Emergency podcast: chaos at OpenAI

26:31So I always had it in the back of my head. Like at some point, this has to hit some kind of resolution because Microsoft will not demand it, but will just create a resolution to say to its shareholders. So there's all this like vetting of these like old heads. The OGs come to town, right? So the vibe I'm getting, and again, we don't actually know if everyone was all together. We know there was a lot of opening high people at that headquarters. We don't know if the board was there, actually. But the vibe we're getting is people are firing names at this board, and the board is not taking it seriously.

27:06And in the meantime, they are running their own search for a new CEO because their interim CEO, Mira Morati, has sided with Sam in the meantime. like publicly during the hearts campaign on twitter she's posting the heart right so she's what a sentence she's gone over i mean it is just absolutely childish when i say these are very flawed human people like we're gonna win this fight by posting hearts on twitter is i don't know what to say about it one day i will like have had enough time to process that situation but mira has gone over to team sam like very publicly so the board needs a new ceo so that they're getting tossed these names and everyone is hoping that they will accept some names and resign and the new names will take over.

27:47And in the meantime, they are looking for a new CEO. And that is more or less what is happening all day Sunday as the 5 p.m. deadline draws ever near. So we hit that deadline. And again, nothing. Alex, what would you have? Where was your head at sort of the end of Sunday? Like, obviously, things get crazy several hours after that deadline. But like, where were we at the end of that day on Sunday? Do you think? I was feeling like if we didn't have an announcement by five, it wasn't going to work out. And we were kind of getting back channel that by midday Sunday, things were taking a turn. And I didn't obviously no one could have expected what actually happened.

28:28This is the most like I was saying at the top bananas thing. But yeah, I think people were generally people who were close to the story were thinking like, okay, he made it very clear publicly. this is the last time he's setting foot in the building. The deadline is passed and it's not looking good. And there's a version of that that would have been a very dramatic weekend that ended in a relatively okay thing, right? Sam and Greg go off to do something. They bring some OpenAI people with them. OpenAI hires a new CEO. Everybody moves on with their lives. There's a version of this where a lot of people had a lot of feelings, but it turns into a relatively normal corporate change, right?

29:08But that is obviously not what happened. Right. The negotiation to bring Sam Altman back has failed. Mira Morati continues his interim CEO. Microsoft wishes Sam the best and says they'll support him in the new venture. Again, the thing that needed to happen by Monday morning was a Microsoft statement to the market. And you really just cannot underestimate how much pressure that was applying to the situation. So I'm expecting that statement, right? Like, here's the forcing function. We just have to be ready for this deal is going to fall apart. Microsoft is going to issue some holding statement and say, you know, we have a deal with OpenAI.

29:46Everything's fine. Our contracts are rock solid. Brad Smith, our chief legal officer is a great lawyer. Like, whatever Microsoft is going to say to calm everybody down. But instead, and all credit to Bloomberg, Bloomberg has a report, I don't know, minutes before the thing actually happens, that Mira Murati has hatched a plan as interim CEO to just rehire Sam and Greg as employees while the board is out calling people to interview them to be the CEO to replace Mira. And so what comes out in the end, I'll just like fast forward a little bit, is Mira's plan was to quickly hire Sam and Greg as employees again, forcing the board to fire all three of them, which would have led to presumably lawsuits.

30:33Like, who knows what was going to happen in that moment? But that was the last swing of chaos when it was, I think, obvious, oh, this just fell apart. Like, we're not in a place where we're negotiating four people resigning. We're in a place where we're actively trying to create legal leverage for a lawsuit to come. and the board is actively trying to replace the person that they just hired to replace the ceo they fired sorry it's just it's i know it's just like this whole thing has been such a blur for nila and i like when you say all this out loud it's just truly it's insane and it does seem like i think your instinct there's right now that it's at that point what you're saying is i'm not gonna quit you have to fire me yep like i'm i'm going to cause so much trouble for you that you're gonna to have to get rid of me and then i'm gonna have ammo kind of in whatever direction i want to use it yeah exactly right and the rehiring of sam and greg is like a deeply funny idea like you have to fire them again it's just like just like hire them as interns and watch them get fired yeah and all three of them will then sue the board and this is when alex and i just started sending out hundreds of text messages this is about to fall apart like this is absolutely about to fall apart and it fell apart and it fell apart in the weirdest way possible, which is the board just didn't even announce the information broke.

31:57Yeah. This is just like coming from sources. Like the board has been in a bunker somewhere. They don't have a crisis comms team. They don't have like people speaking to the media. They've just been radio silent, but it starts to trickle out that they've actually named a new CEO and it's Emmett Shear. Not on my bingo card for who was going to be the next CEO of OpenAI? Not on anyone's bingo card. I think it's very safe to call this one out of left field. Emmett was the co-founder of Twitch, which is a live streaming video site, not an AI company. He's not seen as a AI leader. He posted shortly after it leaked that he was being named CEO that he got the call for the job that day and took a few hours to decide.

32:43And, you know, he's a free agent. He left Twitch earlier this year before mass layoffs. They've had two rounds of layoffs since. I can confidently say that the vibe within Amazon is that Twitch has been a, how should I put this, shit show since Amazon bought it. And so no one really thought of Emmett except the board, apparently. He is now the CEO. And he sent this note internally to employees saying that he was going to conduct an independent review of the board's actions, which is hilarious because he was hired by the board. I don't think that's possible to have an independent review when you are the person you're represented by the people that you're reviewing.

33:27I mean, it just doesn't make sense. And also those people can callously fire you whenever they want, which they've proven twice in 48 hours that they can do. Okay. And we've been dancing around this this whole time. I really think at this point we should just say who these people are. Because at the end of the day, three people set all this in motion. Three people have caused OpenAI to explode from within. So should we just get into that? Yeah, do it. Yeah. Okay. So OpenAI is very strange in how it's structured. And this is something that people haven't really been paying attention to because we've been so focused on just the success of their products with ChatGPT.

34:01But OpenAI started as a nonprofit. And so it has this weird structure where there's a nonprofit. There's a flowchart, which is like perfect for Decoder. There's a flowchart on OpenAI's website. And I challenge you to look at this flowchart and try to make sense of it. It is one of the most confusing. I just want to be very clear. You can definitely make sense of it. And they've drawn it to be more confusing than it actually is. Okay. Well, that is 100 % true. We're going to redraw the flow chart now because it's all different. So there's this nonprofit with this board that controls it. And importantly, the board of OpenAI does not have equity in OpenAI, which is a really wild thing.

34:38And so that's the context of the board. There's three of them really that aren't OpenAI people. So one of them is Adam D 'Angelo, who is the CEO of Quora. He operates, I should note, a competing AI chatbot platform called Poe. He was the original CTO of Facebook. He's a known quantity in Silicon Valley. He's a really well-connected guy. There's a woman named Helen Toner who has ties to the effective altruism movement. She used to work at Open Philanthropy. She's now at Georgetown. She funds AI safety stuff. She was actually on our stage at Code in September with Casey Newton talking about AI safety.

35:14And then the other one is a woman named Tasha McCauley, who is the former CEO of GeoSymSystems. And as far as I can tell, basically no one in tech that I know knows who she is. Okay, so that's the three people. And they are the ones who basically decide to bring Emmett in. Because as we find out in the next turn of the story, the guy that we thought was the architect, the mastermind of all of this, flipped, which we can get into. Yeah, so Emmett's a weird choice. Emmett's a weird choice. and also openly a decelerationist, which is a phrase that the AI community loves to use. There's videos of him being like, this is terrifying.

35:54We should slow it down. This will extinguish all value in the cone of light, which is a real thing he says. Wow. It's amazing. And his point of view is we should slow AI innovation way down so we can get a handle on how dangerous it is. So you can see where the board is, right? This is a split that we've been hearing about the entire time. OpenAI is playing with dangerous toys and Sam is running too fast. So we brought in Emmett Shear to slow this whole thing way down. Yeah, I think that's right. And then on the flip side, very early this morning, Neil, I think to your point correctly to get something out before the stock markets opened, Microsoft announces it has hired Sam Altman and Greg Brockman to run, I believe the phrase is an AI research lab.

36:41An advanced AI research team is their statement. And can I just read this statement from Satya Nadella? Please do. Which just contains magnitudes. We remain committed to our partnership with OpenAI and have confidence in our product roadmap, our ability to continue to innovate with everything we've announced at Microsoft Ignite and in continuing to support our customers and partners. We look forward to getting to know Emmett Shear and OAI's new leadership team and working with them. And we are extremely excited to share the news that Sam Altman and Greg Rockman together with colleagues will be joining Microsoft to lead a new advanced AI research team.

37:14We look forward to moving quickly to provide them with the resources needed for their success. In the journalism business, we call that burying the lead. And importantly, Sam Altman is being given the title of CEO at Microsoft, which is inside Microsoft's corporate politics, like a big deal, right? There are not a lot of CEOs at Microsoft. Usually when they acquire a big company, they give those people the CEO title. So the person who leads LinkedIn has a CEO title. The person who leads GitHub has a CEO title. Phil Spencer is now the CEO of Microsoft Gaming. That's a big deal. You can go listen to that Decoder episode where I ask him what that title shift means.

37:53And it basically means he has his own resources, right? He's split off from Microsoft. Your own P &L. Well, I don't know. You can go listen to that episode. It basically means you have your own resources and you are more of a free agent to run your little division like it's their own little company. So this is the arrangement Sam is getting. What I will tell you is I read this statement, especially that we look forward to getting to know Emmett Shear. They don't know. They don't know these people. They don't know what's going to happen with OpenAI. They don't know if Emmett's going to turn the pace of innovation way down, but they are able to tell the market, hey, the face of the AI winning that we've been doing as Microsoft now works at Microsoft.

38:31Does Sam Altman have a contract to work at Microsoft yet? Like, I don't know the answer to that question. Does Sam Altman, the guy who used to run Y Combinator, who has his hands in every startup in the universe, has multiple investment funds, thinks of himself as the guy who was running the hottest AI startup in the world on its way to making AGI. Does he want to be a Microsoft employee? Like, I truly do not know the answer to that question. I do know that this statement utterly worked to not only calm the market, but to send Microsoft stocks skyrocketing. And we are now sitting here saying, do we need to pre-write Microsoft is now a$3 trillion company?

39:09Because the stock, as we are speaking, is like to the moon. So this statement worked. I'm just cautioning everyone. This, in my view, is a holding statement. It's still a holding statement. Well, I think knowing what we know about Sunday, knowing that Microsoft really wanted to have this buttoned up by markets open, this was decided like it was like almost like 1 a.m. Pacific. Yeah, I was asleep. Like straight up. I was like, I'm done now. I'm thinking that like, OK, surely nothing more is to come. And then like around one, Nadal tweets this. And luckily, Tom Warren is waking up in London and gets it on the site.

39:44It's nuts. But but I think there's an important thing here, which is we know that Microsoft wanted to get this done. I have to imagine that whatever got Sam and Greg to agree to go to Microsoft was a lot. I imagine they had all the leverage in that situation because they'll just go do their own thing, right? Like we've all been saying. Microsoft cannot make it look like the company that they have literally bet their AI Azure future on is imploding before their very eyes. and so I have to imagine this is going to go down as one of the best packages ever from a big tech company to to a team to come there because they had all the leverage that night I agree so this just look like this brings us to winners and losers wait we're not done with the story yet we're not there's one more turn but we're going to get to that but first we're going to take a break we'll be right back

41:00Thank you. With just a few prompts, you can turn any idea into a code-backed prototype or app that anyone can explore, iterate on, and share. And it's all connected to Figma. So you can start from scratch or from your design file. Figma Make is made for designers, product managers, and builders who want to move fast and explore more ideas. Try it now and see how quickly you can make your ideas real with Figma Make. Just go to Figma.com slash Vergecast. That's F-I-G-M-A dot com slash Vergecast.

42:01Try Trileca by 1Password. Trileca by 1Password provides a complete solution for SaaS access governance. And it's just one of the ways that extended access management helps teams strengthen compliance and security. 1Password's award-winning password manager is trusted by millions of users in over 150 ,000 businesses from IBM to Slack. Plus, 1Password is ISO 27001 certified with regular third-party audits and the industry's largest bug bounty. Take the first step to better security for your team by securing credentials and protecting every application, even unmanaged shadow IT. Learn more at 1Password.com slash VergeCast.

42:46That's the number one password dot com slash VergeCast.

42:57all right we're back and we are back with the last and current turn of the story which is a letter so much of this all is just happening people writing texts to each other it makes me very happy nila since you're our quote reader for the day do you want do you want to read part of this letter yeah it's truly spectacular from a bunch of open ai employees they they wrote an open letter to the board keep in mind this all starts with you know a 5 p.m deadline which is there's going to be mass resignations unless sam comes back they've blown the deadline twice and sam has now announced to go to microsoft the markets think microsoft has conducted a uh the greatest act will hire in tech history with no regulatory oversight like i have a version of that idea from like 5 ,000 people in my text message inbox.

43:46Like, Satya Nadella is a genius. Okay, what no one is counting on is like, oh, people still want to do the plan, right? Like, if you say you either hire Sam or we all resign, a lot of people are going to try to resign when you don't hire Sam back. So 500 plus employees out of, what, 700 employees? Yeah. I mean, by the time this episode comes out, it will have been the whole company. The letter says, we, the undersigned mate, choose to resign from OpenAI and join the newly announced Microsoft subsidiary run by Sam Altman and Greg Brockman, Microsoft has assured us that there are positions for all OpenAI employees at this new subsidiary should we choose to join.

44:21We will take this step imminently unless all current board members resign and the board appoints two new lead independent directors, such as Brett Taylor and Will Hurd, and reinstates Sam Altman and Greg Brockman. Importantly, Ilya has signed this letter, and he has tweeted, I deeply regret my participation in the board's actions. I never intended to harm open AI. I love everything we've built together and I will do everything I can to reunite the company. Mira has signed this letter and Sam has now quote tweeted Ilya with three heart emojis, which I just saying this, I want to point out, we are talking about 10 to$80 billion in shareholder value is happening with heart emojis on Twitter.

45:03It's the most emo thing. I've listened to Letter to Elise like 10 times this weekend because when I was in college, I would just use the lyrics to Letter to Elise by the Cure as my aim away message. That's where we are emotionally with this situation. This means the board might succumb to this pressure campaign because they're going to lose their entire company. All these employees are tweeting OpenAI is nothing without its people. The board might succumb. They might reinstate Sam. This whole Sam is now the CEO of a Microsoft research team might not come to pass. I truly do not know what happens next.

45:34And that's what I mean by Microsoft needed some clarity and they got it today in a move, which maybe seems like Nadal is a genius, but also may have just created enough stability for the next chapter to play out. So, Alex, there are two things in here that jump out to me that I'm really curious about. One, this means Microsoft has made very clear, very loudly that anyone who wants to leave OpenAI can go work for Microsoft, which is a pretty ruthless thing to do to a company that you've been saying we, you know, are committed to our partnership and look forward to getting to know the new leadership team.

46:06And the other thing is, this is the final form of the kind of the board against the world thing. This is not, quote, tweets and heart emojis. This is literally the vast majority of the company saying in no uncertain terms, we will leave unless you do this. I'm always of the mind that it's one thing to make big noises about quitting your job. It's another thing entirely to quit your job. This does feel like the board is going to have to call a lot of bluffs in a way that it's probably going to lose at this moment in time, right? Yeah. I mean, what I was hearing over the weekend is that Ilya is really the wild card here.

46:41He has a lot of respect internally. He's a co-founder. He runs the research teams. the fact that he flipped and is now saying, I'm also going to go to Microsoft means there really is no open AI anymore. Because that was always the key man is like, if Ilya is still there, there's still a chance that, you know, they're not going to be this like breakneck commercial entity that they have been under Altman, but they at least will do leading research because the research community respects him. If he is going, it is literally Adam D 'Angelo, Tasha McCauley and Helen Toner those people are going to be all that's left of open AI yeah he was kind of the truest believer right like if you wanted to pick somebody who was like the most held on to the original vision of open AI you'd probably pick Ilya right so the Atlantic has this incredible in-depth story about open AI that everyone who's listening to this should go read and there is a scene uh where they had a leadership off-site I think it was last year and Ilya has a wooden effigy that he literally sets on fire at the meeting to represent non-aligned AGI, which is, you know, this is like the idea that, you know, they're creating AGI in an unsafe way.

47:51I mean, it's kind of a religious thing in more ways than I think people get in the AI community. But yeah, that's, that's, I think for you, if you want to understand kind of the cultural divide, which is maybe where we go next here, but like the culture divide within OpenAI has become somewhat religious in the last year or so as the company's become commercialized. I told Alex that a real headline we should write after this is the rich people are crazy. Because this is religious fervor now. These are people who believe that they are going to create the power to destroy the universe. And that is how they talk about it.

48:26It's not like an imposition of class warfare on billionaires. It's, no, you should listen to the billionaires talk. And you should listen to what they call effective acceleration and deceleration. And there is a schism there that is beyond cultural, right? It has reached a level of religious fervor that is making them act irrationally. And that might be appropriate. Like if you thought that the thing you were building could destroy the world, you should probably reach into some like principles that are not just made up, which is like how you find religion. But that's like there's no moderating force on that.

49:08Right. There's no like centuries of church dogma or liturgy, like whatever. There's some forum posts about effective accelerationism that they're all reading that they've just made up. And like, that's weird. And it's worth some scrutiny outside of the sort of boardroom or chart battles that, you know, that's where I like to focus my attention. But there's something here that is making these people act irrationally and with not just like religious fervor, but with like martyrdom fervor. Like they will kill the company because they believe this is so right. Right. And what's great about this story is that this tension is actually personified in the org chart.

49:47So going back to how ridiculous it is that the way that OpenAI is structured, they have this nonprofit board that controls the company effectively. They can fire people, etc. They get to decide when the company has reached AGI, which is a very important power to place in a company that's whole mission is around creating AGI. And so you have that one side of the equation. And on the other side, you have the for-profit entity that OpenAI created under Sam Altman to raise all of the money from Microsoft, hire all of the top talent from all of the big tech companies that they have in the last year to build wildly successful commercial products.

50:26And Sam Altman is on stage at Dev Day. I was there a couple weeks ago saying, we are going to be the app store for AI. We are going to start sharing our revenue with you. We are going to give more of our technology to our developers. We are going to be a platform. So inherent in the structure is this tension where the commercial entity is charging ahead with something that the board is tasked with protecting and ultimately deciding. And incentives drive behavior here. And so it's just really interesting to look at like, maybe this was always going to happen. You know, the favorite theory I've heard over the weekend was like, this is actually Elon Musk's poison chalice because he named OpenAI.

51:07He really set it up, gave it its first donation, set up the nonprofit structure. And maybe this was always going to happen. Well, and then importantly, stopped giving that money when he stopped being part of the company. It was a big part of like Semaphore's reporting from a while ago. Right. That was part of when OpenAI realized we need a lot more money to do the kinds of things we're doing. They went to Microsoft to get both the capital and the compute to do it. And all of a sudden, you're just kind of off to the races. And as soon as you decide you want the money, it becomes very hard to not want the money anymore.

51:40That's one of the things we've seen over and over. And like this idea that you can have a nonprofit overseeing a for-profit company is not unheard of in tech. It's fairly unusual. But like Mozilla is a version of this kind of org chart. Signal is a version of this kind of org chart. Like it's possible to do, but it comes with a really complicated set of trade-offs in every case. Like Mozilla's thing, and Nila, I know you and I have both talked to Mitchell Baker, the CEO of Mozilla, about this. Mozilla's for-profit company gets most of its money from Google, which to Mozilla, the nonprofit that wants to make the internet better, feels really bad.

52:12But without that money, you can't do the work anymore. And so this tension is never going to stop going away, or this tension is never going to go away. And so you eventually need somebody in charge who is either so mission driven that they stay that way no matter what, or you get into the Sam Altman thing where you eventually start chasing the money and then the money gets big and then that's what you become. So I said this earlier, and I'm not going to try to describe an org chart on a radio show, but it is very true that the flow chart that OpenAI has published about its ownership structure is consciously more complicated than it needs to be.

52:52There's only three boxes on the structure. Most of the extra boxes are just labeling who owns what. So there's OpenAI, the public charity of 501c3 that owns a holding company, which co-owns OpenAI, the commercial entity with Microsoft. That's it. That's the whole structure of this thing. There's a charity that owns a holding company so that employees can have some equity on the holding company. And then there's the commercial thing that goes off and does all the work. That structure exists all over the place. It does not exist in Silicon Valley. So outside of Mozilla and these other things that are run is, you know, they're sort of like not capitalistic in their way.

53:32Right. In Silicon Valley, everything is like a Delaware organized C Corp with VCs on the board and now all over the place, super voting shares for founders. Like there's a way to do it. And they do that because this stuff happens when you don't. Like that's why they do it. Right. Like founder drama with their VCs. Like the story of Uber exists. Right. The story of Apple exists. Like, do you know every single year Meta shareholders vote to remove Mark Zuckerberg? Yeah. Yeah. And he just outvotes them because he has super voting shares. Like, it just doesn't matter that the shareholders of Meta are like, get this guy out of here year after year because he has super voting shares.

54:09Right. So you get a structure like this and these shenanigans become possible. But then you look around the world and you see functional structures like this at very big companies, at very important companies. Someone pointed out to me that Bosch, which is a huge German engineering firm, very successful in the car world, in the engineering world, is organized basically like this. There are companies that are owned by the towns they're in in Europe. All these other structures exist. In this org chart, people look at it and you look at it and you're like, this is crazy. If you just spend a few seconds looking at what these boxes mean, and I don't know why some of them are ovals and some of them are squares.

54:48It's just three boxes. Yeah. Right. At the end of the day, there's OpenAI, the charity that owns a holding company that owns OpenAI Global, which is the commercial entity. And Microsoft is a big investor in that entity. And that means Microsoft does not have a lot of control over this board, which I think has become a real problem for Microsoft at this point in time. It means that Sam Altman did not have super voting shares to stop it. But it also very importantly means that the board was too small and too inexperienced, made a hasty decision, and there was no governance check on their behavior.

55:21And I think that is like the fundamental mistake of the structure, not the structure itself. Or we find out that – and I don't think this is the case, but I just want to play devil's advocate. Or we find out that like OpenAI invented God in the last two weeks. There's this clip of Sam Altman at a conference in San Francisco the day before he was fired saying that there's been a few moments in his career where they've gotten to push back the veil of ignorance. And he was talking as if they had made fire. And he said one of these moments happened in the last couple of weeks. So everyone is reposting that clip saying what if they realized they had AGI or what if they realized the model was about to just accelerate in a way that no one could have predicted.

56:03I will say that Ilya has been doing a press tour and being pretty open that the model capability is going to progress a lot faster than everyone thinks. So there has been a lot of theory to that. But what it looks like right now is inexperience and the board not realizing what was going to happen when they made such a rash decision. And I just want to put this out there. The other thing I've heard is it's not that Microsoft and the commercial entity of OpenAI and Sam didn't think safety was a problem. Right. Like Kevin Scott, the head of Microsoft, was on stage with me at code and we talked about AI safety and the limits and like how careful they should be.

56:44This is a thing that Microsoft is publicly talking about. So if OpenAI's board is saying, okay, we're going too far ahead, their entire job is communicating that there's a misalignment. Their entire job is saying to Sam, hey, you're over the line. Like, pull it back or we're going to fire you. Instead, they just fired him. Instead, they just went ahead. They gave Microsoft one minute notice that they're going to fire Sam. And so the fundamental job of this board is to communicate and set limits, and they just didn't do it. Yeah. And this is another good moment to put a button on what you said earlier, Neal, which is that we actually don't know the whole story behind why he was fired in the first place.

57:26And Emmett Shear, in his weird, long tweet announcing that he is CEO, at the very end says, PPS, before I took the job, I checked on the reasoning behind the change. The board did not remove Sam over any specific disagreement on safety. Their reasoning was completely different from that. I'm not crazy enough to take this job without board support for commercializing our awesome models. So before that, I'm like, okay, OpenAI wants to go back to being essentially a research organization. Fine. Like, whatever you want to say about that decision, that is at least a decision. Now it's like, okay, well, then what are we doing here?

57:57Like, what fight are we having here? I'm sure there are people who know, but that part does not seem to have come out in any way that I find at all sort of satisfying. What I think we will find out is that this is a classic business dispute. I mean, yes, there are obvious concerns on different sides of the debate here about AI safety. And I do think that played a role. But what we just saw was a couple of weeks ago, OpenAI hastily, and I can say this confidently because I was part of the press scrum and getting pre-briefed and all this about the App Store stuff. Like very hastily, OpenAI announced a major business expansion that Sam Altman led, right?

58:38They had this huge conference and were like, we are pushing ahead. We're going to be a new platform. And if I had to guess, that probably led to a lot of this because Sam was clearly caught by surprise. There was nothing, you know, normally if this had been like a long gesticulating thing over months, it would have leaked in dribs and drabs. And it really just came like a thief in the night. And so there was a breaking point here very recently. And the thing I can point to is Sam getting on stage and saying, guess what? We've been going 100 miles an hour. We're going to go 200 miles an hour and we're going to become an even bigger business.

59:15And what seems clear now is that a bunch of people at Microsoft, including Satya Nadella, saw that and said, hell yeah. And at least one person inside of OpenAI said, hell no. Yeah. I mean, this is like, essentially, this is capitalism versus hippies. I mean, that's what this story is. It's the story of Silicon Valley, man. Yeah. I think that's too reductive. Again, the religious aspect of this is just not to be underestimated. Microsoft sees this and says, yeah, go even faster because every one of these queries runs on an Azure server. Yeah, right. Right? That's great for them. They're ready to go.

59:49The OpenAI board sees this and says, OK, you're going to give people the ability to make computers go take actions in the real world without any oversight and let them build whatever AI tools on our model that they want and resell them and create a commercial incentive to do that without any safety checks whatsoever. And that is opposed to our mission. That is like real. That is a moral dilemma. Do you remember? I mean, we were just talking about this on the Vergecast. We were talking about how this store and this agent API thing is going to open them up to a world of hurt. I mean, they literally had to say on stage, we will indemnify you for any copyright violation that may occur from creating bots on our platform.

1:00:30It was very – and like I went to a Q &A with Sam after they announced this at Dev Day, and it was clear that they had not thought through the details, right? Like they hadn't decided on how much revenue they were going to share with creators, how the store incentives were going to be aligned. It was all just being done kind of on the fly. And we were all going like, this sounds like a disaster. Like, it's a smart business play. If you are Sam Altman, the venture capitalist, right? It's like, yes, this is what you want to go do. You want to build a moat around, you know, the fastest growing consumer product of all time.

1:01:00You want to be a platform. But it was clear that they were moving a million miles a minute. Yeah. And also, the problems are not hard to imagine. Okay, a creator sets up a GPT bot that just spams people in their email with marketing messages. Does OpenAI want to be sharing revenue with that creator? Is that a thing you want to incentivize? You can make it even worse. It's connected to Dolly, right? Okay. You want to make deep fake nudes of people and that's an app you want to sell in the GPT store. Does OpenAI have the content moderation ability to stop it at scale when millions of people realize this is a thing they should try to sell?

1:01:38The most wild thing I heard at DevDay was that OpenAI uses GPT to moderate GPT. It uses the AI to moderate the AI. And that's how they were going to approach the store. Yeah, I mean, it's truly bananas. Yeah. So I think there's a lot still to sort out here, right? I think including the question of like, does Sam Altman actually ever work for Microsoft? And I think we're going to talk about this a lot for a long time, I suspect. But let's just put a button on this moment for now, because if we stay on here too long, more news will break and this will all be out of date. So we need to like put this thing on the internet.

1:02:14Just five minutes. I want to do real quick who won and who lost over the last four days. And then we're going to get out of here. I'm just going to throw some names at you and you guys are going to tell me winners and losers. Sound good? I mean, it sounds like an impossible exercise. It's going to be great. I'm excited about it. Here we go. Person number one, Satya Nadella. I feel like big winner of the weekend, right? He might be the only winner. Okay. Everyone thinks he looks very clever today. I do not know long term if having your stock price swing this much on the actions of one or two people is a good thing for Sachin Adela.

1:02:45Fair. Okay. Person number two, Sam Altman. I genuinely can't decide. Loser. Huge loser. Really? Yeah. Without question. Huge loser. He was the guy. And now he might still be. He might start a new company. He might go work for Microsoft. He might take everyone with him. but losing the thing you founded is the worst. And it might be that he returns to open AI in 12 years like Steve jobs. But in the meantime, right. The most likely outcome is he owns next. Is this his Elon Musk style Joker origin story? Like, is that, yeah, I don't, I, there's no, I come back to the very beginning. It's great to think that all these people are, are cogs in a machine and it's a narrative that's playing out.

1:03:26And it's like, you know, people write about it that way. Sometimes we write about that way. These are people, you know, and like whatever he does next is going to be deeply colored by just the emotional damage of this weekend in the sense that he can't trust the people he thought he could trust. And maybe that's just the reality of being the guy who ran Y Combinator for a long time. Maybe he was already like that. You know, he's been in the scene for a long time. He knows a lot of people. But to go from the top of the world to Microsoft employee in 48 hours, no matter what happens next, he is not the winner today.

1:03:57I think he is more a winner than Eli, but I don't think he's a full winner. I mean, I think like Sam, like I was saying, looks like he had all the leverage with Microsoft. Microsoft needed a way to message that this was under control. I imagine he extracted some insane deal terms out of Nadella late last Sunday night. and if he does get the entire OpenAI team to come to Microsoft and he gets to run it like his own little fiefdom, like LinkedIn has its own email, right? Like it's its own building, its own. Do you think he gets to use Google Meet? That's well, maybe, you know, that may have been the biggest concession of all was that Nadella agreed that they don't have to use Teams internally.

1:04:37We'll find that out. I'm sure Tom is on that. But I mean, I'm thinking to why they created the commercial entity. It was because they needed funding for compute to train their models and Microsoft showed up with a bag mostly of cloud credits, interestingly, saying, we will give you all the compute you need. Microsoft has gone on what I've seen some analysts saying is the most expensive private company infra build out of all time with its AI data centers, something like 50 plus billion dollars in the last few years. So if Sam gets the insulation of being an independent company within a mega cap company and all of the resources that he was desperately trying to raise money for for the last few years anyway, and all the people from the company.

1:05:22I mean, it seems like he wins. I mean, you know, like he seems like a winner to me. Certainly this is like for his like, yeah, trusting people, not great, but I think he's more a winner than Eli. All right. I have a bunch more, but I actually think I think you're right. Most of them are pretty easy. Open AI, obvious loser. Google, winner or loser? Google is the most vindicated of any company right now, right? They got so much heat for being slow and not rushing the stuff to market. And now they're vindicated because OpenAI itself just tore itself apart over the idea of going a little bit slower.

1:05:58And Microsoft, which is, recall, Nadella on Decoder, I want to make Google dance. I want them to know that it was me who made them dance. incredible quote is now having to start a new division, set it up, give it research funding while navigating meeting Emmett Shearer, which is literally in the statement, they have to meet this guy. Like if you're Google, you're like, wow, we were right to go slowly and carefully and our moment will come to us. Also, we have better distribution for our AI search products than anybody in history. Like maybe we'll just continue to steamroll the industry. I don't know that's going to happen.

1:06:34I don't know if Google can actually execute many, many questions. But if you are Sundar Pichai today, you're like, phew, missed that one. You're just fine. I think that's right. I think this is the most aggressive wrench of instability into its biggest competitor that Google could have possibly hoped for. Bard still sucks, though. So there's that. Anthropic. Alex, you brought this up earlier. Is the team at Anthropic just giggling and high-fiving right now? I mean, yeah. Anthropic was created for a similar reason. I don't know if there was a coup that was failed and that's why Anthropic got formed, but they all worked at OpenAI.

1:07:10They were the kind of more decelerationist, effective altruism movement. They broke off as a separate faction, created another AI lab, and they haven't had the commercial success of OpenAI. I think it's worth noting that the only AI startup that has been successful commercially is OpenAI. Like there's Claude, there's all these chatbots, like their usage pales in comparison. So yeah, I think they are winning because of the instability. Like maybe they'll be able to hire some of these people. But I really think everyone lost here except in the short term, Nadella and Sam Altman. I really do. In the short term, every Microsoft employee is thinking about selling their stock to buy a house.

1:07:49Yeah. I've seen multiple tweets of just the screenshot of the stock price saying, thank you, Satya Nadella. Yeah. All those people won. Yep. Did anybody else win? Remains to be seen. Fair enough. Okay, last one, and then we should get out of here. Chat GPT. Is this the end of the Chat GPT era? Is it even the end of the story? I don't know. Okay, fair. If we play this out in what seems like the most likely scenario right now, which is that OpenAI becomes significantly kneecapped in some meaningful way, it seems unlikely to me that all of the stuff that we saw two weeks ago at Dev Day comes out the way that we thought it was going to.

1:08:25It seems possible that this organization goes back to being more of a research organization. And ChatGPT, the fastest growing consumer internet product in history, becomes a sideshow for this company again. I mean, if there's no one there to operate it. Yeah, there's certainly that. I mean, that's where we are right now is like there may not be an open AI by the end of the week, right? Yeah. Like ChatGPT probably over, which is wild, but like they'll just recreate it inside Microsoft. And it won't be as cool and it won't be as like, you know, break neck and, you know, move fast break things. You know, Microsoft is like a, you know, regulated monopoly.

1:09:01They're going to move very slow and thoughtfully on like how this is rolled out because they have to. Right. So, yeah, I do think this marks this decidedly marks the move fast, break things era of generative AI. I think that's pretty safe to say. I think so, too. Without question. The like craziest rise and fall of an app I can think of. Unless Microsoft just takes it. Like, I do not know what happens next. Like we're going to end this podcast and I'm going to go think about the org chart a little bit more. And by the time I'm done looking at the boxes, everything might be different. And I think that is, again, the weirdest outcome of all of this is we still don't know.

1:09:39We might be on to six CEOs of OpenAI by the end of this week at the rate we're going. Each one of the board members is going to become the CEO until there's only one of them left. It is nuts. And it is, I think, one of the most riveting corporate dramas of our time, but it's also just, frankly, one of the saddest. Yeah, agreed. All right. I think we're done here. Neal, you want to take us out? That's it. That's the Vergecast. That was also Decoder. We love hearing what you think of our shows. You can send us emails at vergecast.verge.com or decoder.verge.com. We read them all. You can follow us all on threads, where we broke a lot of news this weekend.

1:10:13Sign of the time for threads, I will say. Although most of the action was still happening on X, a lot of it happened on threads. And very importantly, you can subscribe to Alex's newsletter command line, which I am 100 % confident without having asked Alex first is going to be full of scoops about what's happening at OpenAI for the next 500 weeks to come. I hope so. I sure do. That's it. Rock and roll.

1:10:41And that's it for the Vergecast this week. Hey, we'd love to hear from you. Give us a call at 866-VERGE-11. The Vergecast is a production of The Verge and Vox Media Podcast Network. Our show is produced by Andrew Marino and Liam James. That's it. We'll see you next week. The Limu Emu

1:11:26Fascinating. It's accompanied by his natural ally, Doug. Uh, Lemu? Is that guy with the binoculars watching us? Cut the camera. They see us. Only pay for what you need at LibertyMutual.com. Liberty, Liberty, Liberty, Liberty. Savings vary. Underwritten by Liberty Mutual Insurance Company and affiliates. Excludes Massachusetts.