Feb 7, 2026 ยท 2026 #3. 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
Anthropics Super Bowl Ad is dishonest
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Speaker 2
Hello, everybody. It is Saturday the 7th of February. I'm talking to you, as always, from San Francisco. Tomorrow is the Super Bowl, the annual American celebration of all things football. One piece in the Financial Times talks about how the NFL became America's true national pastime. It's the one thing that brings Americans together. And, of course, it's taking place tomorrow in... Silicon Valley in the heart of San Jose. But as it happens, the... Super Bowl is not bringing Silicon Valley together. In fact, it's dividing them. That's been the big story of the week out here. Anthropic have a big ad about the absence of advertising in their product, and Sam Altman has lashed out. Keith, I know you are a keen observer of all things ChatGPT and Anthropic. What should we make of this latest spat between Altman and and Anthropic, Dario Amadei.
Well, the joke on the streets here is, of course, Anthropic doesn't have ads because it doesn't have consumers. And so ads would play no role for it, whereas OpenAI has hundreds of millions of consumers, many of whom use the service for free. So ads make sense. But the more serious point underneath that pushback is to do with the content of Amadei's ads are basically fundamentally dishonest and deceitful. He's making the point that OpenAI plans to have the AI tell you different things based on what it's trying to sell you, which is just not true.
But it's- When was the last time, Keith, you ever saw an honest ad? By definition, those are ads by definition are dishonest. So I don't know why we should object to that. Altman had an interesting ex post. He acknowledges that the anthropic ad is funny and it made him laugh. But he says, and it's interesting that the way he frames it, he suggests that most people can't afford Anthropic. It's too expensive. Only wealthy people use Anthropic. And therefore, the open AI advertising model is actually suited for less well-off people. It's an old argument. It was made back in the Web 2.0 age as well.
Yeah, look, I've advocated for not ads. I don't like the word ads because it conjures up exactly what Amadai is implying, but paid links. A paid link. if relevant to the context, can be a helpful thing. And I see that more and more actually in places like Instagram, where the ads are relevant to something I'm kind of interested in, and I don't object to them as much. So I think there's a smart way for Altman to be right. And I think for sure, Amadeus, it's a good marketing move by him, don't get me wrong, but it's, and you're right, ads are, do lie. But it isn't like the Apple 1984 ad, which was,
truly symbolizing a shift and a change in Apple's position in the world. This is just childish. I mean, it's like playground stuff. So it doesn't really... You don't draw the conclusion that Anthropic is breaking through to a new...
Yeah, it was childish, but Luckily for Apple, it reinforced a true theme in the real world, whereas no one in the real world thinks that Anthropic is looking out for the interests of consumers because it isn't a consumer company.
Well, I use it. I'm a consumer. I'm not sure that's entirely true. Altman's been ridiculed, according to the BBC, at least, for... what they call his online tantrum. Do you think it reached a nerve? Altman's usually a fairly smart, self-controlled guy. Why did he respond so angrily?
Well, it goes to the roots of his personality. I mean, Altman is... you know, a somewhat deceitful and devious person himself, probably more so than Amadai. And when you get under his skin, a bit like Trump, he reacts. And that's what happened there. You know, if you look at the other player, Demis Sabis, of the three, he's the most... you know, I'd say fundamentally intellectually honest of the three.
Well, it's interesting you put him in their company. He is, of course, not the CEO of Google, although he runs DeepMind. He's the Google AI chief. I wonder, Keith, whether one could argue that Altman was already irritated enough this week. Anthropics had a remarkable week, a breakout moment, according to the DFT, with the launch of... their new version of Claude, which has really riled the markets. How big a week has it been, do you think, for Anthropic? And do you think that's the real reason why Altman has lashed out?
Well, look, just to frame it, I think the answer to that is no, because literally the day after that Anthropic launch, ChatGPT launched version 5.3 of its model, which was a very similar leap as the Anthropic one. And indeed, yesterday, I spent most of yesterday playing with both of them. And I'm an Anthropic user, as you know. Are you a consumer, Keith? Never mind whether I'm a consumer, but I am now a convert to OpenAI. I used OpenAI to do the new... But you've always been a convert. No, I've used Anthropic forever. I don't use OpenAI because I see it as a consumer app, and I'm using it for real use cases in my work. But yesterday, I switched to OpenAI because the new is called OpenAI Codex, which is an app that runs on the Mac, not to be confused with the codex they've had for months. And it and Anthropic, and this is both of them, so this is the big story, both of them have introduced the ability for the interface, the chat interface to launch multiple workers, if you will, or agents in their language. So if you give it a task that has multiple skills, it will spin up an agent for each of those skills and they'll collaborate with each other. In the case of Anthropic, they literally talk to each other. In the case of OpenAI, there's a kind of a boss agent that each of the other agents talk to one-on-one. But they're all collaborating on your job and the outcome of the job. And it can take, depending on the complexity of the job, anywhere between 10 and 15 minutes to complete. And it replaces software. you know you don't you don't really use software anymore you use yeah i mean that's
the headline keith say it again it replaces software i mean that's a massive statement you've been hinting at this now for months but it's this seems to be the week where the promise of replacing software has become a reality with the launch of Claude Opus 4.6, and, of course, the new ChatGPT. A lot of the headlines this week have been about, and this one is about Anthropic, Anthropic triggering a trillion-dollar sell-off. When we talk about replacing software, it means that software companies are suddenly being dramatically hit on the markets and in every other sense.
Yeah, I mean, if we define software as... Take my creatorautomation.ai, which I built for myself with AI. I didn't use it this week because the OpenAI agents were better than my software. And that's me grandfathering my own software, or that's the wrong word, isn't it? Retiring my own software, which was custom built for my use case. And OpenAI now does a better job than my software. And by the way, saves me about two hours that otherwise I would have had to use. The other one is ClawedBot, which has now been renamed to OpenClaw.
is hort h-a-w-t well and oh oh i i i put it to work on my newsletter as well too i actually produced two copies of my newsletter one with open ai uh and one with open claw and uh actually a third one with claude and claude wasn't very good for my use case. It's probably good at other things. It's very good in Excel, for example. If you use the clawed plugin to Excel, it's fantastic. But both OpenClaw and OpenAI did actually a really, really good job. Now, the nice thing about OpenClaw is I can embed its persona in a Telegram account and I can talk to it and ask it to do things from anywhere in the world and it's running on my computer at home and will carry out my tasks. So when you put it all together, This week was a massive shift in how you use AI and how productive it can be for you.
Right. And let's talk about the impact on the rest of the market. As I said, some people believe that this latest version of Anthropics Claude triggered a trillion dollar seller. What would it mean to companies like Adobe, middle market companies? Look at the stock price. It's dramatically collapsed over the last few days. And they're not alone. The same has happened with PayPal and many other software companies.
Workday. I noticed that Workday, all of these companies have had massive reductions in their share price. I mean, in a way, Google is lucky to be in the AI game. If it wasn't, it would have had a similar impact. reduction based on workspaces and Google Docs and all that stuff. So basically what's happened is the stock market has moved away from software companies and SaaS companies as well. It can't yet move towards AI companies because they're not listed yet. Well, Google is. Google is. But actually, Google had a bit of a drop this week as well because it's a hybrid.
And, you know, I don't believe it will stay a hybrid. I think it's going to go the whole way and become an AI company, unlike, say, Microsoft, which had a huge hit this week because it's deemed to not be an AI company.
And Amazon. Amazon had a terrible week. I mean, just to put things in perspective, Google's numbers came out this week. They beat Microsoft. expectations dramatically and yet uh the stock declined even though they committed 185 billion dollars this year to investment in ai so it does reflect the complete shift in the way the markets are viewing tech companies and their valuations yeah
Oh, absolutely. And Amazon, I think, committed $200 billion. And their stock lost about 10%. And Microsoft, of course, is increasingly looking, of all these tech giants, the least... able to embrace AI, or at least...
Well, it's become a utility in the AI world, a bit like Oracle or Microsoft. It isn't a player except at the infrastructure layer, and it will be able to charge a rent for that infrastructure through Azure, as it does to open AI, but it isn't going to be the company that transforms our experience.
Yeah, I was going to bring that up. I mean, Elon Musk, as regular viewers, listeners to the show know in our conversations, I'm not a big fan, but you have to admire. I mean, he thinks very clearly on this. And according to The Economist, he's betting his business empire on AI by merging XAI and SpaceX. So he's part of this dramatic week, isn't he?
Yeah, I mean, I think if you measure the effectiveness of the CEOs or leaders in this space, he's the most definitive. He says, this is what's going to happen. Here's what I'm going to do. He then does it. And in his case, it's audacious. He wants to put apparently a million satellites into orbit, which are effectively data centers for AI processing, using the power of the sun to fuel them. And there's no doubt even that's a huge audacious goal. He reckons by 36 months from now, most of the power needed to run data centers in AI will be in space. Yeah,
it's interesting because Google this week came out with more initiatives about their AI energy race investing in earthbound energy. So that's going to be an interesting story. Keith, a couple of months ago, we did a... show on AI. It seems every week we do a show on AI. You argued that the AI race is a myth, and I argued that it's a bit of a wacky race. Every week, there's a new leader. Has anything changed this week? in terms of the order? We have Google, we have Microsoft, we have OpenAI, we have Amthropic, we have XAI. Has anything dramatically changed or is it just that AI is moving ahead of the rest of the tech world?
AI is moving ahead. You know, it's easy to forget this, but the first player who stepped up to this new level of functionality was Kimi. K2.5 is the version, which is a Chinese system that started this concept called agent swarms to put a swarm of agents against a task. That was three weeks ago, I think. So the last three weeks have seen Kimi, OpenAI, and Anthropic leapfrog Google, actually. So Gemini now is a generation behind again, and no doubt is already working on not being...
So it is like the wacky races. Every episode or every week, it seems to be a new leader. One pulls ahead, one falls behind. Is there a dick dastardly here, or is it...? I mean, for me, of course, it's Musk, for others it's Altman, some it's Dario Amidai, or does it really depend on one's own perspective?
Yeah, I vaguely watched that show when I was a kid, but I can't, I haven't got your ability to link it to what's going on, so. So I don't know who is dick dastardly in this context, the bad guy chasing down all everyone else and cheating is probably Amadai, if there is one, because he's, he's just using marketing and lying to try to look better. And when it when he actually isn't any better. In other words, he's as bad as everybody else.
Yeah. And the focus of your editorial this week is that the CEOs are not growing up. But does that reflect, again, the reality of our AI age or however smart AI becomes and it's becoming smarter and smarter? It's not making humans any smarter. It doesn't make Elon Musk or Sam Altman or Dario Amidai. And we should just accept that, that however smart AI becomes, we're always going to be a bit dimwitted. We're always going to watch dick dastardly and wacky races. Except for you, of course, who was always mature. What were you doing when you were watching when you were six, Keith? Shakespeare? Shakespeare?
When I was six. You missed Wacky Races, how? I was six in 1960, so I think I was probably watching Top of the Pops. So you did have a little bit of triviality. You weren't always an adult or as much of an adult as you are now. I certainly wasn't. But, you know, we lost the thread there a little bit when you asked me that question. I had it all teed up and now it's gone.
Are we dealing with these weirdly paradoxical realities that on the one hand, AI is becoming smarter and smarter, and on the other hand, we humans are becoming dumber and dumber?
Yeah, I've got it back. So look, I don't think we should be measuring AI's impact on whether it makes individual humans smarter. I think we're as smart as we are. We're all different. What it does is it makes us leverage AI tools to get more done hopefully done better not always true sometimes it gets done worse hence the slop word um and it doesn't make us smarter it just makes us better at using our smarts to get things done and and i think if it can achieve that
Well, so that is to do with two things. Let's take AI growing up. This leapfrog from a single chat interface with a single AI to AI that can carry out work using multiple AI agents that it manages is a huge change in what AI can do and how fast it can do it and how well it can do it. That's the AI growing up bit. the ceo's not growing up is the mud fight between altman and amadai basically when amadai talks about adolescence in in terms of the tech it's ironic because he's displaying adolescence in his use of advertising and altman is equally in his response showing adolescence when all of them have got better things to be doing they're actually contributing massively But having this side fight, as they do it, that's the CEOs not growing up it.
They totally can. You know, I mean, all of every entrepreneur, and I'll include myself in this in the, you know, in their learning phase, when they having their ass handed to them in all kinds of ways, react based on their past experience, either better or worse. When I was doing real names, roughly at their age, I remember thinking I could beat Microsoft in a fight. And I really thought I could. And I didn't. I lost and had my ass handed to me. And over time, you learn that not everything is under your control. And that is a maturing learning curve. They'll all get there because the world won't be kind to them unless they get there.
Well, one thing for sure is they've all got jobs. And there was an interesting, another interesting piece of this week's news. There's so many interesting stories this week, is that U.S. job openings dropped to a five-year low in December 2025, more and more pessimism on the job front. One big piece, at least, of media news was that the Washington Post essentially committed suicide, at least according to The Atlantic. Is this an important week in terms, again, you've talked about swarms and bots and smart AIs. Is this an important week or will it be seen, Keith, as an important week when it comes to what more and more people are believing to be a crisis of human labor?
I think it's the starting point of that. And most of these things are overreactions based on a misunderstanding of timing. I don't think timing suggests that anytime soon, meaning the next few months, you're gonna see a huge change in the job market.
So I think the answer is there's always an emotionally overreaction and the stock market is the extreme of that because it reprices things very early once it sees a trend. So there is truth in the reactions, but timing wise, probably an overreaction.
Coming back to my favorite television program, Wacky Races, it's made up of all sorts of interesting teams. And in terms of the AI Wacky Races, I mean, obviously we think of Anthropic and OpenAI and Google, Microsoft, Amazon, XAI. But I know for you, it's a more complicated race. The VCs are also participating. And in this week's editorial, you imply that they're not doing very well, that they're certainly not leading the pack when it comes to the AI race and this crisis of software. Can we talk, Keith, also maybe last week about the crisis of traditional venture capital as well?
a massive concentration of capital into a small number of hands in venture capital. Most of the money being raised and deployed is coming through about five funds. Those five funds are sucking the air out of the room. So the venture ecosystem thrives due to seed investors who find these companies before anyone even knows they exist. And the seed investors are being starved of oxygen. And many are threatened with being viable. And there's a lot of writing this week about that. Three very good pieces, one from Beezer Clarkson, one from Dan Gray. And I'm blanking on who the third one is. And they all are navel-gazing about the impact of this structural shift in capital for innovation. And I think that's a reasonable concern. I do think that really, really good angels and seed investors will still thrive. But anyone in the middle is threatened.
Yeah, and this is part... of this crisis. It's the middle market, companies like Adobe is in crisis. It's the five or six AI companies that are dominating everything, whether they're privately owned or publicly owned. There'll probably be two IPOs, Anthropic and OpenAI, this year. So what's happening with venture is a mirror and a cause, or perhaps a cause and a consequence, of what's happening more broadly in tech is that fair i mean it's it's a winner-take-all economy in every sense whether you're an investor whether you're a developer whether um whether you're an ex uh you're an ai company yeah and there's
some public companies in there as well and palantir's results this week were spectacular It did actually suffer a big rise and then a big drop in its share price over two days. You think of Anduril as the AI.
is not going to look anything like it does today. And if you're an investor, you want to be putting money into the future S&P 500, not the current one.
One company we haven't mentioned, actually, is Facebook. Are they a loser in this? Are they a footnote now in the wacky races? Do they even have a team in this?
Well, look, they're a fantastic company. Their revenues announced last week were huge, but their stock price has not gone up because in the AI race, they're also ran, trying to figure it out. They have a brand new team in place, building brand new platform to replace the Lama platform that they've had for the last few years. There's a big internal debate about open source versus closed source models. But with the amount of money they've got and time is on their side, they'll probably figure it out. I think they'll be a survivor, but they're a laggard right now.
You sidestepped the Washington Post question. Are you concerned that America no longer really has, it seems, a viable newspaper in its capital, especially in its... rather precarious political state in February 2026? You know,
I think the Washington Post has been challenged for at least 10 years since Kathy Graham um moved on from it i think i think it's it's challenged for a few reasons but the first is it tied its mast to the biden administration and ended up taking the wrong side of most discussions and became irrelevant in a similar way to the new york times by by abandoning the you know the the um being the source of truth and becoming a combatant if you will in the political wars And so I think America lost a news business. CNN tried to survive by being independent against MSNBC and Fox. But I think across the entire traditional media, you've had this partisan bifurcation of media, and they're all gonna be losers because of it, because most normal people want to make their own opinion from good journalism. And that hasn't been available.
But there's less and less journalism. I mean, isn't that the point of the crisis at The Post? I mean, whether you like them or not, they're capable of high quality journalism.
jeffrey epstein seems to be one well he of course he is one but for the entire media to get distracted from big issues by focusing only on this boogeyman and trying to undermine Trump only because of his possible associations, or likely associations, means that the entire narrative driven by the media's desire to sell clicks or papers or news or TV channels, gets distracted. You know, before it was Russia interfering in the election as the boogeyman. There's always a boogeyman. It's always about the media driving attention. And none of it is about what's really going on, which is so frustrating.
But what you call this attention economy is an advertising economy. And does that mean that maybe Dario Amidai is onto something suggesting that when you insert ads into AI, you're only compounding attention and trivialization on all the other things that have gone wrong?
Yeah, I'm definitely against what we would think of as ads in AI. So you're in Ahmadai's camp? For good reason. Even if you don't like its tactics. No, because I do know that OpenAI has also said it's against that, and that it has no intention of putting what we all think of as ads in AI, but it is thinking of working with... You know, if you think about the digital world, everything that there is available to sell, read, buy has a link. And so we live in a place where in order to have your ideas, products, services surfaced in any conversation with AI, it has to take the form of a link. And Amidai has links, Perplexity has links, OpenAI has links. And now they're saying, you can pay us to put your link there in our free service if and only if it's relevant to what the user is looking for. I think that's okay.
I don't think there's any intent, and it would be counter to OpenAI's economic future. There is no intent to interfere with the AI itself and what it says, does, thinks... at all. So that's where the Amidai ad is, just leveraging a nuance to try to depict something that isn't going to happen.
Well, there is other news outside AI. My tech company of the week is Spotify. I'm not always my favourite company. But their news this week was that when you buy a physical book now, and they've done a deal with a bookstore, you'll also get the audiobook. It's always been one of my irritations that you on amazon at least and audible if you buy the the audiobook you still have to buy the physical book so that's still innovation keith i don't know whether you're a reader of physical books or a listener to audiobooks but this is good news spotify is still innovating
Yeah, if you look at this blur behind my head, it's books and there's a whole wall of them to my right here. So I do read physical books. You know, that is innovation, but with a small eye, I would say. And it's smart as well. It's smart because I think the consumer will much prefer getting both in a single purchase. So I think I do think that's smart.
but in the name of democracy i gotta be fair and give you the last word with your startup of the week this is a startup that raised 20 million dollars from sequoia to build the cursor of crm sounds rather boring tell me about this company uh well
we all know that uh what CRM is, Customer Relationship Management, probably not everyone knows what Cursor is. Cursor is a rapper, not a singer, a rapper as in paper. A rapper with a W. around using AI to write code or perform other tasks. And this is an AI wrapper around your customer list and your prospects list. I don't know if it's any good, but the investors who invested in it certainly... What's its name? The name of the company.
Didn't you say the name of the company already? No, it's your Startup of the Week. It's your company, not mine. Startup of the Week is... I think it's Upstart, I think it's called.
But I will say, before we go, Andrew, I think the Post of the Week is the most illustrative. It's from Om Malik. And You know when we talk about music, we talk about music distribution going from the radio and vinyl to cassettes, then to CDs, and then to streaming. And then back to vinyl. And what Ohm has done is applied that same evolutionary scale, if you will, to software. And in his piece, he talks about the different eras of software distribution. I remember when I started EasyNet, we put a CD sellotape to the cover of the internet magazine. And we also gave the CD to Compact to put in the box when he bought a laptop. And software distribution was really hard. Then mobile came along. And suddenly there were two app stores and you could ship software to the whole world in five minutes. And it took the form of app stores. And then there was SAS, which is the cloud version of software for enterprises mainly through subscriptions or buying enterprise software. And what Om is pointing out is that software is now going to be built by users for their own purposes. And it'll be self-made. And I think his piece is very thoughtful and really good.
Yeah, software then becomes free in every sense, metaphorical perhaps and otherwise, although someone's going to profit from it. One other sector that's getting a lot of attention in the Super Bowl week is the prediction market, the way in which betting has become ubiquitous. Maybe we can talk about that another week. But let's add where we began, Keith, with the Super Bowl. not far from you, in Palo Alto, in San Jose. Who's going to win between Seattle? Do you have any sense between Seattle and New England?
Yeah, I'm not really aware of that. Well, this was the week that... broke the software business. I hope next week it will be equally dramatic, Keith. We need your That Was The Week newsletter. It keeps us up to date with a fascinatingly and dramatic changing world. So we'll talk again next week. Thank you so much. You're welcome. Bye, everyone.