'm glad that's the way that you're looking at it and hopefully it's the way they're looking at it. And I think it goes back to something you said earlier in the episode, which is it's not a zero-sum game. Like hopefully as technology evolves, the Pareto curve expands. It It enlarges the size of the pie and there are just more people creating music, more people consuming music, and the music gets better. And that's that's a good thing for everyone. So, on that

note, let's get to our lightning round. I always ask the first question first, and that is what is the best part about working at Suno? And what's one thing that you would change or as the CEO or or actively working to change? >> The best part is the people. Um I just derive an immense amount as a personal choice for me, but I like I derive an immense amount of personal satisfaction from who I get to work with. And so, um it's yeah, it's a fantastic group of people. Uh obviously most people here really love music. Um and so, uh that makes for fun after-hours things. We have a studio at the other end of the office and it gets used most days to make music just for fun. What's one thing I would change? You know, I I regret like the fact that we are um we have four offices. I love that we are an in-person company, and I wish we could all be in one office, and um geography dictates that that is a little bit hard.

And basically, it means that um there are some people in the company that I love that I get to see in person a little bit less. >> Yeah, this is something I hear remarkably often from founder CEOs in particular, which is a lot of people say that people is the best part about working at the company, but I still love to lead with this question cuz it's always interesting to hear the specifics, and you brought up how music brings people together at Sono, which makes sense. But then, founder CEOs in this era in particular, I just hear over and over again saying, "I wish we could all be together in one place." And it makes a lot of sense cuz you're pouring your heart and soul into the work, and that's just different when you're in person with someone versus when you're doing it with a computer screen. Let's get into music, since obviously that's that's the root of the whole company.

So, what is your favorite band, your favorite musician, and your favorite song? >> Yeah, uh that's that's a really hard one. Uh let me try to give a couple. Like um uh probably by by listening minutes, uh it's the Beatles. They're timeless, they're amazing. Um I also have two little kids, and so be the Beatles is also great for um for little kids. It's like kids music that um won't annoy you. Uh if you want a band, uh it my my default answer is I I've seen Phish like probably 45 times in concert, and I really really like them. And so, um I'm going to keep going back to that one. Individual musician, um I really I mean, will you permit me to say Chopin? I really like Chopin. His picture is in fact behind me. Um and favorite song, that's a tough one. That changes. Um uh it could be Starman by David Bowie.

I'll just go with Starman by David Bowie. >> Those are great set of answers, and I I you know more about music than I ever will in a lifetime, uh but obviously a a very diverse range of answers there as well. So, uh I I love the thoughtfulness you put in that. Also say as a father of two daughters myself, I never thought of the fact that a lot of Beatles music plays really well with kids, but then has all these layers to it. And I don't know if this resonates, but I'm a huge movie uh aficionado, and for me it's similar to Pixar, where you can go and watch a great Pixar movie with your kids. And like your kids are watching one version of the movie, and then you're watching a totally different version cuz it has so many different layers to it.

>> 100%. And so The Beatles is definitely like that. There's a lot of layers to it. You can't explain to your kids what all of those songs are about, or certainly not until they're older. Um just as a funny tidbit, I don't know, at the end of the year Spotify does the the does your like Spotify age, um and I'm 72. >> Oh, wow. I'm going to have to check that out. I wonder what my Spotify age would be. I like a lot of electronic music, like Owl City and Avicii. I don't know, I don't know what that says about me.

>> You might be 25. >> Yeah, I might I might actually be younger on the spectrum. What is your favorite subscription app besides Suno, and what do you love most about it? >> It's probably ChatGPT and Claude. Um if you just did it by minutes, I'm I'm probably a a dow of both of them. I don't use a ton of apps, if I'm being honest. Uh in fact, I I have no news or social media on my phone, um by design. >> I'm actually right there with you on that. I I'm on LinkedIn more than I probably should be, but I don't use a lot of social media at all. I never really have.

Uh and then, you know, we're we're I say we, my team at at Elemental Growth is on Claude code every single day now, multiple hours a day, and it seems to just be taking over everything we do at work, but in a good way, in a lot of ways. Um the music industry has a history of fighting tech companies. We talked about that earlier, from MP3s with Napster to streaming with Spotify, uh and now with AI. Uh So, how do you think the story ends this time, and does it end any differently than last time? We we talked a little bit about that before, but any final points on that?

>> Yeah, I think, you know, the the major theme here of what technology most impacts is the format of music. MP3 is a format, streaming is a format. Before them, there were CDs, before that, there were vinyls, there were tapes. I think about this through that lens. And so, um we fast forward 5 years, there will be a new format, or the new format. In in mind, it's actually kind of obvious that it's um interactive music. It is music that you play with. And um it is more engaging, it is more fun, um it will change the place of music in culture, um it will make music much bigger, and that's kind of the whole point of what we're doing.

>> Yeah. Well, again, as somebody who isn't as deep into music as you are, I may reveal my ignorance here, but it reminds me a little bit of how with jazz or with a jam band like Phish, it's like every time is a little bit different because the humans in the band are riffing with each other, and now you can riff with AI, and that just creates all sorts of opportunities and expands the pie of when you can be musically creative because you can't meet with your whole band at any time of day every day, but but you can open up your computer just about any time and start riffing with Suno.

So, that makes a lot of sense. >> I mean, just on that, like so important is that Suno is on your phone. And it's because creativity strikes at weird times. Um and your phone is usually with you. >> Spot on. Well, last question, what's the most profound change you think AI will have on music over the next 5 years? >> I think the real answer is making it more interactive, which is what I said before. Um and what I think happens from that is like yes, it gets very plentiful, um in the same way that the iPhone camera and Instagram really made imagery very, very plentiful, and how just massively pie expanding that can be, but also how it changes the place of imagery in culture, it will AI will definitely change the place of music in culture. And my sincere hope, but again, I think this is something that we need to work towards, it's not a given. My sincere hope is it makes it a much more prominent part of culture. And if you

think about it today, music is often a background thing. Um and I hope it really brings it to the foreground. >> Yeah, I hope you're right as well. And I guess I'll end by saying I feel like technology in so many ways over the last 20 years have pulled people apart and made us more isolated, more disconnected from one another, whether that's video games, which you brought up earlier, or it's social media. And what I hear you saying is music is one of those things that's been such a through line across human history and at its best music is interactive and it involves humans creating together. And so my hope would be that the technology that you're building at Suno just leads to more of these interactions with people listening to each other's music, co-creating music and and that would be a very good thing if it if it means that people are spending more time with each other as opposed to by themselves.

Any parting words you want to leave the audience with before we wrap up today? >> That was really well put. You know, I think I think about this like at some point today you're going to have a free 30 minutes and um are you going to doomscroll your favorite app for 30 minutes or are you going to do something that is creative? It doesn't even need to be in music and it seems to me that the on the art and entertainment and creativity side, um the biggest thing that AI will do to the world is make you skew toward being creative and productive instead of just doomscrolling and gosh, that's got to be a good thing.

Um last thing I'll say is we're hiring. We've got offices in Cambridge, Massachusetts, New York, LA and San Francisco. So come look us up. >> Well, I love that and it sounds like a very inspiring and exciting company to work for and a lot of great locations to work at. Thanks again for coming on today, Mikey. It was a real pleasure. >> Thank you. I had a lot of fun. >> Thanks again to RevenueCat for sponsoring this episode. If you're building a subscription app, you can use RevenueCat to power purchases, optimize paywalls, offer discounts, handle refunds, manage customer data, run AB tests, generate insights, and unlock more revenue. I work with dozens of subscription apps and most of them are already using RevenueCat to power their businesses. Learn more at revenuecat.com/subversion.