Suno raises $400M+ as engagement metrics surge 50% in six months

Their business is secure. Crowd Strike secures AI and stops breaches.

I don't know if you can actually hear that. The the filter might be a little bit too much, but anyway, we can deal with that later. Uh we are very very fortunate to be joined by Mikey from Zuno. Welcome to the show. How are you doing on a massive day? He's here live with us in the TV and Ultradome. Amazing that you can be here. How you doing?

I'm I'm doing great. I'm excited to be here. This place is awesome.

Thank you. Uh give us the news. What happened today? Hit the gong.

Uh I hit

hit the gong. Tell us what happened

and then come back and then come back and tell us.

I like how it shakes the entire camera. Anyway,

that was by far the strongest hit from a guest.

Yeah, a lot of people come in and they don't they they they don't want to break the gunk to be it's my me.

You told me to go at it. I went that was perfect.

Well, it's a good day for that. That's the new bar.

So, take us through why are we hitting the gong?

Uh we're hitting the gong. We raised a little over $400 million uh announced today. Very exciting. Um led by bond a history of investing in really category defining consumer businesses which is really really cool. Um, lots of new investors joining. All of our old investors, uh, participating. So, really, really excited about that.

IVP, Ferrer, Union Square, Matrix, Quiet, Light Speed, Menllo. Wow. Murderers row. Good job. Uh, what unlocked this round? Why do this round? Are you spending a ton on inference? Are you spending a ton on training? Is there a huge cost? Or is there just so much traction that that unlocked the round? Like, what's the thesis?

Really, really more the latter. there's been a ton of traction um and a ton of actually progress since the last time uh I was on and so um just to just to zoom out here

where yeah where is the company right now in terms of the products

so every time we we release new products we release new models uh more people come to the product but more people stick around and so the engagement metrics that we track say it's

usage retention uh what fraction of our users come three or more days a week

uh session time all of these are getting really really good. All of these are up like about 50% in the last 6 months which is huge and this is just like constant making the product better, making more people fall in love with being creative and so it's been incredible to see. Um and uh yeah it's like a lot of hard work from the team but it's

what is the northstar metric on retention like Dow Mau monthly churn like how are you thinking about actually like what is the correct metric to measure for your particular business

you know yeah different businesses are going to be different we we like we like retention um I think all of these are kind of very very similar um the thing that's like really important here is just to remember yeah like revenue is is a trailing metric for how good is your product and retention is really the best way to think about people are coming to my product over and over again. They're falling in love with it. There's no real hacking that basically.

Yeah. And are you early or you are you far enough along to see like smiling curves in the retention curve?

Yeah, for sure. So, it's been it's been really exciting. So, there's also there's two retentions. There's usage retention, there's paid retention. So, remember we have a pretty generous free tier. And so, uh our free users are also still coming back over and over again. I think we've created them through having that magical moment, their eyes light up. And so it's it's really around how do we let people do that themselves and really feel creative by themselves.

Uh walk me through the paid tier philosophy, a single tier. How like like do you have uh like are you thinking or do you already have like ultimate pro plans where someone can spend a lot of money if they're using this as like more of a proumer tool? Uh enterprise like how how uh how how how striated is the is the monetization at this point?

It's a great question. I think the short answer is we haven't really changed our pricing all that much in in a long time. And so uh it might not even be appropriate. But what we have now is there's a free tier, there's a $10 a month here, there's a $30 month here. $30 month tier gets you not only more usage um but some different power features y um and then

you know AI is this weird thing. It doesn't really fit neatly into SAS pricing and so a lot of sub companies are trying different things

a lot like everyone else when you reach your cap you can buy overages and so we've got lots of people who

they spend hours and hours a day on it and they hit their caps and and they keep paying for more. Yeah, it's such a weird dynamic in the modern era where you will have customers that are effectively gross margin negative potentially depending on exactly where you set the thresholds and then you will have customers that are they they want the best product. They're happy to pay but they come once a month with something and uh and they're probably much higher gross margin and there's just this blend because we're in this like you know real marginal cost era no longer the zero marginal cost era. you know, a few things there, like actually our margins are pretty okay and music models are small and so um it's not really the same cost structure, but the other thing is it's really interesting. There's not really a lot of consumer businesses being built right now and certainly not a lot of consumer entertainment businesses being built right now. And there's a real different I think interaction that the average user has when I'm sitting down at co-work and my boss is paying for it. And so I don't really think or care about how much money this is costing me. And when you're building for consumer entertainment and you're paying for it and it's not for doing a task, it's actually for enjoying yourself. Um, you actually do think a lot about it. And so, a, it makes our jobs harder. We actually have to deliver more. But B, you you you want to make sure that you provide the right experience where you're not constantly thinking about, oh gosh, how many credits do I have left before I hit the pay wall?

If uh if if Ben Thompson were here, he'd probably ask you about advertising. I feel like uh consumers want to be entertained. they don't necessarily want to be productive. It's very hard to get consumers to pay at scale. Obviously, the AI era is like a big narrative violation of that because there's tens of millions of consumers that are paying for AI tools across the entire industry. But, uh there does seem to be a limit to how many consumers will pay even $10 a month. Are you thinking about uh trading credits while you're waiting for a song to to generate ads? How are you thinking about that? We're not thinking at that scale yet. I actually kind of disagree with Ben here, which is I mean he's he's he's awesome. He's brilliant. But I just I just think about like uh 800 million people about stream music.

A little less than half of them pay for it. That's actually a lot of people. Like 300 million people pay for streaming music. That's a lot.

Yeah.

Um

I just think that that like the history here.

So So there's billions and billions of people that don't listen to music online whatsoever. Is it really only 800 million? accounts

or is that like accounts on Spotify that doesn't maybe count YouTube?

That's uh to my knowledge that's like Spotify, YouTube, Apple, and maybe like five others.

There's like a billion records now too.

Vinyl is going to make a big comeback. And um I mean I think as we um change the place of music and culture, as the product gets better, we change the place that music has in culture. We see these creative moments happening. And so you see them digitally now where maybe you saw the Puerto Rico song or you saw um an amazing creative trend where people are making like their cringy text messages into songs and they go viral on TikTok. And I think you'll start to see that penetrate actually the IRL world as well. And so I yeah I'm long vinyl.

Yeah. Uh, I was talking to someone who wanted a job at Sunso and they were uh they were asking me what I thought about the business and uh obviously I've been bullish for a long time like this new entertainment form factor, people love the product, you guys have real revenue, you're growing really quickly. Uh the my unsophisticated view was that I I told the person I can easily imagine this as like a10 billion company. It's hard for me to see the world in which this is a hundred billion dollar company. Uh why am I why am I wrong? I know you're insanely ambitious. You're not gonna you're not going to 2x from here and then be like job's done everybody.

No, the job the job's not done. I think um this is one of those things actually. I think you've just been hanging around investors too much, you know, which is um the thing uh that you know you have to be right for a long time while while a lot of people are wrong. And the thing is every every revenue scale we hit, every usage scale we hit, the question is like, but how many people really want to create? And that was we got those questions at $100,000 in revenue and we get those questions at $100 million in revenue. And the lived experience we have playing with the product with people, but also the metrics and looking at our growth curves and seeing that every time we make the product better, yes, things inlect up, but we appeal to a broader and broader audience. So I feel very

I mean the same the same thing would have been said with early creative tools on the internet which was you know how big you know with Figma was always how big could Figma be you know there's only so many designer professional designers then of course like the rest of organizations same thing with with Canva I'm sure a lot of people passed early on because they just thought like yeah there's just not that many people relatively that uh that do design and then of course like you make the tools more accessible uh and And so so anyways,

so yeah, and so there's a couple things here. One is I think actually we'll have a very large fraction of those 300 million subscribers to streaming also subscribe to Zuno in the long run. But the other thing that's amazing is that for the first time in a long time in music, there's innovation on the end user experience. And so you can have something where there are multiple products out there like right now the end user experiences for music all look the same. They all look like a a streaming provider, right? It's just like I have all of the world's music living on my phone and I can listen to it and that's it. In the future when you can do more things and play with it, you probably subscribe to multiple video games. You probably subscribe to multiple video streaming platforms. Why can't you subscribe to multiple music things? And I think the answer is in 10 years you will.

Yeah.

Do you think Apple would ever make a play in this space?

I think lots of people are going to make plays in the space. Uh it's almost validating to see that people finally think of music as a growth area uh again. And so, but again, I think that there's so much room for innovation here. There's so much green field to build upon that I'm actually really excited.

Yeah.

Can you take me through the story of the Puerto Rico song? What happened there? Like like what were the key inflection points? Actually,

I don't I I don't know is the answer. You know, this is like a song from a creator and and it hits the right uh cultural moment and the right vibe and and I don't want to say it's a good song or it's a bad song, but it is a platform where so many people can be creative on top of it. And that is really fulfilling to see. And that's stuff that it's like new use cases for music that haven't been around for a minute.

Is AI music becoming less controversial or there still people discovering it and getting angry? because I could imagine there was a wave over the last uh 2 years where people were like, "Wait, they're making music with AI. This is a this is meant to be a human creative pursuit or whatever criticisms they've they've thrown at it." But I imagine over time people start to realize that these are fun entertainment products and tools and uh before you form an opinion, you should just try using it and form an opinion based on your experience, not um you know, internet comments or things like that. But my expectation is that uh it will actually become less controversial year overyear as it sort of normalizes.

It is definitely becoming less controversial. There are still plenty of people who have pre-formed opinions and then you know actually playing with the product is the best way to change their minds. Um the thing that has become apparent in the music industry is that so many people are using it without talking about it. And as people realize that everybody knows that everybody knows it certainly normalizes there. And then as a consumer, either when you play with the product and you realize how fun it is, or when you realize actually there's some AI music in the music that I love, all of a sudden you're not so anti-AI. I think this is a lot like people being anti-AII and then they realize that chat GPT can help them with their homework and they kind of soften a little bit

as a useful tool. I mean Martin Scorsesei came out yesterday in a Black Forest Labs video talking about the value of AI in the film making process but the video demos we'll watch it later on the show but it it demos um sort of like a previsualization step almost like a storyboarding step. He's describing this alpine town and it's generating visuals for him. He's not taking it end to end. Of course, that that that that's happening more and more, but uh but he that I I think uh having a filmmaker like Scorsese kind of step up and say like, "Yeah, this is useful. I can see this working."

Absolutely. There's an analog in music actually, which is that you'll have songwriters um make beautiful songs and instead of making demos for an artist, they can make lowfidelity songs like lowfidelity productions which are demos with so much easier. Yeah. and it will get ripped out and produced without AI or with minimal AI when it's produced for the last time. But

do you think there's still an opportunity for major artists to be a first mover in the space? We were talking uh I think a mutual friend of the three of us and I was telling him uh he loves he uses it all the time. And uh I was telling him like why don't you just be the first one to take like a proper leap and do this and like the first person that goes and says like hey this is okay it's not so scary I used it in the creative process we still used a bunch of you know uh traditional producers sing you know songwriters etc. Um, I think that person will get a lot of like credit and potentially, you know, a lot of attention. It will obviously be polarizing, but then the fourth, fifth, sixth mover won't really get any potential benefit. It'll just be normalized by that point.

I think that's totally possible. Like, we've, you know, there are people, there are a few people who are outspoken about it, but I don't know if we've had like the moment yet.

I want like the project like somebody has to make like a hit record. There have been I think what what you're

and they have to they have to go out and say they have to be just very upfront and I what I want is them to document the whole process like I want a video camera.

How much is enough suno in there? Actually

that's the that's the whole point. It's like like a guitar might have only been like one little riff in a song and it could have been some originally like somebody recorded it on their phone and then they brought it into the studio and then like maybe it got transformed and made with a different Yeah, exactly. And so to me to me there just needs to be

it's it's always going to be a spectrum, right? There's going to be songs that are heavily Sununo, but I would expect for like major artists, even if like the the original kind of idea or um some element of the song was made on Sunno, like you said, it'll just get transformed and adapted and evolved. And so I think it's just positioning it as not like a replacement for anything, but uh another kind of individual artist in the room, another instrument in the room uh that's contributing to the project. Here's how I think of it. We we we haven't hit that moment, but at some point somebody will really document and celebrate the use of these tools in making things like we're already in a lot of very popular music, but at some point somebody will really do that actually proactively and that will be a huge cultural moment. And then the next amazing cultural moment will be an artist releases an album that is meant to beed by their fans. Yeah.

And um this is the new format. This is the thing that will pull people in that deepens the artist fan connection and that will be a very very important moment.

How spiky is Sunno's intelligence? I want to know about the training process. There's

we know you have to leave. So final maybe that's the last question. Um but uh the uh there's this like famous story about uh I think Andre Carpathy told it about like GPT4 41 some iteration where at at a certain checkpoint uh it was terrible at chess and then all of the AI researchers just were really interested in chess and they were like there's no reason it should be bad at chess and so they added a bunch of chess training data and then all of a sudden the models got pretty good at chess and it seems like we're in this spiky intelligence race where coding is really valuable. There's a lot of money there. So, the models are getting better at coding, but if you look at the models, like it doesn't really feel like they're getting that much better at copywriting right now because that's not where the boom is. They're getting really good at coding. And I'm wondering if you're seeing like, okay, there's an opportunity for us to nail jazz in the next iteration or us to nail EDM in this iteration. like is it is like what fingerprints is the team leaving on the model versus just we're scaling up and it's getting better at everything and there's surprises and it's very indeterminate. We we actually try not to leave our fingerprints on it. And we have a like the approach is we should not be the arbiters of good and bad music, you know, like a file corrupt or or lowquality sound recording, but like um you can go and you can listen to a song that has 10,000 YouTube plays or 10 million YouTube plays and you won't actually be able to tell the difference. And so we should not be uh the arbiters of that. The thing that's different here also is that um in the chess and the coding examples, there are objective right answers. And that's why the models are so good at getting spiky there. Yeah, music has no right answers. And so

there's no verified rewards, verifiable

rewards. And so um all of those techniques actually don't work nearly as RLVR.

So we're actually in the regime where it's much less spiky. And actually when we find a hole, an anti-spike, then we actually try to fill that in. Like it's not like we're amazing at jazz. It's like you suck at jazz or you suck at country and now we have to go and figure that one out.

Okay, last question. We'll let you get out of here. Uh, what about uh adding on to the viral fuel? When I hear about turn your text messages into a pseudo song, I think why can't I just screenshot my text messages and then you edit the video for me and I get a viral format. That feels like very tool you see very like each video format is going to be a whole process for you. It's going to be hard to like oneshot the full end to end. But what do you think about personally? I want the Slack integration. I want a pseudo song at the end of each day. Then

yeah, why weren't you on the stage at Microsoft Build?

I want my I want the workday at at the office summarized.

You want a workday integration?

No, I want I want the Slack summarized and I want to be able to listen to an hourong song on my drive home about what's happening at work.

We'll try we'll try to make that happen. We're not doing a workday integration, but I will try. The thing that's amazing is actually this is a platform where people are creative on top of it. And so we're not going to come up with the right cringy text message videos. Somebody is going to do that. Well, uh, that's a challenge for you at home. Go make the cringey text message. Go make the workday integration. Prove Mikey wrong. He says he won't build it. You can. You can go and build.

You can go and build the workday integration yourself.

Well, I want to react. Imagine keeping up

at work.

Yeah.

Through music, John.

Be fantastic.

What's more powerful than that?

You know what's more powerful than that? Console.com.