OpenAI Chaos Continues with Sunny Madra and David Sacks | E1852

29:53This is very, very simple. LinkedIn equals business. Business equals LinkedIn. So I want you to get started today and you will understand very quickly why LinkedIn is the place to be for B2B. We're going to give you$100 in credits so you don't have to spend any money. You get to test it yourself. LinkedIn.com slash this week in startups. No spaces, no dashes. You type out those four beautiful words this week in startups and you get$100 terms and conditions do apply. Yeah, it's clearly not a threat to civilization. And therefore, these guys on the board, the folks on the board, jumped the fence.

30:26They did something rash. They're completely inexperienced based on what I've seen or probably have never been on the board of a trillion-dollar company or a$100 billion company. With the exception of Adam D 'Angelo, obviously, Cora is a very important site, and he's a pretty experienced person. CTO of Facebook before that. CTO of Facebook. I mean, he's legit. But this is a dysfunctional board built on a dysfunctional corporate structure. and if they did have a split with Sam, there are 10 different ways you could have handled this that would have been better. The easiest one would be to say, Sam, it's pretty obvious that you want to be doing these other things.

30:59We as a board have decided we want to have a full-time CEO who doesn't have conflicts. Therefore, we'd like you to consider maybe moving into another position and segueing over the next year and what would be a good way for us to approach this collaboratively to make it great for you and obviously not have any reputation damage We certainly don't want to damage OpenAI or the foundation. So how can we collaborate on this? Instead, they shivved him in the middle of the night on Google Meet. I mean, the ultimate of insults. So, Sax, how would you have him? They couldn't even use Zoom? Is that what you're saying?

31:32Yeah. I mean, have you used Google Meet? I mean, that's janky. Well, and Microsoft put$10 billion in the company. They couldn't even be forced to use Teams. Or how about in person? How about we do this at a dinner, at an offsite, or we have a thoughtful one-on-one. We do a walk and talk. If you're going to ask someone, Sam Altman's profile, who put his life into this and his reputation to move out of that top spot, how would you handle that, Zach? If you did have this philosophical arrangement, what would be the proper protocol here? Well, look, we don't even know that they did. I mean, we're speculating about something where they haven't even provided an explanation.

32:09I mean, again, based on Ilya retracting, it seems like there is no explanation. But look, I think that if one were to look for a root cause of this whole thing that goes beyond just the immediate events of the last few days, I would say it has to be the nonprofit structure, the original nonprofit structure, and this funky corporate structure that's been set up with this weird board that consists of a bunch of people who don't really have skin in the game, like no VCs as far as I can tell. and people with fancy titles and effective altruism or whatever, basically people whose main credentials seem to be liberal arts majors and stuff like that.

32:53I mean, for one thing, I guess when this first happened, I was really surprised that Sam could even be fired because I just kind of assumed that he had control over the board. I mean, as the principal founder CEO, I just figured he had control. So I was very surprised that something like this could even happen. You know, on our pod, I speculated that he must have ownership in this. It just can't be the case that he has zero ownership. And I guess I was wrong about that. I mean, I guess he really doesn't have equity and he really didn't have control. I mean, I guess that's probably the most surprising thing to me about the whole thing.

33:29And I guess that springs from the fact that it was set up in this funky nonprofit way. and I just think it just all goes back to the point that you can't do better than a Delaware C Corp you know whenever I see founders trying to innovate on structure, legal structure I know it's going to be a disaster, like that's not where founders should be innovating founders should be innovating on product and technology, not in coming up with you know, newfangled corporate structures and org charts and governance and all this kind of stuff and I think there was a skepticism. Before ChatGPT came out on November 30th last year, I think there was kind of a general skepticism about what OpenAI was doing, not because, you know, the people involved weren't seen as serious, but just, it was just seen as bizarre.

34:20Like the idea that a nonprofit foundation could somehow be on the cutting edge of technology and lead to some sort of, you know, technological revolution. I think most people would have thought that was the really fanciful idea. I mean, Nonprofits have their place in the world, but it's just not generally pushing the boundaries of discovery and innovation, right? It's more about distributing products that already exist to needy people. So in any event, I mean, the structure was always really weird. And it appears that the weirdness never fully went away. I mean, they tried to paper over it. They tried to create these complicated crisscrossing structures that allowed Microsoft and others to invest.

35:01but they never fully solved the problem that the alignment that was set up in the beginning just wasn't very good. And I think one of the alignment problems is that Sam didn't have a proper comp package and I think the board wasn't set up as a proper board. I mean, for all the shit that VCs get, like VCs, and I think you made this point, Jason, VCs wouldn't fire a founder where everything is going up and to the right. Like, that's suicidal. Quite the opposite. What did they do with Elon when it was up and to the right and he was fully vested. They created a comp package to keep him there, right?

35:36You're talking about Tesla, yeah. I'm talking about Tesla. Tesla created a comp package where the better Tesla did in terms of his public valuation, the more stock he got. So he had these very far out of the money stock options. I think Tesla was worth, I don't know, tens of billions at that point. And his comp package went all the way up to$600 billion, meaning that if the company would ever be valued at$600 billion, he would invest additional owners. So they gave him these far-out-of-the-money stock options. So the better the company did, the more ownership he got. Yeah, I mean, everyone had skin in the game.

36:13Everyone was aligned. I mean, this is the beauty of the – I think of the way that the startup ecosystem works from just a legal and structural point of view is it's – I don't think that Silicon Valley would work without the innovation of the joint stock company. You create a company and you have widespread ownership by founders, VCs, and employees. And everybody's aligned. Everyone has skin in the game. And then within that framework, the founders run the company and typically have very broad latitude to do what they want. And everyone's comfortable with that because everyone's aligned. and when everything is up and to the right and working no one messes with that it's only when things descend some sort of crisis or you know there's some sort of criminal issue or something really really horrible do you ever like typically get a big intervention but otherwise when things are working you just leave them alone because again everybody's aligned for that big outcome this is the key point you know the structure here sunny in silicon valley has been perfected over decades and it's been battle tested right and the delaware sea laws are the you know the manifestation and the codification of that this franken structure franken structure they came up with you know which they i think they modeled off of mozilla and the mozilla foundation because they started printing hundreds of millions of dollars in google adsense revenue this is the original sin then you put a bunch of you know non uh stock price incented individuals on the board and you get this weird outcome you can say all the terrible things you want in the world about capitalism and vcs but the focus let me just add something to that okay there was a widespread belief or a belief that a lot of people are indulged in and maybe even sam and this could have been his downfall that somehow open ai would be a more trustworthy company if it did not intend to make a profit.

38:14This was sort of the premise behind all this foundation nonsense, is that if open AI does not intend to turn a profit, we can all trust it more. And you heard this repeated over and over in many different ways. And I think even Sam may have bought into this. And I think ultimately, that was the undoing here. If we level up the conversation, Sax, if you think about operating systems for building things in the world, for changing humanity, say what you want about capitalism, but it aligns perfectly with hard driving people and capital allocators. It aligns perfectly. And it is a more trustworthy situation because you can trust that everybody's working towards increasing the share price.

38:56And the share price obviously represents something in the world. It represents progress. So again, you can hate capitalism. it's imperfect but it's the best system for building companies and changing humanity in the best way and i think that's ultimately what we'll see here is this is going to be um what do you call it when you have like an ultimate decision there's a word for this but it would be the ultimate decision on capitalism versus non-profits just uh butchers your point there let me quote adam smith uh what adam smith wrote this is in the 1700s is that it is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own interest.

39:40In other words, we can trust people in capitalism, not because they're going to do something altruistic for us, but because they're going to pursue their own self-interest. And in so doing, they're going to produce something that we all want. I mean, that was basically what Adam Smith said. Right. And if you don't, what happens? It's that profit interest that allows us to understand their motivations and to align their incentives with something that's good for all of us. Now, as soon as you take away that profit incentive, say, for example, by putting a bunch of effective altruists on your board, now you don't really know what it is that they're optimizing for.

40:19They claim they're optimizing for some principle or other, but it may just be some crazy idea in their head that we shouldn't trust, or it might be some bureaucratic incentive. These people— It's for social credit. Let's call it what it is. They're doing it for credit or they want to be some resume-packing item or some—who knows? Who knows? It's some sort of—but you're never going to take human incentives, motivations out of something. You're just going to replace the profit motive with some other kind of motive that might be even more nefarious. And you're wondering, like, how did this happen?

40:56I think one of the reasons why it's so inexplicable is because we don't know the motivations of these board members. And you pointed out before, Sam did not have a package. This is, if you were on a board and your CEO who was doing this dealmaking was so important and critical, and they didn't own 10, 20 % of the company and their future incentives were not aligned and you didn't have them locked up for the next five years, that would be job one of the board is get Sam a package and lock him down. And maybe the reason he was doing these side deals was because he wasn't compensated at OpenAI. Totally.

41:30I think that's an excellent point. Look, we don't know what side deals he was doing, but he was running around making a lot of deals. I guess there's been a couple that have been reported. But if you're worried about that, if that was the worry, re-up the guy massively. Give them a huge stock package. Give them the Elon package that they did at Tesla. Yes, absolutely. So that he's singularly focused on making open AI successful. But again, this goes back to the motivations of this foundation and this funky setup they had. It's not clear what it is they were optimizing for. I mean, I guess some vision of societal good, but who judges that?

42:07And so at the end of the day, it just becomes a highly political thing where you've got a room full of people. And, you know, they're optimizing for some sort of, you know, bizarre political dynamic. Are you building the next great AI product? Well, if so, you know how expensive APIs can be for their model training data. Training AI is pricey. That's a fact. We all know it. So you have to try the Brave Search API. Yes, I am talking about Brave, the privacy browser that I am obsessed with. Brave's browser has 65 million users. Think about how much data that drives for Brave Search, which is the only global scale independent search index outside of big tech.

42:48And that index is available to anyone with the Brave Search API. The Brave Search API can power your chatbots and train your models, inform answers to real time queries, and it will serve images, web results and even rich text snippets. The Brave Search API features an easy, intuitive data structure, and its data is populated by real human interaction, not web crawlers, all for a fraction of the cost of the major players. It's free for up to 2 ,000 queries per month with paid plans for as little as a$3 CPM. That's cost per thousand. So if you're building a next-gen AI app or chatbots, you got to try the Brave Search API.

43:24Get started today. Brave.com slash Jason. Ooh, Jason's brave. I like it. Brave.com slash Jason. And get the browser while you're at it. It is awesome. It's also got a VPN built in. That's pretty cool. David, a big question now for, I think, a lot of the folks that tuned into this regular show with me and J-Cal is portfolio companies. You guys at Kraft have made a really big push into AI companies that are building around AI. And obviously, most of them are building an open AI. How are you rethinking that? What have been the discussions inside Kraft around what should companies do? and that are reliant on this?

44:03What are your initial thoughts there? Well, I don't think that's going to change. I mean, I think that what OpenAI did with the launch of chat GPT was this huge proof of concept for large language models. And now that they've kind of opened the door, no one's going to go back. So I think probably 70 % to 80 % of what we do these days is AI-related, at least for new investments. or probably 70-80 % is looking at it as something AI-related. I don't think that's going to change for anybody. But in terms of what I think the practical impact on the ecosystem will be, I think this is probably going to be a huge leveler.

44:43I mean, you figure OpenAI was probably a year ahead of anybody else, including some number of months ahead of the whole OpenAI source ecosystem. So if OpenAI is kind of unraveling, and then Sam and his team have to kind of put it all back together again and some new division of Microsoft, that's going to take some period of time. I don't know how long it's going to be, six months, a year, whatever. While that's happening, everyone else is going to catch up. So I think that the practical impact of all this is going to be a big leveling in the startup ecosystem. Obviously, this will be hugely beneficial for OpenAI's competitors, whether it's open source as a movement or maybe some of these other competitive foundation models.

45:28Obviously, it's probably going to be better for them. And then you figure there's a bunch of startups in the ecosystem

45:38that appear to have some kind of privileged access to open AI. I don't think there were a lot of them, but there were startups that were sort of anointed by open AI. I mean, I think open AI had a fund that was making investments in the ecosystem. And I think those companies got a lot of momentum and went on to do really high-priced future rounds based on the expectation that they were kind of first and would stay in the lead because of this, again, privileged access. And so if the rug's getting pulled out from OpenAI or it's just the whole thing's getting kind of level set now, then those startups will be impacted as well.

46:15And David, just building on that, like Chamath had a tweet earlier today, right, where he talked about, so the impact of small companies, you know, smaller companies building and startups building versus large companies and where large companies get caught up in their legal departments. Do you see now what's happened with OpenAI? Obviously, they were operating like a startup. Do you see them slowing down? And does that open opportunity for others to kind of come from the startup side of the world versus, you know, we know the larger companies are here. Google is here and obviously Microsoft's been there.

46:47But what are your thoughts on that? Well, it's a good question. I mean, I think that in terms of creating foundation models, that seems like a big company activity is just so expensive. Right. But the open source ecosystem has a lot of startup activity in it. And so, you know, if the open source ecosystem now gets further ahead because OpenAI is slowed down, maybe that could redound to the benefit of some startups that are playing in that ecosystem. I think it's really hard to see. I think for startups that were building on top of OpenAI, this probably slows them down. Because there's a whole product roadmap that was about to be built at OpenAI for developers, and that's been derailed now.

47:33At least from where we're sitting at this moment, and maybe this all kind of gets put back together. So I think that to the extent that you were competing with OpenAI, this is probably certainly a net positive for you. But if you're building on top of open AI, it's probably going to create some problems for you. At a minimum, you're going to want to diversify and build on top of multiple models, which people are doing anyway. So I think it kind of depends on where you're sitting in the ecosystem. Jacob, what do you think? I got my first email from one of our companies. It's called Roam Around.

48:07And they are API partners, like everybody else, with open AI. and they basically just told their investors, hey, we have other models we're using. If OpenAI collapses or we have problems, don't worry about it, basically. So that's the first time I saw founders who maybe had a dependency here or maybe people perceived to have a dependency then letting their investors know, don't worry, we've got escape hatches here. It's not gonna be a killer for us. I do think most people would say the technical advantage when I asked people on the pod that OpenAI had, most people would say six to 18 months lead time.

48:44So let's put it at 12. Then you have their momentum, which was people wanted to work there, Sam's cult of personality, the other leaders being a cult of personality. They had a lot of momentum and momentum in startups, in poker, in the entertainment industry. Momentum is very important. And the momentum they had was, if you're talented, that would be at the top of the list of places you want to go. If you're an investor, that's one of the places you wanted to own shares in. Momentum is so important in startups for so many reasons. And that momentum is now gone. And now it's the opposite. It's gone from momentum to being a risk factor.

49:22So the secondary that Thrive was going to do, that's off. That was a momentum-driven secondary. People wanting to pick, hey, who do I align myself with as a startup or a corporation? That momentum is gone. They're not the lead player anymore. They're the incapacitated player. They're the dangerous choice to make. So now Claude or Bard or Falcon or an open source project, whatever's on Hugging Face, that becomes the default. So this has been cataclysmic in the damage it's done to OpenAI's lead. That 18-month lead, I think now is zero and or a negative. And I think you can't understate that. Yeah, I totally agree.

50:03And you're right. And like part of the momentum they had was in hiring, right? I mean, they were slurping up so much of the tech talent. And AI engineer is the hardest job to hire for. And a lot of it was continuing to go to open AI. And now that momentum is going to slow down or stop. I mean, presumably, if Sam and his team set up a new entity at Microsoft, they will still attract a lot of talent to that. And the people who are loyal to Sam on that team are going to go. But if you're an engineer now, you have to really think about, is this such an obvious choice anymore? Okay, let's go to predictions here.

50:42Sonny, you've been doing a great job here moderating, but I want to hear your prediction. What happens in the next 72 hours? Everybody has basically said they're going to quit. I thought half the people would leave. It's like 90 % have said they're quitting. So that means OpenAI is dead. Microsoft says, hey, they have this option for everybody to go there. And then, of course, there is the reinstatement of Sam and the firing of this incompetent board. What do you think, Sandeep Madra, happens next? I think the company reemerges. And yeah, that's my prediction. I think - Open AI. Open AI. Sam and Greg and the other folks are back.

51:27Return of the King. Return of the King. and there is a real conversation around governance structure, nonprofit, profit, and all that becomes clear forever. I think that's where I go. Saks, prediction next 72 hours? I think there's what should happen and then there's what actually happens. And I think what should happen is probably about 50-50. I think what should happen is that OpenAI brings Sam back because, I mean, like you said, the whole company's basically now signed a petition saying they're going to leave if he's not brought back. So you bring him back. That means that the whole board's got to go.

52:04He's got to reconstitute it. I would go further and say they have to get rid of this whole Fakakta nonprofit structure that created all the problems to begin with, give Sam a proper comp package, get rid of all these crazy crisscrossing dotted lines and so forth, and just normalize the structure of this company. That's what should happen. What actually happens, I don't know, I think it's probably 50-50 at this point whether they bring Sam back or not. If they don't, I think we all know that it's going to implode. But again, with each passing day that they don't do the obvious thing, it's more likely that the obvious thing won't happen.

52:42I mean, look, I thought it was obvious by, I don't know, what was it, like Saturday or Sunday that they needed to bring Sam back because the whole company was groundswelling in favor of him. And then they did this like total non sequitur of hiring this like Emmett guy, like out of nowhere to be a CEO. So that was wild. That was wild. So it's like, wait, they didn't get the message? And David, I don't know if you saw this, but on Twitter, Benny Off, like everyone is out, like every AI. The sharks are out, blood in the water. So if you allow that transfer to happen to Microsoft, then, you know, pretty much you're doing that, you know, for money at that point, I guess, right?

53:23because you're going to try to get some comp package from them. But I think there's so many options for those folks that they're going to weigh those things. There's so many options. And now you're just kind of working for Microsoft. Is that really? I mean, the bloom is off the rose. The bloom is off the rose. So I'm not saying that they won't be able to reconstitute a pretty impressive team over there, but it's not going to be this undisputed leader. They're going to have to spend, I don't know, six months is getting back to where they were at least so this is the great leveler it's i think my my tldr is that this is gonna be a giant leveler in the whole space because the company that was running away with it is now you know hamstrung or imploded yeah that's correct sacks the entire poker table has been flipped the blackjack table got flipped chips drinks cards it's anybody's game now.

54:20It's literally if OpenAI implodes, anybody's game, anybody could win AI. And that actually might be better for the marketplace if you think about it. Jason, I got to ask both you guys, Jason and David, XAI, right? You guys are close. You guys understand it. Obviously, Elon started OpenAI, put the first money in. What does this mean for XAI from your guys' perspective? Zach, you go first. Look, I mean, I don't think XAI would exist if OpenAI, had done right by Elon in the first place. He did put in the first$40 million and he ended up with no shares. Now, I don't know that that was necessarily nefarious.

54:58Again, I think it was a result of this crazy non-profit structure they created. But I think that as part of correcting this mistake they made on Friday, I think they should undo all the shenanigans. Let's fix all the shenanigans that happened here. Again, normalize the structure. I think Elon should get shares for his $40 million that went in. That's what should have happened. It should have happened a while ago. And Sam should get a package. Yeah, but Sam should get a package, for sure. That's what should have happened. They should have cleaned this stuff up a long time ago. But you know what happens is no one ever wants to clean anything up.

55:36They just want to keep papering on top of it. They papered a bunch of crap on top of the original mistakes and now it's come back to bite them in the ass. But look, in terms of what does this mean for x.ai i mean so that that company's already been created now it's off to the races and so the genies are the bottle and i mean it's gonna be good for them right it's gotta be good for them yeah it felt to me sonny when we played with it that they were six months behind maybe maybe argue nine something like that and now if this continues in chaos it's going to take six to twelve months for all the dust to settle there's going to be so many lawsuits.

56:16Every employee who didn't get to sell in the secondary offering is going to be able to sue the board. Sam is going to be able to sue the board. Microsoft is going to be able to sue the board. The investors who invested in this are going to be able to sue everybody. So the number of lawsuits that will come out of this for lost money, remember, they raised$300 or$400 million from investors other than Microsoft. This is going to be... And then Imagine the discovery. Imagine how much chaos that will cause. It'll be a full employment program for lawyers, for sure. But Jason, play this out a couple of moves beyond that, right?

56:51So now you've got this, assume like Sam doesn't come back, right? And so you've got 90 % of the employees have left. And you've got this shell entity that's running OpenAI, but there's no one there. So they're barely keeping the lights on. And then meanwhile, there's all these lawsuits or threats of lawsuits, and they're getting all sorts of legal letters. What do you think that management does at that point? I think they just kind of sell it all to Microsoft. Yeah, they quit. Sell it for parts, shut it down. Versace to be their white knight just to clean up the whole mess. So then Microsoft just acquires whatever's left, which would be substantial.

57:24I mean, you're talking about the source code here to OpenAI, which is doing over a billion in revenue and it's got thousands of... No, they've got over a million developers building on top of it. i mean this is a massive asset huge asset and it you know here's the other thing sex you know this if this was going to go that route they would need to their fiduciary which using the word fiduciary with these idiots i'm just call it what it is is you know um i don't think they know how to spell the word um they would need to run a process and they would need to let five or six people bid on it and then what are those people bidding on the source code that microsoft also has access to, but not AGI.

58:03The Franken structure of this makes it really hard for people to bid on it. But you know, Benioff might come, Google might come, Apple might come and want to bid on this. Zuck's going to bid on it. So now you have to go with the whoever bids the most in order to get the shareholders who've been wrong their money. This is so chaotic. There is no way to run a process because who assumes the liability? And I guarantee you, the IRS is firing off massive investigations into what happened here because they don't like when people run circles or make them look foolish and i think this is starting to make them look foolish non-profit generates 190 billion dollars in equity value then there's an auction for the assets it's insanity well i think the way it was structured and there are there are versions of the corporate org chart that have been floating around you know x is that you've got this foundation that is a non-profit and then it somehow controls an LLC.

58:59And that LLC is a for-profit entity, but whose membership interest is kind of controlled by the foundation. And then the investors came into that LLC, and Microsoft had some sort of cap profit arrangement with that LLC. So I guess what would happen is that if you could sell off the parts, that the LLC is the entity that would be sold. So I guess either you just sell the LLC as a whole and everyone sells their membership interests, or the LLC sells all the assets, okay, and then distributes the cash to basically dissolve, right? So I think that's probably what you do is you have the LLC sell all the assets, take in a bunch of cash, and then you distribute all the cash to the membership interests of the LLC and wind it down.

59:48And then what happened is you end up with this foundation, the OpenAI Foundation, that's got a bunch of cash sitting in it. And it's a non-profit controlled by, I don't know, whoever stole and controlled this board, I guess, would control all that money. That is a big pot of money there. This is a, what an absolute, utter mess.