Oct 11, 2024 · 2024 #36. Read the transcript grouped by speaker, inspect word-level timecodes, and optionally turn subtitles on for direct video playback
Edit labels for this show, save them in this browser, or download a JSON override for the production folder.
Transcript Playback
Open AI is a Multi Trillion Dollar Opportunity
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Speaker 3
Hello, everybody. It is Friday, the 11th of October, 2024. It's been quite a week in technology, especially for AI and Google. Earlier in the week, two men, including Geoffrey Hinton, the so-called godfather of AI, won the Nobel Prize for Physics for their work in foundational AI. Hinton had been at Google, no longer is there, but many of the most important ideas for LLMs were developed at Google. And then the day later, I think it was Wednesday of this week, DeepMind, which of course is owned by Google, Demis Hassabis and John Jumper won the Nobel Prize for Chemistry. But all the news this week hasn't been on that more positive side of Google. It's about the breakup of Google. According to the Financial Times, the Google... Breakup reads like antitrust fan fiction. The FT has an interesting piece, which is in the This Was the Week newsletter, which it imagines Google split up. And in spite of all the Google news, Keith in his That Was the Week newsletter focuses on AI. The title of his newsletter and his editorial this week is Open AI is a Multi-Trillion Dollar Opportunity. Keith Before we get to OpenAI, what do you make of this week for Google? And isn't that ironic that in a week where all the headlines are dominated one way or the other by Google, you chose to write about OpenAI?
I thought you were going to say about the past, Keith.
Words and timings
Ithoughtyouweregoingtosayaboutthepast,Keith.
Speaker 2
Then you would... Well, it certainly is both. And it may be the future. That's still an open question based on what happens. But I chose OpenAI because of its impact on the future. Now, the Google stuff, it's interesting, the FT piece you mentioned, because... And I agree with this impressionistically, but I'm not enough of an expert in law.
Let's quote the beginning of the FTPs. Imagine a world where Google, Pear, and Alphabet, at $2 trillion, the world's fourth largest listed company, is torn asunder. Its search engine goes one way, the Android operating system another. Web browser Chrome, advertising businesses, all set free in the name of fostering competition. It is a world Google's detractors might welcome, but not the one investors see.
Yeah, now, if you imagine the conversations we've had about the Americanization of tech with companies like Palantir and Anduril doing military stuff, and the discussion we've had about the nation state, you would think the American nation state, if it was self-aware, would want a strong Google. But apparently not. Apparently the American nation stayed.
Well, if it had a sense of its own history, vulnerabilities, strategic interests, if that was the nature of the American government. It probably wouldn't want to break up Google. It probably would see Google as a global asset in a world market. But clearly, it doesn't. Some ideological overlay is driving them to want to break Google up because it's too good. And I think reading the FT piece, it seems very likely that that will not happen, that The DOJ's wishes are unlikely to be realized. But they are driven by a kind of zealotry, which is somewhat anti-American, if you think about it.
Well, now you're beginning to sound like you know who about make America great again. And anyone who criticizes anything in America is anti-American. I mean, there is a legitimate argument for it, at least maybe not breaking Google up, but certainly some antitrust case. Would you reject that entirely on the Google front?
I think even when you look at it through the framing of competition, Google is a weakening giant, not a monopolistic giant. Its search business is being ripped apart by Amazon and TikTok and Facebook in terms of advertising. The actual search business is being ripped apart by AI. it is struggling to build its own AI to even have a chance of competing in the next phase of technology.
Well, that might be a little bit of an exaggeration. One of the pieces that I encourage you to put in, you chose not to, was a very interesting Wall Street Journal piece on Google's new Notebook LM AI, which in my mind is just as good as anything that OpenAI has.
No, I'm teasing you. I know you've got a lot of material, and it's a stacked newsletter this week, a really good one, full of both interesting essays of the week and news of the week. But it's still – I thought that that – Wall Street Journal piece on Notebook LM was interesting. And as I said, I'm really impressed with Notebook LM.
Yeah, well, if you remember a couple of weeks ago, we actually played one for people, and we've talked about it before. So the Wall Street Journal piece is mainly noteworthy because... It's the Wall Street Journal, which means that it's getting into the public awareness that this thing exists.
It's more than that, though, Keith, I think, because Notebook L is easy to use. And one of the things that I think why the Wall Street Journal is probably the best newspaper for technology is it's a very good interface for learning about cutting edge technology, which is accessible for ordinary consumers. I'm not as techie as you and I can use Notebook LM.
Yeah, last week, actually, I gave it a show after we'd recorded the show and it did its own version of the show plus crib notes. And to be honest, it's good enough for me to consider using. You're going to replace me, Keith? I'm not going to replace you, although I do suspect that we did. I didn't put in the newsletter, but hey, Jen, H-E-Y-G-E-N, this week made it possible to record yourself and create your own avatar. And then you can give your avatar a script and it will look as if it's you speaking.
Yeah, this week, my wife, who, as you know, is a very senior person at Google, was at one of her senior meetings. And she told me that she was blown away by new technology. She wouldn't tell me what it was, but I'm guessing it has something to do with Notebook LM. So they've got a lot of stuff in the pipeline. In other words, I mean, to go back to what we were saying earlier, I don't buy your idea that Google is somehow and also ran in the AI in the AI race, which is what you seem to suggest.
Well, let me just correct you one notch. I think Google's old products and revenue streams are threatened. I will not at all write Google off from being a major player in AI. It can certainly happen. However, it comes with some challenges. Google have to replace revenue streams, whereas OpenAI just has to build new ones. And therein lies the rub. Google has 300 billion, if you can believe it, annually to replace if search ends up being pushed down. And OpenAI has, you know, a very good opportunity to make that kind of money. And so it's going to be, it's not going to be a one-way street.
Right, and that cannibalization, as we've talked about this before, it's the innovator's dilemma. But sometimes it's good. I wonder, in retrospect, I mean, maybe we can do a whole show on this, In retrospect, whether the DOJ's actions against Google in the long run might actually be seen for good for Google because it forces them to address some of these issues. But anyway, let's get to the core of your editorial. It's not about Google. It's about OpenAI. The title, Keith, is OpenAI is a multi-trillion dollar opportunity. Well, I might respond to that. Well, you don't even need Keith Teer to tell us that. Everybody knows that. So what?
in the late 20s, adding $141 billion to the cash flow, being enormously profitable. And I'm quoting you, what would have to occur for OpenAI to be this big? It's not that much beyond its projections, and those projections could either be too low or too high, and you have their bold revenue forecast. But they're not unrealistic. You've talked about that. I mean, so what?
Well... Firstly, the world doesn't really have more than a couple of multi-trillion dollar companies right now. I think it has Google, Apple, NVIDIA, and Microsoft. And most of them are one point something. It's very likely that OpenAI will... benefit from the fact that it's bringing, last week we talked about AI seeping into all the cracks in society as in like a virus. And that's already beginning to happen if you look at some of the job consequences that I discussed this week in this week's newsletter. So the total addressable market for open AI is the GDP of the world.
Yeah, but that's Keith Teer talk. I mean, so what? You could have said the same about Google in the internet age since the internet was taking everything over.
Not really, because Google was search advertising. Open AI is functional. Reid Hoffman's piece about super agency kind of gets into that, but it's functional as in it's going to do things and get paid for doing them, just like humans do. And that is a very, very large opportunity set at which I think he's underestimated. I think those projections of 100 billion by 2029 are going to end up being wrong on the low side.
So one of the pieces that you cite is Eric Newcomb's The Bear Case for Open AI at $157 billion. What's Newcomb's argument and why perhaps don't you agree with him?
Well, sadly, Newcomb is a paid substack and I don't pay for it. So I only put the... You haven't read it? I only put the top in, and the top is the bear case. Sorry, the top is the bull case. This says the bear case. So I don't know what his bear case is, but he does a very good job of articulating the bull case, which is why I put it in. And I would say to Newcomer, it's foolish for you to be paywalled because no one actually knows what you think.
Yeah, without the B. Oh, I'm wrong. His name is Eric Newcomer. So he's a comer rather than a cum. So what's his argument? Okay, so we got the bear case. What's the bull case?
The bull case is the one we've been making week after week.
Words and timings
Thebullcaseistheonewe'vebeenmakingweekafterweek.
Speaker 3
You've been making it? I haven't been making it. You haven't? Well, I'm not making the case that there's something inevitable about it becoming a multi-trillion dollar company. It may be an opportunity, but there's no guarantee that it's going to win. It's almost inevitable.
Yeah, but when you're thinking about startups, what matters isn't how much you raise. It's what is the ratio between how much you raise and how much you make. Okay,
well, I take that, but I don't know if you have the numbers, but I would bet a great deal of money, maybe not a trillion dollars, that that ratio between Google and OpenAI benefits Google.
Exactly. This is a Keith Teer slide, which is just.
Words and timings
Exactly.ThisisaKeithTeerslide,whichisjust.
Speaker 2
No, there'll be a spreadsheet underlying.
Words and timings
No,there'llbeaspreadsheetunderlying.
Speaker 3
I'm not saying you made the slide, but this is so speculative, Keith. Talk about other products that don't even have any. And this is other products for $25 billion.
I can tell that you've never invested in a startup because you've got to read this through the eyes of an investor. And there'll be a spreadsheet underneath this with line items. And there'll be whole pages that correlate to those line items. And the other products will be known internally. They're just not prepared to talk about them.
I take your point, although I'm guessing that the people who have put billions of dollars into open AI are not reading the slide deck. I mean, they're believers or they're not.
I would absolutely invest money in open AI if one could.
Words and timings
IwouldabsolutelyinvestmoneyinopenAIifonecould.
Speaker 3
You can't because... No, I know that, and you know that in the editorial. So you think it's a good investment opportunity. Most of the VCs just don't have the kind of cash to put in.
Minimum check was $250 million, and most funds are smaller than that. But Koestler put money in, didn't he? Yeah, but Koestler has several billion in funds.
So you're bullish. And I've talked to you about this before. And again, another piece that was in the FT this week, which you didn't put in the newsletter, but I thought was interesting. OpenAI pursues public benefit structure to fend off hostile takeovers. I wonder whether... One of the weaknesses of open AI, there's still a lot of questions and uncertainties about its corporate structure and governance.
Yeah. It is in the newsletter, by the way, Andrew, that one.
Words and timings
Yeah.Itisinthenewsletter,bytheway,Andrew,thatone.
Speaker 3
Oh, it is that one. I thought it was a good one.
Words and timings
Oh,itisthatone.Ithoughtitwasagoodone.
Speaker 2
It's in the news section. Now, well, a public benefit corporation means a for-profit that has goals above and beyond profit. So it's actually moving from a not-for-profit position to a for-profit, but with public benefit as one of its goals, which means that it can't be accused of undermining shareholder value. If it chooses to do something that's good for the public, but bad for the shareholders, it will be allowed to do that free of any lawsuits.
So I wonder, I mean, I take your point that the sky's the limit for open AI because the future is the sky. But then they're also competing against everyone else. They're obviously competing against Google. They're competing. It seems as if their relationship with Microsoft is increasingly competitive. So they have to beat the Microsofts and Googles and Facebooks and Amazons of the world, don't they?
Well, the way I think of it, Starting with Google, the main impact of Google's Gemini is on people who use Google products, like spreadsheets and documents and such. The main impact, Microsoft doesn't have a competing product. They literally use open AI inside. Amazon really doesn't have a competing product. Facebook does, doesn't it? Facebook does have a competing product, but it's made it open source and free. So Facebook is betting on glasses and social media as being the right place to put AI into. OpenAI is putting it into everything. It's totally into everything, including Google Docs. You can put OpenAI in Google Docs as well. So the canvas that OpenAI is addressing is the world. Google's addressing Google users. Microsoft is using OpenAI for Microsoft users. And everyone else, like Claude from Anthropic, is really good, but very much second.
But what happens if, I don't know what the antitrust people would do on this, but if Amazon, for example, bought, and I think they're a major investor in Anthropic, if Anthropic gets snapped up by one of OpenAI's competitors? bigger competitors and amazon and apple i mean google wouldn't be allowed to buy even in microsoft i guess that probably wouldn't work either i think that's more
about the doj or the ftc than it is about actual competitive issues i i think i think the foundation layer of ai there are sufficient players that one of them being acquired shouldn't actually impact competition at all.
I mean, you can argue it either way in the sense that OpenAI is everything, so it should be valued as everything. But what about the human dimension? There's the question of Sam Altman, his CEO skills, his relationship to the board. I mean, anything could happen to OpenAI. It could screw up. It could make mistakes. It could get sued. Altman may quit, may get fired. There are so many different... variables on this, human variables.
I think it might get there, but at this stage of a company, you need a kind of a fighter who doesn't listen. The more you listen, the less likely you'll succeed.
Yeah, I think last night they showed off the CyberCab and the latest incarnation of their humanoid robot, Optimus. And they were actually both very, very good. And Grok, which is his AI platform, is also very good, by the way. I use it now for image generation before anything else.
So anyway, so you and I have talked about this. Many times before me, we'll talk about it many times in the future. Keith believes that open AI is a multi-trillion dollar opportunity. Don't be precautionary. All in on open AI. I'm not enough, as he's noted, of an investment expert to make a call on that, although I would urge a degree of caution because there is the human element and who knows what will happen in the future. Anyway, we'll come back to that. Lots more this week. Very interesting piece by Charles Duhigg, the New York Times writer and the New Yorker on Silicon Valley as the new lobbying monster. And the heart of that monster is somebody called Chris Lehane. a very close friend of my wife. Tell us about Lehane and this piece. I know you were quite impressed. Some people might see it as a bit of a hit job on Lehane, but he's presented as this Machiavellian figure in Silicon Valley.
Yeah, I mean, this article is really good for capturing the moment of Silicon Valley becoming sufficiently influential in Washington, DC, that it can alter the course of legislation. And Chris Lehane, especially in the case of Airbnb, it's documented quite well how his role in getting politicians to understand the underlying power of tech has succeeded. And at the end, the article makes the point that Airbnb has more voters in its orbit than most politicians. And given that most elections, you can swing an election with as few as 50,000 votes one way or the other, technology lobbyists are super powerful. probably more powerful than any other prior lobbyists because most lobbyists don't have access to large numbers of people who they influence. So it's kind of a moment in history when we all know politicians are feckless.
Right, and the subtitle of the piece from Duhigg is From Crypto to AI, the tech sector is pouring millions into super PACs that intimidate politicians into supporting its agenda. It's interesting that Lehane is part of the Democratic Party and the liberal democratic social network culture of the San Francisco area, and yet some of at least his expertise is going to bashing Democrats and progressives. So I think there is some political payback in the Duhigg piece. One thing that I think is really important, Lehane was the head of government affairs at Airbnb, did a great job there. Now he is the right-hand man of a certain Sam Altman. So I think the Lehane legend now, I mean, he is... He's got the top job anywhere. He is explaining to Sam Altman how to present open AI to the world and particularly to Washington, D.C.
Did you think it was a bit of a hit piece or do you think it was actually quite simple? I think Lehane would be... tickled pink by it because it adds to his legendary status. But I'm guessing his wife probably wasn't thrilled with it.
You know, I think it may have started off wanting to be a hit piece, but there isn't really any facts that run counter to normal practice in D.C. So it ends up making him look good at what he does.
Yeah, and what Lehane does is he takes the Silicon Valley message and he makes it compelling in Washington, D.C., and it perhaps comes back to your observations on Google. Maybe Google should be hiring Lehane, Keith, in terms of their arguments against antitrust.
And I would say Google, in my view, Google has significantly underplayed its narrative to government, which I hinted at earlier when I said what I thought. But it really underplays its hand with the public as its backer. I think the public like Google a lot more than they like Facebook, for example.
Or even OpenAI, which is still a bit of a mystery. Anyway, well, Keith Teer suggests to Google they hire Chris Lehane. He may not be available. Another really interesting piece this week is by another Silicon Valley heavyweight, intellectually and otherwise. Reid Hoffman has a new piece in LinkedIn, his own company. So he's publishing on his own platform, rather like Elon Musk putting on X, about something he calls super agency. What is super agency, Keith?
Super agency is when humans leveraging AI extend their capabilities beyond what an individual human can do. And so it's really about the agency of human beings expressed through the use of AI.
And of course, nobody knows the AI world, particularly the investment in AI world, as well as Reid Hoffman. He had his own AI startup. I think he's an investor probably in everything, including OpenAI. I think he was involved from the beginning at OpenAI, wasn't he?
No, the opposite. He's contesting that and saying that you can't really separate AI from humanity because it's built to serve humanity and it will extend humanity's abilities even to the point of solving problems humans can't. But it only can do that because humans are behind it and creating it and programming it. So he's really saying that just as the wheel took humans from only being able to travel short distances to much larger, AI is the same.
I wonder if Reid Hoffman is a very prominent backer and investor in the Harris campaign. I wonder if Harris is indeed elected, whether just as Musk has volunteered his services to Trump, that Hoffman might get involved on the AI front. Harris is still struggling for a real message. I wonder why she hasn't really picked up on AI.
Yeah, well, this week's Video of the Week speaks to that. The Video of the Week is an admission by Democrats that they messed up with Elon Musk by not inviting him to the electronic vehicle summit at the White House, which, when you think about it, is just bizarre that they wouldn't invite him, given
But, yeah, I wouldn't be surprised if... if Hoffman becomes more involved. Although, as we noted, Super Agency sounds like a Reid Hoffman kind of book, a best-selling book. So maybe you will see that in your local airport, a book by Reid Hoffman called Super Agency, suggesting that technology is going to make us all super agents. Speaking of super agents, startup of the week, Keith. Tell us about, was it Miro, the innovation?
Yeah, the company's called Miro. Its past life is as a kind of a whiteboard stroke canvas for product teams to develop and iterate on ideas. It's very sophisticated whiteboard with lots of bits in it. What they did yesterday, and there's a video behind that link, is they launched the next version, which is an AI-powered whiteboard. And I put it in because it really shines a light on how jobs are going to be impacted. What this whiteboard now does is it takes all the conversation that teams have using Post-it notes and tables and charts and whatever else they might do drawings and it creates product specification documents from listening to the conversation and so it takes it takes days and days and days of work out of the product development process using ai and
I should disclose I actually own Miro shares because I was part of a company they acquired and I got their shares in return. And that certainly means I pay attention to them. So there may be something.
They're the product of the week. And finally, our post of the week. I asked you to change the title from X of the week. But of course, you did that. And yet now we have a post of the week all about Mr. X, a certain Elon Musk. Tell us about this post. It's from somebody called Colin Rugg, two L's, two G's. I've never heard of him, but he seems to be quite a... a pro-Trump person on X. Exactly.
So this is what I was just referring to, and it's coming from a mainstream media TV show, by the way, and it's pointing out that the Democrats were the natural home for Elon Musk. He's always been quite liberal, actually, on most ideas. But the Biden administration decided to go to war with him it relates to your point about what is Kamala Harris's message, vision, if you will.
I've got to get her name right. She's your next president. I'm from Yorkshire. We always flatten our vowels. We can't help it. So there's no politics in the way I say the word. Anyway, this was a conversation on CNN, which is worth listening to. There's a video if you click through. And it's the Democrats self-acknowledging their failure.
I mean, it was a panel. It's a classic X. No, but let me just finish what they say. They don't just focus on Musk. They make the point that the Democrats are no longer the innovation party. that they look at tech as a threat, not as an opportunity. And the must thing is just used as an example. I think there's a lot of truth in that. I think the Democrats are kind of becoming dinosaurs when it comes to tech.
Well, I think the Democrats are very conservative. So I think what they're trying to keep is America as it is. And Trump seems to think that everything America is needs to be changed. And Trump, of course, is increasingly bullish on crypto. There was an interesting piece this week on some of his commitment to crypto, which sounds a bit dodgy. I mean, this Colin Rugg guy, though, I mean, I...
I mean, one of his other posts is a picture of Elon Musk jumping behind Trump. He calls it a photo for the history books. Should we trust a guy like Rugg? I mean, I can't comment on this particular piece, but... It's this narrative that somehow the Democrats conspired to exclude Musk, so he had to go to Trump, which sounds nonsense to me.
Yeah. I don't think you have to trust him because he's not really stating an opinion. He's more like playing the role of a reporter. I don't think there's any opinions.
I'm getting the sense, Keith, that you are... tiptoeing into the Trump camp. Hold on, let me finish. You said that the thing that you care most about is technology and innovation. You suggested that the Democrats are against technology and innovation. They're the anti-American party for their encouragement to break Google up. You're a big fan of Musk. Why won't you vote for Trump?
Trump doesn't represent a positive future at all. Trump is a street fighter focused on himself, and there's no vision he holds that he won't drop the next day if it doesn't suit him. What the world needs is true visionary leadership able to execute human progress.
I don't think anyone of the two main parties has a clear narrative that you would recognize that in, either one of them, which is sad, right? It means, I mean, as a sociologist, I almost conclude that America's self-confidence has gone and is not expressed in the narratives of either main party. So they've both internalized a kind of a doom point of view, which is inappropriate for a time when technology could open up opportunities like that.
Maybe I know that Lehane is very well connected in the Democratic Party. Maybe Lehane should quit Sam Altman and go and work for Harris. I'm sure he could if he wanted. I wonder what La Haine would do with Harris, what he would tell her to say and not say. Because at the moment, she doesn't seem to be saying a great deal. It's still hard to pin her down on anything.
Yeah. I mean, I was on the Gilmore Gang last week, and the question was asked, why is the election so close? And it was implied that there's a lot of racism and sexism against Harris. And I said I didn't think that was the case, although there's always racism and sexism. So it's going to be some of the case. But the reason that it's so close is because she's so insubstantial.
Yeah, I'd rather her be more female and black and maybe more of a Stacey Abrams. But that's another. We'll have to do, Keith, we've got, what, three weeks till the election?
I actually, my own view is it's not a particularly consequential election. It doesn't really make any difference who wins, but maybe I'm wrong on that one. It's been an interesting week, Keith. A fascinating week. Lots of things happening. And then next week we're going to talk your friend Elon Musk, right?
Have a good week, Keith. Don't... Don't do anything stupid. And we will see you next week to talk more technology, more open AI, Google, Elon Musk, and everything else. See you next week. Thanks so much. Bye, everyone.