Inspired with Alexa von Tobel
00:09:22 · Alexa von Tobel
[00:09:22] S1: It is, by the way, having also built the company and sold it to Northwestern Mutual. I think for an acquisition years later for people to all feel good about it is is rare, and can... Congrats on being in the 1% of founders selling a company that goes really well. I think that really speaks to who you are. But before we go into Suno, alongside your work at Kensho, you were also a lecturer at MIT Sloan since 2018 teaching machine learning. In just a nugget, give us a sense of what you love to teach about machine learning. Make us all smarter. What sort of was your core concept as a as a professor?
00:10:02 · Mikey Shulman
[00:10:02] S0: Yeah. Totally bananas that they let me do this. You know? And So the course that I teach is a very applied course. It's in the business school, and it is for students who are largely going into quantitative trading. So it's meant to be very applied, it's meant to be very hands on. And the world of machine learning moves so quickly, but it's so particularly for natural language processing, as more and more documents get put out in the world, there is more and more edge to be had by understanding those. And there are tons and tons of classes in quantitative modeling, in econometrics, and just not enough very applied coursework in the nexus of machine learning and finance. And so with a professor at Sloan, of realized that this was like a gap and and made a very applied course. And I've actually, unfortunately, in the last year stopped doing it for lack of time, which actually bums me out greatly. But... Yeah. I I mean, I I loved every minute of it. Know, I'll tell you something personal. In 2020, the course went all remote, obviously. And it's like fine, I hate teaching over Zoom. And I remember in 2021, the first lecture back, and it's like fairly crazy draconian rules, you know, like everybody is like, we're in a big auditorium, everybody's six feet apart, everybody's wearing a mask. It's, like, not... And I remember at the end of that first class, I'm, like, so excited to be back in person. It's four to 7PM, so it's, like, an evening class. It's three straight hours of lecturing, and I come out of that class, and I'm like so energized after having been back in person with everybody after a year on Zoom. I couldn't fall asleep that night. I fell asleep at like three in the morning, and it's just like it's like... It's just an amazing data point for myself on, like, how fun it is to be around people when everybody is, like, into what you're doing.
00:11:57 · Alexa von Tobel
[00:11:57] S1: I'm just smiling because the entire Inspired team, very similarly, when it was very draconian and sort of, you know, we went back into the office in 2020 because we were like, we can't be a part. This is so much less fun. And part of the incredible thing of working with people who are jazzed about the same thing in person, it's really, really fun. So I appreciate that. But what a good insight about yourself, Mikey. So, Mikey, I wanna rewind to a few years ago when I met you and you... It was, you know, very early to what's happening to big LLMs and you were like Alexa, I'm gonna build one of the first audio LLMs and you were not even totally even sure at that point what the product was gonna be exactly. And I remember you were like, we're gonna see where it goes. Luckily in that moment I was like no matter what I I want to. I want to be on this guy's team and and invest in anyway. But talk about what is Suno and how did Suno start? Obviously you guys are technically brilliant. But where did Suno start? Where is Suno today? And kind of just give us a sense of that journey.
00:13:09 · Mikey Shulman
[00:13:09] S0: A lot to say here. So I think of us as we are trying to build the next set of super engaging digital experiences around music. We are trying to make music a much bigger part of your life than it is today. Music has a particular place and culture, and that is different from other forms of media. And it is only for lack of imagination that tonight you're gonna go home and instead of watching Netflix, why wouldn't you do something around music? And we are best known for giving everybody the ability to create, and that is one of the most engaging and valuable experiences that people can have. But it's actually much more than that in the grand scheme of things. And I think about this really as if we are doing our jobs right, eventually, will stop talking about creation and consumption differently, and we will just be saying, like, I'm going to go and do music. I'm very struck, you know, as a kid, I I... So I grew up playing in a lot of bands in high school and college, and some of the most fun I think I've ever had is just making music with my friends. And it is trying to bring those feelings digitally to billions of people that is what gets everybody here out of bed in the morning. You are right at the beginning. You may even remember we said we would need a 100 times more compute to be able to produce good audio. And so we thought that that was years away, and then a couple months later, we turned around and said, just kidding. I think we can do it now. And so that is the... You know, that research breakthrough, honestly, was the seed that got the whole thing started down the path that we are now. But I think if you zoom out, the thing that we really realized is that a, audio is everywhere. It is... It's like, you know, you can shut your eyes. You can't shut your ears. And the amount of audio produced and consumed is immense. But b, in the world of machine learning, audio is like a bit of a backwater, or it certainly was in 2022.
00:15:11 · Alexa von Tobel, Mikey Shulman
[00:15:11] S1: Absolutely. [00:15:12] S0: We realized like it is it it is only for stupid reasons that audio isn't like a first class citizen in machine learning. And we said like, why don't we why don't we try to make it a first class citizen? Why don't we make audio be as good and as important as we think it should be? And that is the thing that got us started down that path. That is the thing that made us realize that we thought it would be too hard to make good music, and that is ultimately the thing that drove us to, like, push through that barrier and and make that research breakthrough and be able to do it. And the rest is sort of history. [00:15:42] S1: Let's talk a little bit about just sort of the idea that a moment to be like, everyone is running over text and content, LLMs. Let's go do audio. And you're absolutely right. There was nobody, near to nobody, focusing on it. Where did that insight come from? [00:16:01] S0: Honestly, I think it's just because we liked it. We, like... We were doing a little bit. Maybe 10 of what we were doing at S and P was audio related around earnings call transcription. And we just realized, like, this is messy and annoying, but we freaking love it. And so it was... It wasn't... I don't know if there was, like, a first principles thing. It was, like, just this realization that, like, we will keep doing this, and we like it, and no one else likes doing it, and this can be very, very big, and everybody's just, you know, overlooking this maybe because it's less sexy somehow than than text. And I think that proved true in the long run.
00:16:01 · Mikey Shulman, Alexa von Tobel
[00:16:01] S0: Honestly, I think it's just because we liked it. We, like... We were doing a little bit. Maybe 10 of what we were doing at S and P was audio related around earnings call transcription. And we just realized, like, this is messy and annoying, but we freaking love it. And so it was... It wasn't... I don't know if there was, like, a first principles thing. It was, like, just this realization that, like, we will keep doing this, and we like it, and no one else likes doing it, and this can be very, very big, and everybody's just, you know, overlooking this maybe because it's less sexy somehow than than text. And I think that proved true in the long run. [00:16:37] S1: Got it. Let's quickly talk. So for those who are not part of the now 100,000,000 plus users of Suno, which is incredible growth by the way. I you've gone from zero in revenue when we invested to 200,000,000 plus and and I'm sure that number is already changed. A few quick stats because this is crazy. Suno was founded. I really call it 2022 released to the public in 2023, and in less than three years, 7,000,000 tracks are generated a day. And, obviously, you just raised it almost a $2,500,000,000 valuation. What's happening, Mikey? Like, let's just start. What is happening?