Anthropic Beat OpenAI to the IPO

They Might Be Self-Aware, Anthropic Beat OpenAI to go public. Sorry, you have reached a number that has been disconnected or is no longer in service. Mainlining the static between salvation and the margin call right here on They Might Be Self-Aware from the blur I am. Hunter Powers joined by the static himself. Daniel Bishop. I was going to try and make, yeah, does that, is that good audio? Just me and Hunter doing that for half an hour. I mean, that's basically the episode today. Well, that, and we're going public. We've been keeping it private. Not a lot of people know that we're going to take this podcast public. As in like record from a park one of these days? Well, I was just going to make the RSS feed public. So that, well, I mean, technically it already was, but we never officially announced it. And everyone else is talking about going public. So I just thought it was like a natural next step. Yeah. Okay. So we're going to be offering an IPO at what? Eight point. Ten trillion dollars valuation. Is that we're worth that? Right. We're we put AI right in a lot of our marketing type. That is we say it enough. Although I, we've got to be worth at least a few billion dollars because we say AI this many times. We say it a lot, but it's not in our name. And that honestly, that's like knocked us down a few notches. Our name is, they might be self-aware and some people got to figure it out. Like it's not immediately obvious. Do we have a dot AI domain? Well, yes. So our parent company, the blur can be found at the blur dot AI. And that's actually where the website for this podcast is. It's an AI company. Yeah. Yeah. We're pretty legit. Four hundred and twenty point six nine trillion dollars. I think. I think that's it. Yeah, we did it. We said AI. All the AIs are going public. I mean, Google obviously already was. But we're talking about. Well, Anthropic's the big one that people are talking about right now because they. It was going to be open AI. Everyone was. Oh, maybe open AI is going to be IPO wing soon. And then Anthropic said, we're going to beat you to it. And then they did. Yeah. Well, maybe. Yeah. So I guess. So first, Anthropic has now officially said that they are going public. They filed their S1. That's what it's called. Your S1. We really need like a business lawyer or something to come on this pod. Hey, Hunter. Anthropic keeps it every article. And there's a lot of them right now about these. Hey, Anthropic confidentially files to go public. If it's confidential, why does everybody know about it? Hunter. OK, so when you are going to go public, you have to file something called your S1, which is mainly a disclosure document. It's sort of your prospectus about what you think the future is and how you're going to take over the entire world. Yeah. But also at the same time, you have to air all your dirty laundry. And companies tend to like very legitimately use this as an opportunity to air every bit of dirty laundry. Yeah. Actually, we don't make any money. Uber was a famous one for this. When Uber did their S1, they're like, yeah, so we're going public and we've got this giant evaluation. But we will be worth so much someday. Well, actually, they even said in there, there may not be a profitable business model here. It's so wild. And we're just going to have to try to figure that out. But we do have a large percentage of the market share. We got if market shares how you want to judge us, we are doing great. OpenAI a year ago, I would have absolutely said would be in the same boat. Maybe not necessarily profitable, but everybody's using them. We'll talk more about everybody using them later because there's a number that I literally can't believe attached to them. But that's for later. So S1, you have to air your dirty laundry. And Propix said, hey, we actually might make an operating profit. Not a, I don't think that means actually profit, but you know, making some money. Good for them. But they did not file the public S1. So there's this other thing you can do when you're going to go public is, so you have to file an S1, but you have the option of filing a confidential S1 before you file the public S1 when you file the confidential. So with the SEC, the Securities and Exchange Commission, they have to review and approve your S1 documents saying, okay, you guys are ready to go public. So you can file confidentially and people still know that you file, but they can't read it. It's not, there's this database called Edgar. I'm sure you've, you're familiar, I've heard of Edgar. I have. Yeah. And many companies, they're always like, can you scrape Edgar? Is that an option? Almost everyone on this podcast is currently saying, I don't care. And you know what? Me either. The point is, right. I was just going to go like an hour on scraping Edgar. That was my, we shouldn't. I mean, I'd find it interesting, but confidentially filing an S1, it is the dirty laundry, but we're not allowed to see the boxer briefs up on the line. Right. And the rumors are they're really bad. And also the rumors around why the open AI is like, why we're playing these head games of who's going first is whoever goes with the public S1 first is going to take the hardest part of the like, oh my gosh, that's what's really going on at this company. Because when, again, because the other ones will also be pretty similar. They're both horrible. The disclosure part of it that whatever we're, we're spending, you know, for we're losing $4 billion a day or something like that. We're not going to be surprised if we see that, but I think we're going to see something like that. Then all the media is going to react. Oh, look at Anthropic or look at open AI. Oh my God. It's a dumpster fire. And then everyone will calm down and then everyone will get excited. And then the other company will come in and say, Hey, we're going public too. In fact, they're only losing 3.9 billion dollars a day or just losing like an extra half a billion a day. Sure. No big deal. And the market will calm down. And then, I mean, we shouldn't gloss over. There's a third company, third AI-ish company going public. SpaceX. SpaceX. Oh, for sure. Yes. There's the ish, right? So SpaceX, the famous once just was rockets, now is rockets and AI and also used to be Twitter stuff all kind of rolling. They're basically AOL at this point. Hmm. Hmm. No. No. I don't know if I can quite follow that one. You lost me. But they have the best satellite internet. Right. Tesla is also kind of getting in there too. Well, yes, but SpaceX and it's one of their most profitable parts of their company is their internet. Their satellite internet. SpaceX has Starlink. And that's one of the most profitable pieces of their company. That's where I'm getting the AOL connection. They are. Sure. You brought me back around. An internet provider, Daniel. America Online. America Online. Hmm. But they're going public first and at an outstanding valuation, right? In the trillions of dollars. I think. I mean, I don't know what. I'm not much of an SEC filings guy. How much do companies usually say their valuation is? Is it maybe not this much? Maybe not usually. No, I don't. It sounds like a lot. Yeah. I don't think it's usually this much. And I mean, what it really matters is they'll they set the starting stock price and then it's trading openly and it will go up or it will go down. But it's right. That starting price is. And with the SpaceX one in particular, because of some deals they've worked out, a large percentage of that stock will have to be purchased by various pension funds. And these. Yeah. So. Right. And at this point. Everyone take a look at your Vanguard account. And see how much just got shoved into AI. AI. I. So I know someone actually in this kind of like. 4. OK, kind of adjacent sort of space. And I've been told that, like, you know, anyone with 401k, you know that you have some degree of like, I want more risk or this or certain like sub types that you can do. I've heard that there are now like entire funds that are specifically like, look, not AI. AI. It's all the other stuff before everyone went AI crazy. The rest of the S&P 500, that kind of thing. And I, as I've mentioned before, one, this is not a financial sort of podcast. Two, if you are thinking that I'm saying anything even close to financial advice, not only should you not follow it, you probably should do the opposite of whatever I'm suggesting. But to me, a very uneducated in this kind of, you know, economics space person, space. Does it feel to you like we're going to try and cash in while the bubble hasn't yet popped and then get our bag of money and run before it all comes tumbling down? Or, and like the only reason why I even say the or is because supposedly Anthropic, who I thought was definitely just chucking money into the never ending fire, might have an operating profit soon. Like, is it possible that they are one point whatever trillion dollar companies? Or to you, does this feel like, look, we're going to go public. The five people up at the top are each going to make their $72 million over the next couple of years. And then once it's all gone, well, at least they got theirs. Yeah. It feels suspicious. But, you know, we can go. That's a succinct way to put it. Sure. Because they're all kind of going at the same time. And it's definitely there is what's called investor pressure. There's what was Anthropics last round where they raised. So they went out and said, hey, we need to raise some more money to keep this company going and keep growing. And they raised $65 billion. And that was, you know, to get them to the next year. Hey, we just need a little bit more money. And you and I have worked in a number of startups where it's not uncommon that, well, A, startups almost by definition are losing money. And they hope to one day make it up. And you're in these situations where it's like, yeah, we got to go out and raise our numbers are usually lower. We need to raise like $5 million, $10 million, $20 million, not $65 billion, but still giant numbers. They're still giant numbers. So then those investors, they want their return. They want to see something back. And Anthropic and OpenAI have so much money that there is got to be a ton of pressure on them to turn a profit. Oh, sure. And by profit doesn't mean actually make money, by the way. Profit means have a liquidation event, an exit event for their investors where the investors can profit. Right. Because every startup, you know, all that VC money going in, they want to see that 10x, 100x, 1000x exit. Yeah. And if you're already starting at one point, whatever trillion dollars, I think it's going to be a little hard for all those people who've poured in the billions in the first place to say, and this is worth it to keep that company going. Like, I don't know. Is this the equivalent of the Hollywood blockbuster that put in $400 million to the movie and then made $401 million on like the ticket sales? There are elements of that. So, Louis can pick on Anthropic as an example, but this is... You're picking on my favorite? Come on. Be nice to them. Well, all right. So, there's a lot of investors in Anthropic that are very upset right now. And they are upset because Anthropic is going back through their books and identifying certain investments and saying, yeah, we're basically nulling this out. We're not counting it because you didn't follow all of the rules. You didn't follow on at this point. So, one thing that's not... And honestly, it still shocks me. I've seen it done multiple times. So, like, I know it's a real thing. You invest in a company. You would assume that by investing in a company, you buy 20% of this startup. Sure. That, okay, when the company finally sells to Microsoft or Google, Google buys it and they buy it for a million dollars. Well, you're going to get 20% of that million dollars back because you own 20% of the company. Yeah, that's how that works. You would think that's how it works. No. That's not how it works. No, right? At all. What? Not in the slightest bit. You only own 20% of the company for, like, a short period. And then when they go to raise more money, sometimes you have the option of you investing more money to keep your 20%. And sometimes they're like, no, no, no, no, no. We're not even going to give you that option. This would be called going pro rata. But keeping up... It'd be a shame if something was to happen that 20% to a company you own. You have to keep investing to keep the money up. And when you don't invest, when the next round comes around and you say, you know what? I'm going to just... I want to sit back this round and see how things are going. And come back to me with your next round. I'm still interested. Now it's five. You tend to be looked at very negatively, especially if they find success in that round. They're not coming back to you and they will do everything they can to see that your percentage of ownership goes down. Okay. This whole thing's a scam. Yeah, it goes down. This entire thing is a scam. All right. So this is all, I would say, fascinating but horrifying is probably the more correct option Or wildly complicated. How is this going back to... Are people cashing out here? Right, right. Is this the beginning of the big cash grab and panic exit? Does this signal the beginning of the end? Is the AI bubble because these companies are going public about to pop? Or is this a sign that we are reaching a level of maturity in this industry where this is now going to be not just the like nascent titan of industry, but something that truly sticks around. Is this part of the economy forever? And I don't mean that a little bit. I mean like a, it deserves to be 30% of the entire stock market. The truth is probably somewhere in the middle. All right. So obviously not all AI is just going to go away. And I don't think it's going to run the entire economy forever. The overwhelming opinion on all three of these public offerings is that they're probably going to have a short-term bounce. Meaning when they go public, they're probably going to go up a little bit. But when we look at them just one year out, we're expecting to see them down from their opening price, not up. And so that's like first just if you want to, I want to take the average of what were all the experts saying? That's what the experts are saying. They're predicting they're not going to hold these valuations. They're valued too high. It's not real. I mean, have people been saying that about specifically Tesla? Now granted, we're getting into the automotive industry, but there's plenty of touch-up against AI and self-driving and all that. And people in this industry. There's also strong rumors that Tesla is going to get wrapped into SpaceX, but. Yes. But like. I'm not saying that specifically these other companies associated with this technology. Like industry are like way, way, way, way, way, way overvalued. Yeah. Isn't it all just made up? Well, people absolutely say that. But there are companies. I mean, Uber is doing pretty well. Like I would, again, after seeing Uber's S1, I was like, this can never. They literally come out and say. Yeah. There may not be a profitable business model. And by the way, right now we're going to run out of money on this date and we will have no more money except again. Now we're going to sell some stock and that's going to give us more money. But otherwise we're running out of money and we don't know how to not run out of money in the future. But still that company seems to be pretty, pretty well thriving. Are they profitable? I never actually bothered to follow up because I remember hearing years ago, like we're super duper not profitable. Did they ever become profitable? I mean, we can ask the overlords. Is Uber? My guess is yes. All right. While he's looking that up, everyone place your bets. Yes. Uber is profitable. They reported $263 million in net income for Q1 of this year. Good for them, I guess. They really pulled that one out of the fire. So. But here's my. I'm curious. I'm not bullish on these companies. I'm not bullish on open AI. I'm not bullish on Anthropic. SpaceX is like a different beast. Are we bullish on AI in general? The concept of AI? Wildly bullish on AI in general. It's just that it's that competitive moat thing. I don't. Yeah. They have the users. I'll grant them that. They're right. Anthropic is now technically like bringing in more money than open AI and open AI cross whatever. One billion users and one eighth of the world has now used chat GPT or something. They've got the users. And I don't want to pretend that that's not terribly valuable. But what they don't have is any, at least in my mind, any sort of real stronghold on the technology. And as a technology guy like that's the most. Or the mind share. The hearts and minds. I mean like a year ago we were just basically. I was already using, you know, the Anthropic models then. But everybody was just saying chat GPT for everything. I'd still say no normal people know what Claude is. When I talk to normal people. I very much believe that's changing. And the numbers say that that's changing in terms of like year over year growth. Yeah. But that's enterprise. Chat GPT is still growing. Enterprise is what's driving the Anthropic. It certainly is. In the business world, I would suggest normal people have heard of Claude. But the outside of business world, when I say Claude or Anthropic, people are like, what's that? What's that? By the way, these are people that are not actively. They don't use chat GPT, but they know of chat GPT. They don't know about Claude. I'm like, well, it's like chat GPT. They've heard of chat. Everyone's heard of chat GPT now. And apparently one eighth of the world has used it. So, right, chat GPT, one billion global monthly active app users, according to Sensor Tower. I don't know who Sensor Tower is, but they're apparently counting people. They've got a tower, Daniel. It's an entire tower. It looks like a cell phone tower, I guess, maybe. Sensors, right? What other sensors are there? It's only cell phone towers. Do they have a .a domain? Mm-hmm. And that's important. So they're worth a few trillion dollars. But let us suppose for a moment that core conceit is true. A billion, one capital B billion global monthly active app users. There's only like 8 billion people. True. I literally do not believe that to be a true number. Well, users, I think, is then different from people. Like, it's not one out of eight human beings on the planet is using the chat GPT app from the app store. It can't be that. It can't be that. It has to be a bunch of people, many people, lots, loads of people, but not a billion people, have logged in from five different devices with five different IP addresses. Or the same cell phone in four different locations that connected to a different, dare I say, sensor tower. Ha ha. But it can't be a billion people. Do you disagree with me? Do you dare, Hunter? Yeah, I mean, I don't know that I care all that much. And it sounds like it's plausibly in the ballpark. And so, I don't know. I haven't really zeroed in that point. And Claude's saying they're doing 56 million users a month are using the Claude app. Right. Way, way, way less. With still, you know, huge year over year growth. And I do think that in terms of, like, mindshare and so on, Claude is very much the star on the rise. And opening, I think, is, even though their growth is still positive, I think they are very much coming down, both at the enterprise level and the, like, regular people level. I don't know. A billion active users, sure, maybe that helps drive that insane valuation. 56 million users of an app with a free tier on it maybe doesn't drive that same level. But, you know, enterprise. You wave your hand vaguely and you say, it's enterprise users. There's a lot of money to be made there. I also think that both OpenAI and Anthropic, and OpenAI to a bigger degree, but I think Anthropic too, they're kind of trying to be the everything companies. You know, we have an AI that does everything. And I feel like... We'll do pictures. We'll do videos. Maybe music. Everything companies don't tend to work long term. Yeah. We tend to like companies that are more specialized in something. Oh, you go to this company for that thing. And I think part of it's also just the management problem that, okay, now you're going to have a company that's doing everything. So, all right, well, now Anthropic needs to have an automotive division and a kitchen division and a legal division and a doctor division and all these divisions to manage all of these products to do everything rather than being able to focus on what they're really great at. And it just sort of... It becomes very generic. What's the phrase? Jack of all trades, master of none? Yeah. Right? Even for the big companies. Even for Google. Alphabet. Like, yeah, they have a million little divisions. But even Google doesn't do everything. And Google famously makes all their money from ads on search. And then the rest of the company is just about, well, okay, you can lose money. That's fine. But funnel some traffic to search and we'll call it even. So, that's a very unique position with that. But, yeah, they're not an everything company. And Google products fail all the time. So, I just think as... And the technology is still getting a lot better very quickly with many open weights models being released. I think it was just yesterday. Google released another one that... And by the way, Google. Like, the competition, the publicly traded companies being like, Yeah. I've heard of them. Here's another free model. This one, we specifically optimized it so that you can run it on your laptop with just kind of like a minimum, very normal spec laptop. I think they said as long as you have 16 gigs of RAM, you're going to be able to run this model locally. And you can have a chat GPT-like experience for free. We have chat GPT at home. Yeah. So, right. In the last few days, honestly, even multiple times in the last month, I think, Google has released a couple variations of some model family. Gemma? Gemma? Gemma. I say Gemma. Potato, tomato. Tomato, tomato. So, G-E-M-M-A. However you want to pronounce that. We'll agree on the spelling. Four. And it's family of models. And here's the dense one. And the mixture of experts. And the active 3 billion parameter. And the big one. And the medium-sized one. But, right. I keep talking about how the local future is one that I think we all really need to be focused toward. While there's also always going to be the cloud future. But if you have an open AI company that's trying to be the everything company. And open AI. If you have open AI and they're trying to be the everything company. And we do video. And we do images. And we do code. And we do chat. And we do this. And we do that. And speech to speech. And so on. And then Google. Or one of the Chinese labs. Or whoever comes out and says. I mean most people could run something that'll probably do most of what they need. Here. Have it. It's all yours. Enjoy. Google's been installing them in your Chrome browser without you knowing it already. And they're just distributing them freely. And so this is the technology. And I love this. The technology is continuing to get a lot better. And continuing to be more and more democratized. And what I would speculate is going to happen is that leaders will start emerging in different markets for. Okay. Like we're sort of the spreadsheet office. Like there's an office company focused on that area. Oh we're the tax guys. But anything that you do there could be subsumed into the broad capability set of any of these other ones pretty quickly. It could. But my suggestion is it won't be as good. Sure. There always is a best AI company for that one. But we're going to have the best foundational model first. And that will sort of generate the best. Let's say do your taxes experience. So we're always going to be sort of at the top. And yeah TurboTax inevitably is going to have their own AI that will do your taxes. But we think that our six month one year that we can stay one year ahead of them at all times on the foundational model that's powering it. And that that alone will be enough for you to say you know what I'm going to have my taxes done by Anthropic or Claude versus going to TurboTax or whoever the new company is that will do your taxes with AI. Yeah. I think that there is a lot to that. But still it's this local be it Google or like NVIDIA semi recently announced and released a different model. You know Quen and other like Chinese labs have not rested on any laurels. More releases have been coming out there. Quen 3.6 supposed to be very very good. I haven't actually tried it. But if you can let's actually say it's not even AI. It's not coding. I mean suppose that you had a 3D printer that could straight up make you an entire car and it's going to cost $5,000 in materials of like it you know it costs something. Daniel I want to be on the record here. I would never download a car. Okay. That old gem. If you could download you know here is how I would make a car. That would be very legal. Opener make me a car and it's let's call it 90% as good as a Toyota Corolla but it costs you a fourth, a fifth, a tenth the price. A lot of people would probably go for that right. If we enter into a future where you're supposed to be paying $200 a month for the really good frontier model and you say I don't want to spend or can't spend $200 a month but I do already have a laptop and I can run something that's pretty much as good for most of the same use cases for free. Why wouldn't you? I don't think open AI it's not even like something something valuation and stock market and whatever they can't survive as a company because they don't provide something that we can't get for basically free almost already and definitely for free a few years from now. What's frontier intelligence that we have? Yes you can't run that on your own computer. You never will be able to but it won't matter because you could do something almost as good yourself. How can open AI exist in that world? Well so first off again for the average consumer there's always going to be you know enterprise you want to pay for the best and so on. At the end of the day I'm in agreement I'm not bullish on them as I said it very clear I'm not bullish on them but if we want to look at counter examples you know Microsoft Windows. Why was anyone buying that when Linux was available? And was free and you had to spend tons of money. Like Libra office instead of Microsoft office or Google Docs. This disparity has always existed and there's has been a solution that is let's just call it 80% as good entirely free and a little more complicated to set up has existed for a long time and yet the company that has a giant marketing budget and is honestly 20% more polished. They've continued to thrive. And also by the way we're seeing all these record users for Anthropic and OpenAI right now consuming their models. It's not like DeepSeek 4 doesn't exist and costs literally one one hundredth the cost. So there's all and we'll call on Anthropic in particular the majority of their money is coming from enterprise. So these are all people that literally could just say you know what I'm just going to switch over to DeepSeek 4. I'm going to slash my budget by 99% and yeah I understand it's not going to be quite as good but they don't do it because it's not yet comparable and there's other things at play. It's not just the budget part. Right. It isn't even the but it's a Chinese model like you don't have to send the money to China. There are US based like data centers that will host those models and that you could just will hook up to that. And you're right people aren't doing that as much so I as much as I want to say why aren't we just using the local models. They're almost as good. There's so much to be said for convenience. And so because it's just on the app store and like well I've heard of open AI before click install ask question get answer like great. My my convenience level is satisfied and I got the thing I was looking for there. There is something to that I suppose. Here's a potential reckoning and a potential support of the idea that the companies have determined that if they're going to have an exit they need to exit now like because of the bubble being popped. So at the in our current environment many companies internally companies are being just encouraged to spend on AI to spend on tokens to have you should have the best AI available to everyone at the company. And if you're not you're doing something wrong. You are failing as a company you're failing as a leader. You need to figure this out. Yep. And at the same time we're now having very large companies come out. Uber I think was actually one of them where they said you know what we burnt 90 percent of our IT budget in the first quarter like for the entire year on tokens. And we're having a lot of difficulty in defending the ROI math on it. When we look at just nuts and bolts what's actually coming out that's actually moving the needle. It's really hard to justify and more and more companies are coming to this sort of like yeah we're look we're spending the money but profits aren't up. And right. Yeah. We cut some profits don't 10x or you one tenth the amount of money being spent on your employees because you fired them all. Then where is that where's the actual value proposition. Sure. And I don't think it's none but right we're figuring out how much value this actually brings and at what price point. This is where inevitably people are going to go to the next stage is OK look we're all lying that there's something here and AI super valuable but we're just like we're not actually at the point yet the moment yet where we can just say the AI model is going to replace this many employees and they'll just they'll literally do the exact same job that we used to pay all these people for. Right. Right. So and costs are out of control because we we need to be a profitable company where I think they're going to start looking at hey deep seek v4 over there that's one one hundredth of the cost. Let's try that out for a quarter and let's see we know it's going to drop our cost by 99 percent. Let's then and the current one's not making us any money anyways so. We can at least not make more money for less. Yeah. And maybe it's good enough. And I think that you will start and not this year but next year you're going to start seeing more and more of that as the experimentation where either it's a they have these automatic routers being set up. So the really hard problems we still will pay open AI or Anthropic or whomever to get the best answers. But 90 percent. Frontier intelligence for five percent. And then local thing I put on my own computer or at least the cheap cloud provider for the daily drivers. Well yes so for the. That's how business works Hunter. We've we've been doing that with people for forever. You have the more highly paid senior employees that help drive your architectural decisions and so on. And then the more junior people under them that help drive a lot more of the day to day work. All we did was rediscover like what tiers of varying levels of seniority. The last element I think of this that I've definitely since the change in the discourse about recently is the idea that the AI is going to directly steal so many jobs. And let's we can call on Anthropic or open AI. If we go back like one year they were out there in the public saying in another year 75 80 90 percent of engineering jobs are going to be done by AI. No one thinks that at this point including them who are they're actively hired record saying hey maybe that was a bit over. Yeah we were. And so I I'd suggest that. I've been saying this from the beginning. It's a tool. This is a tool that helps people not replaces people. Some people but like not the this is not the extinction event. This is you don't need as many farmers to plow a field. Yeah. And so that's the other reckoning that I don't know has completely hit the markets yet. The valuations because the last valuations are still on the round. We're going to we're going to replace everyone and AI is going to do everything. And it's basically as good as a person. But I don't know. Try to stand up a Hermes agent that actually does something. Useful and productive and doesn't embarrass you constantly. Just give it a shot or open claw or whatever you want. These are very useful tools that you should be integrating into your workflow as a person that works at companies that has things brought to you and outputs from you that have to go out. But it's not replacing you. No. Anytime soon. Or if it is that's going to be bad news for the company that did the replacing. And I think more folks are going to find that it's not only going to be worse. It's also not cheaper. And then I think the last question is around our you know does this change when they become self-aware when we if we're able to in the next year which is a possibility plant the flag that we've achieved AGI and sort of universally agree. We've achieved AGI. What happens then with these these companies and the market as a whole does that accelerate the devaluation of these companies or does it. Rocket ship to the moon them. If we get to actual AGI. I think all bets are off all of them because like yes the companies that currently exist that made the first one will have insane valuation for 11 and a half minutes before we go straight from AGI to ASI. And just like that's that's the ballgame at that point. Right. I mean we're done. If we get to actual actual AGI a full everything a person could do you know especially like via computer whatever that it could do it can think and it's creative and it has a memory and so on that lives and grows over time. I think what is the stock market price of the company that first came up with it becomes irrelevant shockingly fast. So your prediction would then be that let's say in the next year one year from today we achieve AGI and generally everyone agrees we've achieved it. Technologically speaking. I don't think one year. I don't think LLMs quite get us there. We're getting like oh it feels so very close but there's going to be something else. There's going to be another just like when transformers the not the you know Hasbro whatever. I mean like the the architecture for neural networks when that came out it was a seismic shift in my industry in natural language processing and then still using transformers under it but like attention and so on that made chat GPT possible GPT 3.5. Whenever that came out another huge shift in this industry. I think we need one more huge shift in this industry before we get to what would be AGI. But once we get AGI like that's then the apocalypse scenario of all of the white collar jobs are gone. I don't mean some of them. I mean literally all of them because the white collar jobs are I could do it at a computer. Those are gone. Those are just gone if we get actual AGI. AGI. Even if it is more expensive because it'll be more expensive for one iteration which it will improve itself on because it'll be AGI. And then six months later it's twice as good and half the price and then three months later it's twice as good and half the price. Yeah. While I'm trying to wrap up this episode you call on like another really good point which some people are seeing early on now. I mentioned the Hermes agent and there's been a number of these examples where companies. Like the fashion company? The what? I think they do like. Oh the fashion. The handbag thing. No. I think it's called Noose Research is the company behind Hermes which is a competitor to Open Club. Just think of it. I haven't heard them in a while. AI assistants. They're AI assistants. And a number of companies have tried to stand these up to do jobs. Okay you're going to be a customer support agent. You're going to receive all the email. You're going to come in and you're going to do reports every day and you're going to send things out. And they've had some success with establishing this and setting it up and it actually can do a lot of this. But what they're discovering is that it's very expensive to run because you can't use any of these subscription plans in general. There's some more crowns. But with this you need to actually buy the tokens. And at this point. More and more. All of the all you can eat plans they're going away. And you've got to buy the really expensive tokens. And so we are actually already getting to the point where yes you can run the AI employee but the human employee is cheaper. Yeah. And that's something that could accelerate and throw a real wrench in this. Yes. AI can do this but it's not cheaper. It's not. If you could use. If we can get them to learn I think that changes. And when I mean learn I mean through the course of a conversation truly like incorporating all that knowledge. One of the main reasons why this is so expensive is if I just say with a new no context window filled up write me a Python program that does X Y or Z. It cost 20 cents. And if I wrote it it would have taken me an hour. And so like in terms of time and money that was way cheaper. But if you say I need you to look up this database and here's this wiki entry and here are you know 10 relevant emails and you need to log into Zendesk to look up these tickets and also GR to look up those tickets. And then compile all this stuff and then write a plan and then start iterating on it. And then you have to check your thing and then run the test and then and then and then and then. Then you've used up a gazillion tokens and every new turn in the conversation is this exponential growth because turn one there was no context turn to there was 10,000 tokens that you have to send out and it thinks about and then starts outputting. And then turn three you had 50,000 tokens that went to it and then came back and then turn for a hundred and so on that that growth there. That's where a lot of this spend comes from. If you just ask it make me a thing apropos of nothing it can do that cheaply. If you say take all the context and I need to do that every time I have a new thing for you then it takes a ton or cost. And so that's that next seismic shift if we can get to I only have to teach it about my company and about the database and about the whatever once and then it just gets it. And I don't have to spend all that extra money having it relearn that every single time. Then we're on to something. But how do other people learn when we're on to something Hunter? This has been a long one. This has been two and a half people and I'm not going to explain the half for the two and a half people that are still listening. And you probably are if you've if you've listened to this long there's almost like zero chance you haven't already subscribed. So thank you. I hope you're still listening because right at the end there some of the best conversations we've ever had. And you were here for it both of you. But on the off chance the outlier chance that and maybe it's that half the brain cells in the Petri dish that's playing Doom and listening to they might be self-aware. If you haven't if you are brain cells in the Petri dish and have not yet subscribed or followed us use all the appropriate language please do. Because we keep talking about these things and hopefully bringing you some unique perspective. And we would love you to be a part of the conversation to shoot us an email post something in the comments write a review. Give us some feedback. We're here for it. We're certainly enjoying this journey and we hope we hope you do too because we will be back very soon with yet another episode. They might be self-aware.