Prem Akkaraju, Chief Executive Officer, Stability AI, Dana Settle, Co-founder and Managing Partner, Greycroft Moderator: Allie Garfinkle, Senior Reporter, Finance, Fortune; Co-chair, Fortune Brainstorm AI
Category
🤖
TechTranscript
00:00Prem, Dana, I am so excited to be here with you, in part because this was not on my bingo card for
00:052024, talking about a turnaround for this company. Less than eight months ago, stability was in a
00:11rapid decline, but here you are. You both stepped in and decided you really wanted to take it on.
00:17Why? It wasn't on my bingo card either, actually. So no one's bingo card. Yeah,
00:22it's no one's bingo card. I think that, well, it was really because the technology and the team
00:29were too good. And I've been partners with Sean Parker for over a decade, and we have just
00:36did a very similar play at Weta Digital, where we created an application layer for film and TV.
00:43We made Avatar 1, Avatar 2, big movies with that company. And we sold that company in November 21.
00:51And after that, we were like, okay, we should probably be looking at the AI version of what
00:56we just did. And how can we use AI to make film and TV production, gaming, as well as marketing
01:03advertising more efficient? And then this opportunity came about, and I got heavily
01:09involved with Dana on this earlier this year. And this is where a lot of the outweighed returns live,
01:18is when you're willing to go in and do the very hard work to kind of turn it around. And we did
01:21that very quickly. And in opportunistic situations. And I think I'm fortunate to be
01:28part of the partnership now, the triumvirate, along with many other partners. But like Prem said,
01:34that is where the outsized returns live, is actually in jumping into opportunities that
01:40other people don't always see. And especially in an area like this, that's moving so quickly,
01:47and where the technology here, like Prem said, it was too good to let go.
01:52Dana brings up, I'm sorry to interrupt. She brings up a very good point. Opportunities are
01:59things that only a handful of people see. If we all saw them, they wouldn't be opportunities.
02:04And so that's kind of the definition. And I think that we saw it, and everything that we've done
02:09for this, how many months has it been now? It's been...
02:11Six months.
02:12Six months, slash six years.
02:14Six years.
02:15How long in AI time?
02:17Exactly. It's like 152 years in real time. But we've seen extraordinary results in such a short
02:24period of time. And it was really, it's good to see that our assumptions are really coming true.
02:29Well, and what was the financial state of the company when the two of you came in?
02:33What were we dealing with here?
02:34Well, she's the investor.
02:35Well, yeah, I mean...
02:36What were we looking at?
02:37The financial state of the company, they needed funding. And that was where we had to really
02:42move quickly on analyzing the opportunity. And so we moved very quickly, but also deeply,
02:47because we wanted to make sure that it was what we thought it was. And like Prem said,
02:51it's proven out to be, I mean, everything we thought and more. And I think in particular for
02:56us, it was really backing an incredible leadership team in Prem and Sean Parker to come in and lead
03:02what is essentially doing what they had done just a few years earlier in the SaaS space.
03:10I mean, you said they needed money. What for? How bad was it? Because the reporting at the time,
03:14my colleague, Jessica Matthews, and I spent a lot of time reporting on the story.
03:17And the narrative was that stability was a death's door.
03:21Well, I mean, look, all AI companies require a lot of capital. I mean, that's no surprise. I
03:26think you can read it anywhere. But I think what happens when you come in with a really incredible
03:33experienced management team, leadership team, is they know how to get in. And this is Prem. I mean,
03:38Prem really was working with the team at Stability even before we closed the transaction
03:45and helping to identify exactly where were the areas that they needed to focus on for this
03:49opportunity that we saw in visual media. I would say that people forget that Apple
03:56was 90 days from bankruptcy in 1997. Actually, I think they were probably closer to bankruptcy
04:00than actually this company was. And again, it all comes down to leadership. And you saw what happens
04:06when leadership comes in with a very succinct plan, a product vision, and the ability to attract
04:12capital, obviously. But the cleanup, we have now a clean balance sheet. We have no debt. We have
04:18nothing. And to negotiate out of those deals or within those partnerships, it was no miracle by
04:25my part, even though I led the transaction. It was a miracle. Give me an example of what
04:29we're talking about. How much of a miracle? Again, this wasn't on anybody's bingo card.
04:35Again, well, I'll actually give the credit. Well, I appreciate that. Thank you. But
04:42while I led that effort, we cut deals in the best way that you cut deals, which is
04:52alignment of interests. And everyone on the other side of the table that I had to deal with
04:57to clean it up and that we got there saw that this was essentially too important to the ecosystem
05:03to fail. Stable diffusion represented 80% of every single image that everyone in this audience saw
05:11in 2023. Just last year, 80% of them came from stable diffusion. That's gigantic. And when you
05:18think about what that implies with a 5 million person active community that works on our models
05:24every single day, that's growing every single day, that's a gigantic ecosystem that the
05:29infrastructure of this industry could not take a hit on. So I think it was the alignment of
05:38interests with those parties and our determination to make sure that that was articulated.
05:45Well, but I mean, the reason I say you should take a lot of credit for it is, I mean, really
05:50bringing everybody along at a time like that is, I mean, that is true.
05:54That's hard.
05:55And that is hard. And it's in doing it too when, you know, I mean, frankly,
05:59you weren't even CEO of the company.
06:02Correct.
06:02And so, I mean, I think when you think about how difficult it is to lead a company during,
06:07you know, tumultuous times and during super high growth and change, that's hard when you are the
06:13CEO. But what about when you're actually just coming in and sort of coalescing, you know,
06:18all of the different groups from, you know, investors to employees to partners, et cetera.
06:22And I think that, you know, that's what gave me greater and greater conviction
06:26at every step along the way was just seeing how Prime actually executed through that.
06:31Well, controversial past leadership has left the company, but a lot of the early investors
06:36in the company, including KOTU and Lightspeed, they left and then they came back.
06:41Yeah.
06:42How did that go? How did that happen?
06:43Is that incredible?
06:44I mean, I think I'll be honest, I'm surprised.
06:47That I think that says it all, you know.
06:49And I think they really re-underwrote to this leadership team,
06:52but also understood deeply the opportunity.
06:56KOTU left the board, they didn't invest anymore. Same with Lightspeed, same with SoundVentures.
07:01So, not only did all of three of them reinvest into this round, their pro rata that they were
07:09allowed only to invest. We did it as a rights offering, in some cases, even more than their
07:14pro rata. And not only did they do that, but KOTU actually rejoined the board and is on our
07:19board again now too. So, it was extraordinary because I think they saw that progress being made
07:25and kind of the positioning of re-establishing the leadership position.
07:30The reason that everybody's interested in the company is because they were so successful.
07:33Yeah.
07:34If it weren't successful, you wouldn't have these, like no one would care.
07:37What were those conversations like though? What was the vision that you sold them on?
07:42Well, again, it was alignment of interest. I think for the other investors, it was really,
07:47well, there was really my previous success.
07:50Yeah. And I was going to say, I mean, I think one of the really exciting opportunities here
07:54that we saw as a new investor coming in was essentially where we think that there is great
08:00opportunity in general in AI right now is really when you're sort of coupling the models with the
08:06application layer. And I think these models were phenomenal, are phenomenal, continue to be.
08:10And then having a leadership team that's so deeply-
08:14There you go.
08:15Just throwing stuff around.
08:18Understands the domain.
08:19Someone can catch it.
08:20Exactly. Heads up.
08:24Is really an extraordinary opportunity. And I think that's, again, why, I mean,
08:27for us as new investors coming in, and we are based in Los Angeles, New York, and Silicon Valley,
08:32but we really understand the media ecosystem and knew how deeply Prame understood and was
08:39connected in that ecosystem.
08:41Yeah. That's really that transition because we were, the first evolution of AI was really
08:46the foundational, the frontier models. And then it was quickly seeing this base because people
08:52always talk about the application layer and it is true that value is there, but being able to
08:56understand and fine tune models all the way into a core model is extremely important for you to set
09:03up success on the application layer. So I don't think, well, some people believe
09:09foundational models is a race to the bottom. I tend to agree as a standalone.
09:16We're really that application, the applied research and application layer,
09:19which is a race to the top. But that's a race to the top that we can have an advantage winning
09:24when we understand the entire software stack all the way down to the chip.
09:30Now, I want to talk a little bit more about the board. Kochi's back. You, Dana, are on the board.
09:34James Cameron is also on the board.
09:36Who's that?
09:37Yeah. Who here has heard of James Cameron? Everyone, I assume.
09:40Please raise your hand.
09:41You probably need to say who doesn't know who James Cameron is.
09:43Actually, who doesn't know who James Cameron is?
09:45Avatar.
09:46Avatar. Yeah, just scream Avatar. How did that happen? What does he bring to the board?
09:53So when I was CEO of Weta Digital, we made Avatar 2, made every frame of that movie with James
09:59Cameron. And so when I built the application layer at Weta and eventually sold it to Unity,
10:06you know, he had a front row seat at everything that I was doing with me and the engineering team.
10:11And so when this came about, I was in constant dialogue with James and in fact pulled him in
10:21early in the year before I joined as CEO. And I said, hey, I think this actually could be...
10:25And he said what?
10:27He said he agreed.
10:28He said he agreed?
10:29You mean about that this has the potential, of course. Well, I don't think he would have
10:33invested or joined the board if he didn't.
10:35I assume, yeah, that makes sense. That tracks.
10:37Yeah.
10:38But even back then, before you joined, before this full turnaround happened.
10:42Yeah. Well, no, it was definitely... It would mean he agreed.
10:44It'd be hilarious if he did join the board without thinking that, though.
10:47No, no, no, no, no, no. Well, there was no board for him to join at that time, right?
10:50I wasn't even the CEO of the company. So what I mean was he agreed was
10:55the question that I posed to him, which was, hey, I think this is worth looking at.
10:59And like, let's really dig deep into this. And like Dana was saying,
11:02we dug very, very, very deep into the talent pool.
11:06Like we didn't invest and I didn't join the company until I met every single
11:12person at the company, myself and Sean.
11:15And we dug very deep in the talent pool. It was so deep.
11:20I was so surprised because there's a lot of also, like, and no offense,
11:23the media kind of over exaggerates in both directions many times.
11:27And then so...
11:28Why are you laughing?
11:29I was reading...
11:30Well, she's laughing because it's true.
11:33But I was so shocked because there were some departures and there's other
11:39spinoffs that came off of it. But we were so,
11:44so happy to see that not only did the depth of talent still stay there,
11:47and there was over 90 people within the core and applied research team,
11:51but they had two other things that are beyond just their genius.
11:55And they were dedicated and they were very passionate
11:59about stable diffusion. And I found, and I told this to Dana just recently,
12:04during all that, you know, that was what feels like years ago,
12:08was months ago when I was meeting everybody and when we closed the deal,
12:12there was somebody on the applied research team that said,
12:14we knew you were coming.
12:16This is even before the actual deal was announced and closed.
12:18And they said, and what the...
12:20Yeah, I mean, it's an amazing statement to make.
12:22Meaning not me, but like, you know,
12:24somebody was coming that was experienced leadership that could bring the vision,
12:28that could bring the capital to the table and clean up the company,
12:31because it was just, it was too important.
12:33Well, they might not have imagined that it was going to be you,
12:35Sean Parker and James Cameron.
12:37Yeah.
12:39But somebody.
12:39To be fair, who could have imagined that?
12:41Yeah. Again, everybody's surprised bingo card.
12:44What does stability do today and what is the opportunity?
12:48Distill it down for us here and now.
12:50So our vision is, we're doing twofold.
12:54One is we're exploiting the commercial opportunities,
12:57significant commercial opportunities with our enterprise licensing and API revenues.
13:02And the second is we are now building the first full stack AI production pipeline
13:10for professional visual media content.
13:13Professional video media content defined as film and television, obviously,
13:18gaming and marketing and advertising.
13:21What about you? The opportunity?
13:24Where does this land for you?
13:25What's the end game?
13:27It's amazing because it really is the thesis that we had laid out at the beginning,
13:31which is that there is a huge opportunity on the creative and production side of media
13:37to leverage AI in the creative community.
13:39And really, if you think about it, we have this incredible product marketing
13:45genius and James Cameron on the board, deeply involved.
13:48And having that combined with this technology is so unique.
13:52And being able to then apply it to those three verticals,
13:55verticals that we at GrayCrop know very well,
13:58having been involved with companies like Scope.ly,
14:02which is a large gaming company, which was sold last year.
14:06Early days of YouTube with Maker Studios, AwesomenessTV.
14:09So this is an area that we're mounted on the advertising side.
14:12We just think there's such tremendous opportunity.
14:16Sorry.
14:16Yeah, no, no.
14:17It's just said perfectly.
14:18But I actually don't know how to do it any other way.
14:22I think it's pretty strange, because I did it at Weta that way.
14:26There was no space between the art and the science.
14:30We worked with some of the biggest directors in the world.
14:32We made the last 10 Marvel movies, and obviously Avatar 1 and 2.
14:36And we worked very, very closely with the directors.
14:39And that was such a great experience for me,
14:43because people ask me all the time after I sold the pipeline to Unity,
14:48it's like, how did you know what products to build?
14:51Who designed your product pipeline?
14:52I was like, it's easy.
14:53I just asked the directors what they wanted.
14:55And so it's true.
14:57So it became the product strategy, the product roadmap,
15:01the go-to-market strategy, everything.
15:03And also the QA.
15:05Everything is sort of built into one.
15:06And the confusing thing I find is,
15:09how am I supposed to build artist tools without an artist
15:14as a partner at the highest level of the company?
15:17So that's just, it's my muscle memory.
15:21Now, let's imagine, for the sake of argument,
15:24we're sitting here in a year.
15:26Us and a couple of our nearest and dearest.
15:30What has stability achieved?
15:31What does stability look like a year from now?
15:33I think robust managed service on our,
15:36which is growing at multiple, multiple.
15:39The head of marketing told me,
15:41I couldn't tell you how, in front of everybody,
15:44how much we've grown, but it's been extraordinary.
15:45But you can, are you sure you can?
15:47Exactly.
15:48I would like to.
15:49I would like to, because it's a gigantic growth rate.
15:51I think everybody wants to hear.
15:52I know, I know, I don't blame you.
15:53Yeah, exactly.
15:54I'll tell you in a year.
15:55Maybe she'll let me then.
15:56Honest here, you can ask her.
15:59So she wouldn't let me tell, but it's incredible.
16:01It's triple-digit growth.
16:06And so I think we're going to continue to see that.
16:08And I think that we're going to be seeing
16:11several breakthroughs in individual shot elements.
16:17We have a very different approach.
16:19We have a professional approach to content creation.
16:21We don't have one text to video model
16:24that has one output that combines
16:26like a gazillion different workflows into one.
16:28That's actually not how professional content's created.
16:31Professional content's created in shot elements
16:33and then they're composited together.
16:34By the way, this is where he's going to go really, really deep
16:37because he does understand this thing.
16:38We can talk about texture deformation models now if we want.
16:42We're sadly at time, but that's for next time.
16:45It's okay, I can go over.
16:48I mean, they all came.
16:51This is way more interesting than next panel.
16:55Thank you, Prem and Dana, so much.
16:56That is all the time we have.
16:57I wish we could.
16:58Thank you so much, guys.
17:00We'll do it in a year.
17:00We'll do it in a year.