Oct 25, 2025 · 2025 #40. 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
What is a Browser? What is a Bubble?
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Andrew Keen
Hello, everybody. It's Saturday, the 25th of October, 2025. It's That Was The Week Day. Our summary of tech news with my old friend Keith Teer, publisher of That Was The Week newsletter. Last week, we talked bubbles. Keith, I think, is more skeptical that we're in a bubble than I am. But I think we... out bubbled ourselves today uh we're not going to talk bubbles part of uh his editorial this week is about bubbles he says this is not a bubble he may be right he may not but i think more interestingly in keith's editorial this week um is here are his thoughts on browsers and the nature of the ai revolution keith i generally don't like your AI created images, but I actually thought this image was a good one for your newsletter. So explain why you juxtapose this is not a bubble with and this is not a browser, focusing on the browser aspect rather than the bubble one.
Yeah, and this is not a pipe, it's a picture of a pipe. And, you know, when you look at this front page, by the way, I credited myself this week. I got AI to animate it, but it's all my own work. And what I was pointing at is that It looks entirely like a bubble. And the browser looks entirely like a browser. But this is a week when, and I'll focus on the browser, ChatGPT joined with others in having its own AI browser. It's called Atom, I believe. Is that right? Atlas. Sorry, Atlas.
It's so memorable. You can't remember its name. By the way, as you were speaking, I... scoured Wikipedia for that reference I missed it thinking I'm smart but of course you're smarter than I am and it comes up the René Magritte picture of a pipe this is not a pipe under the treachery of images so This is the treachery of images. What appears to be a bubble is not a bubble, and what appears to be OpenAI's new quote-unquote browser is not a browser. Is that fair, Keith?
Yeah, and the one on the right there, that's actually Google's Chrome. And the point is, up until this week, it was a browser, but the browser is now being redefined as a different tool, a tool that has integrated into it not just integrated, but stealing the front page of the user interface is an interface to AI. Google does have that, but you have to click on AI mode to get into it.
Yeah, and of course, all this is doubly ironic given the fact that Justice Department was focused up until a few weeks ago about splitting Chrome off from Google. which is certainly about fighting yesterday's war. Keith, you're a veteran of this. This is one area where you're really very knowledgeable and smart. What exactly is a browser? Because a lot of people might not even be able to define it.
A browser is, I mean, technically speaking, it's a rendering engine for HTML and various additions to HTML extensions, like CSS, which is the markup language that gives you the style, for example. So a browser is a rendering engine for content that is published in HTML.
Yeah. I mean, Tim Berners-Lee, when he invented two things, HTTP, that's the protocol that allows a server to be communicated with and to return content. That's the protocol there. And HTML is the language the content's written in. And so when Mark Andreessen first started life as a well-known techie, he was at a university in Illinois.
Right. And there were other browsers at the time as well. Two or three other ones. And then the browser walls happened. Microsoft, which had missed a beat.
Just before we get there. And again, I'm thinking out loud here, Keith. The browser didn't create the internet. The internet created the browser. So when we got the image of, I don't know, Magritte's pipe saying, this is not really a pipe. When we looked at a browser, what were we looking at? It looked like maybe a newspaper front page, but it wasn't, was it?
No, we were looking at, mainly in those days, the Yahoo directory. which itself was HTML, with links. And the innovation was that you could click on a link and switch to a different HTTP server.
Well, initially... it didn't have images. The first browsers only supported text and link. And the text could include bold and italic and stuff like that. So there was some design, but not very much. There was no capability of creating sections of a page. It was just straight down.
And Google... And so it's no coincidence that Google had Chrome. It's... Although... Google could have still been a pretty successful company without Chrome. Is that fair?
Yeah, it would have been vulnerable to the browser owner choosing a different search engine. So having its own browser locked in, ownership of both the browser and search. And Chrome came from nowhere, literally nowhere, to become the world's leading browser over a period of time.
And so just to go back to the 90s, and don't start telling me about your legal stuff with Microsoft, because we've done that before. But the battles... In the legal battles in the 90s, particularly between Netscape and Microsoft and then Google, these were the battles for power in this new age. Is that fair?
I'd say for commercial leadership and power and commercial leadership are intertwined, of course, but it's really for commercial leadership. And a lot of it is defensive. I think Microsoft was pretty aggressive.
I understand, but I'm talking, let's just try and sort of define our terms here to historicize it so we can get to our current moment. I'm just trying to think in broad terms. These were, I mean, would it be fair to call the battles between Microsoft and Netscape and Google and I guess AOL, were they platform wars?
a platform war, the platform being the web. So there was a single platform war that the browser was the tool that it was fought through. And of course, the government played a major role by suing Microsoft for monopolistic behavior, which at that point, Microsoft's browser had 98% market share worldwide.
Yeah, same as the operating system, which was the DOJ's real focus, because Windows was also very dominant. Apple was almost going out of business at that point.
Although, I mean, Microsoft's understanding of the world operating or the world depending on an operating system was wrong in the Internet age, wasn't it? Because the operating system of the PC age was being replaced by something else.
I wouldn't be as binary about that. I think there's a correlation between the operating system, the web, and the browser, and then the cloud. They all kind of emerge together. And if you can dominate the operating system and the web, you have a very good chance of being dominant in the cloud as well.
Okay, so for 30 years we had a history of, I don't know if you want to call it, platform, browser wars, Google, Microsoft, Yahoo and AOL were defeated, Google and Microsoft, and then in a weird way Apple, but it's a different kind of company, emerged. Are we now, Keith, when it comes... to the current situation where we're on the verge of an AI age. Are we at the end of the browser moment in digital history, do you think?
I mean, to come back to your Magritte image. At what point does it not become a browser? I mean, are we getting lost in semantic issues? Might we need to come up with another word? Because it's not really a browser anymore, is it?
It's an interface to content. A part of that will continue to be browser-like. Like the new ChatGPT browser, you can type a URL into it. and it will produce a web page, but that is not its default behavior. Its default behavior is to ask it a question and it will give you an answer, or to have a conversation and it will participate with you. And the web is subordinated to the next layer down, which of course a lot of people are exercised about. They believe it's the death of the web, not just the death of the browser. And furthermore, some people believe it's the death of the cloud.
Right, and I know one of your essays this week is from Anil Dash, often a critic of a lot of digital developments, who describes ChatGPT's Atlas, this new quote-unquote browser or post-browser product, the browser that's anti-web, although maybe it's the browser that's anti-browser.
Yeah, Anil has a history almost as long as mine of being a significant and thoughtful person in tech. And he is of that generation that believed the web was independent and therefore fantastic because it allowed people like you and me to publish and be seen without restriction. And his fear is that The centralization of interface around AI will mean that the web as an independent publishing platform degenerates. And the business models that live on the web also go away.
Yeah, I mean, the advertising model. You start your editorial this week. The web front door has been morphing into AI for a while now. This week it accelerated with Atlas. What is it? Again, maybe I've asked this question before and we've talked about it in previous shows. What does that mean? It's morphing into AI.
Well, it doesn't feel like as abrupt a change as, for example, using the original iPhone was compared to using a desktop. Or using email or Netscape or something like that. Yeah, there was a certain organic element to change. Up until this point, even the initial iPhone only used web apps until native apps were introduced later. So there was an organic element. This feels like an absolute change of paradigm. And not only a change, but a change for the better from the point of view of the user. So the victim here is the publisher, not the user. And so most of the narrative that doesn't like it is publisher centric narrative. You're a publisher. That was the week. Why do you like him? I like it as a user, as a publisher. You know, it doesn't really play a role for me except in creation. But, you know, I can't get my content distributed on ChatGPT or Anthropic or Gemini. because they don't have a way of letting me do that yet. That's the biggest weakness.
But aren't you trying to fight yesterday's war? The idea of thinking of these things, these post browsers, as vehicles to get your content on, it's the wrong question to ask. It's like thinking of television as if it's radio and ignoring the fact that you watch it.
That could end up being true, what I hope is true. and may not be, is that all of us who contribute to knowledge that AI is trained on, ultimately become referenced in AI, insofar as we're relevant to whatever is going on in the conversation.
Keith, you might say yourself, we're all part of humanity, whatever that means. You talk about the browser becoming an agent, not a link map. Again, what exactly does that mean?
Right. Now, so this week I downloaded the Atlas browser. And at the top it says, Ask Chat GPT. And if you click that, it opens a left-hand panel. Well, now you can be on a page and say, give me a summary of this page. Tell me the history of this person who wrote this quote, almost anything. Or if the page lends itself to it, you can say, book me a United Airlines ticket, San Francisco to...
When we talk about agents, is it like having someone... in your basement who is looking at a traditional browser and knows you and is making decisions that they think will benefit you one way or interest you.
Yeah. So it added two other things this week. The first thing, and Anthropic did this as well, it added memory. So you can turn memory on now, which means everything you've ever done is accessible by it. So it profiles you in a way. And they also did the same for their company accounts. They added a feature called Company Knowledge, which opens up, if you choose to, opens up your file system and document store so that you can ask questions of your own document base and email as well, by the way. So it's turning from...
I would have thought, I mean, I'm thinking out loud here again. I don't think you've got any links to this. But I would have thought in that sense, then Apple would have been really well positioned or are really well positioned. For example, I store all my, I do everything through Apple. It's the one company I trust. So they have all my emails, they have all my photos, they have all my music. So wouldn't Apple be certainly in some ways much better positioned than OpenAI or even Google for this?
Yeah, Apple would be very well positioned for a similar feature. Google is well positioned for company level stuff because Google Drive, Google Docs, Google Sheets, Google PowerPoints are all one click away now inside either Anthropic or ChatGPT. You just got to link them and then that is a knowledge base that can be drawn upon. And Apple, unfortunately for Apple, doesn't really have a corporate play with its documents like pages and numbers.
Okay, but they do have a consumer play. They have a consumer play, yeah. Which presumably will come out very soon. It has to. I think ChatGPT might bring it out before Apple. Yeah, but ChatGPT doesn't have access to my emails and Apple. What are they playing with? No.
Can you give ChatGPT access to Apple to your email? Will Apple allow that? Oh, yeah. Apple totally allows that. So this is the war that's happening, Keith. It's a war for information and access to data.
Yeah, where are you going to get the data? So for me, for example, I live and die on email. I always have done for 30 years. And I've been on Apple for almost all that 30 years. So if an agent had access to my email, it would know a lot more than I can know, be able to crunch and analyze and say, well, why haven't you talked to this person recently? And what are you thinking about this? So it offers a lot of exciting opportunities. You note in your editorial, it requires, and I'm quoting you here, the web's economics must reprice fast. We've talked about this before. One of the other links is, This week is from Fast Company about AI about to upend Google's AdWords cash cow. This is not the first or the last kind of article here. What does this mean? Again, we've talked about this. I've never quite understood it. The web's economics must reprice fast, not just reinvent itself, reprice. What does that mean?
No, you said web. I just want to emphasize that the internet's business model is largely business to business through the cloud, through various services that you subscribe to. The web's business model is partly subscription-based, but it's mostly advertising-based. And that advertising takes the form of you and me clicking on links that somebody is getting paid for. And insofar as the web is pushed down a level and the user interface becomes a knowledge conversation, those links don't get surfaced unless they do. Right now, the default is they don't, mostly they don't. And that could mean that the almost trillion dollar annual revenue from the web's business model will disappear unless it can transition to AI. And that's an open question. Matthew Prince is quoted this week in the editorial as asking the UK government to separate Google's ad platform from its search engine to make it so that Google doesn't own both, which is a little bit, I think, last year's news. What he should be doing is trying to put in place a paid links model that can know be beneficial to ai and to publishers and that's a bigger project and a longer term project but but that business model is either going to disappear or transition
and it's ironic of course that while ai according to fast company at least is upending google's adwords cash cow the core of its business The Google stock is up, what, I think it's almost 50% in the last year, 30% since the beginning of the year. So the markets are actually very bullish on Google.
Well, that's because Google's playing a very good AI game. This week, Anthropic committed, I think it was $10 billion to Google's cloud-based NVIDIA equivalent. It's called a TPU, which is the chip required to do AI training and inference. And so Google also now has this button called AI mode in Chrome. And if you click on that, it will change into a browser that looks more like that.
And it's the top of the page, at least the top of my Chrome. I don't know whether mine is personalized. Yeah. So it almost is forcing you or pushing you into... using quote-unquote Chrome as an AI agent.
Right, there's a very gray area between agent and browser.
Words and timings
Right,there'saverygrayareabetweenagentandbrowser.
Keith Teare
Yeah, and so you can see that Google deserves its share price because it's being brave in undermining its own business model. You know, probably not fast enough, I would say, but nonetheless, it is going in that direction.
And then I don't want to turn this into another show about OpenAI, but in terms of this latest release of the AI non-browser, OpenAI non-browser browser, does that make you even more bullish on OpenAI? Are they doing it right? Or are they just sort of doing things that you'd expect without any real economic business upside?
Well, they're adding... and arguably this browser is a feature, but it's a dominant feature in terms of user interface. They're adding features that add value to people like you and me to use it. Like my wife this week said, I showed her how to use it with a problem she was having. And she said, wow, maybe I should switch to chat GPT from Anthropic because she was impressed with what it did for her. So it's adding features that add value and make it compelling to use. And that is driving demand and adoption. I mean, you know, they're already very big. They're still growing very fast and their revenues are still growing very fast. Some people predict they're going to get to 300 billion in revenue, which is huge.
Well, some people, that sounds like Donald Trump, some people believe blah, blah, blah. It doesn't mean anything. Meanwhile, the issue of platforms and control and regulation is still very real. Interview of the week on my Keen on America show is with Tim Wu, one of the leading antitrust thinkers, lawyers in the previous Biden administration, associate of your friend Lina Khan, Keith. He has a new book out, The Age of Extraction, How Tech Platforms Conquered the Economy and Threaten Our Future Prosperity. He also has an interesting New York Times article op-ed is actually out today. Big tech's predatory platform model doesn't have to be our future. It basically makes the argument, it's the same argument that's been made for 20 years, that I actually was one of the people who developed this argument, is that companies like Google are increasingly monopolies on their platforms. But how does AI change it? Is Wu in his age of extraction and is thinking, do you think he's fighting yesterday's war? If this new world of post-browser browser may not even be a platform, it's something different.
Well, look, if it was a history book, I think it would be pretty good. But he presents it in his New York Times piece, at least, as a future-looking policy roadmap. And he particularly focuses on Facebook, Google, and Amazon as what he defines as big tech. And I think in that sense... Yeah, and I'm quoting him.
This is from the New York Times piece. It is simply too risky to count on just a few platforms to invent our future. Yet that is the bet we're making by sitting back and letting Google, Amazon, and Facebook dominate. But he could have written that 10 or 15 years ago. Today it sounds kind of, as you suggest, rather archaic.
Yeah, you know, I think Amazon is pretty invulnerable except to Chinese lookalikes like Temu. But Facebook is definitely vulnerable. And Google is doing a good job of trying to escape from its vulnerability. But if it hadn't made some of the moves it's made, it would have been super vulnerable. And it definitely isn't a monopoly in AI because it's a laggard. There are at least two companies that we always mention that are way ahead of it. So I think... This idea, well, you know, deep inside Tim Wu's intellectual framing is this concept that government can help growth. And he argues explicitly for that in the New York Times piece. I haven't read his book, but I did listen to your interview, and he seems to imply that government can be...
He's good, isn't he? I mean, you may not like Lena Khan, but you've got to give someone like Wu credit. I mean, he's articulate and very well-versed in all this.
He is, but to sympathize with him, you have to believe that government would be better at managing the future than big tech. And I actually think no matter how badly you think of big tech, it's a more reliable producer of the future than government would be. And so his framing alienates me. I don't want anyone to encourage government to get a more hands-on approach to tech because we'll end up being like the EU. where innovation is frozen and money doesn't flow due to regulation.
Yeah, there's a great deal of virtue, but that's it. I mean, he compares and he says, and I'm quoting him, history supports this conclusion about regulation. The golden ages of American building, the canal age, the railroad age, the electricity age, all featured flourishing economies that ran on neutral network infrastructure. Wu invented the term network neutrality. So he knows this stuff inside out. Maybe the idea of network neutrality worked for the web, or at least for the browser web that existed between the mid-90s and today. But what is network neutrality, even Wu's idea of network neutrality, what does it mean in an AI age, in a post-browser age?
Well, funnily enough, if you go back to that internet and web distinction, the internet, which is the network, as opposed to the protocol of the web. The internet is neutral. The very fact that we have OpenAI and Anthropic challenging the previous generation's massive companies is testament to the fact that TCPIP, the underlying thing the internet runs on, allows new companies to build content platforms and knowledge networks and compete with massive rivals And so ironically, he's right about network neutrality, but the rise of big tech is one of the benefits you get from that. I mean, you know, who did Google displays? Probably Thomson Reuters.
The AI age is still sitting on top of the internet, not the web as much, but also the web. And so the AI age is the next manifestation of the internet. By the way, the same for smartphones. They all run on TCP IP and they all support HTTP and email and FTP and all those protocols. So the underlying internet architecture and protocols is the big win that allows all these things to grow. And by the way, allows challenges to arise to replace them.
they're going to make the Googles and the Microsofts and the Facebooks and the Amazons look like small fry in terms of their power and wealth and monopoly.
I think that's a reasonable assumption. I mean, every generation of platform change has led to larger revenues sitting on top of the now new platforms. There's no reason to believe this will be different. It should be exactly the same. There's still a lot that can go wrong. There's still a lot of competition. It's not clear, you know, even though I talk about OpenAI being far ahead, I wouldn't go as far as to say they've won already. Well, that might be a little presumptuous, even for you.
Yeah. Yeah, we will see. Certainly Sam Altman gets it, but then Andreessen gets it as well. He was around at the beginning. It's a fascinating struggle. And then, of course, there still are people who just don't trust AI at all when it's banned. This week, there was a public letter. I'm always suspicious of public letters, probably because no one ever asked me to sign them, that included Steve Bannon and Steve Wozniak of all people, Meghan Markle and Prince Harry calling for an AI superintelligence ban. There's a general hysteria out there about AI, isn't there, Keith? It's kind of like ridiculous that anyone would care what Meghan Markle or Prince Harry or Steve Bannon would think about any of this stuff.
Well, they're also... Jeffrey Hinton's on it as well.
Words and timings
Well,they'realso...JeffreyHinton'sonitaswell.
Andrew Keen
The Godfather of AI was on my show a few weeks ago. Yeah, it's pretty odd, but I guess we didn't see that. We didn't see public letters from... We didn't see these... letters from heavyweight public figures when email and the internet and the web
came along did we so it is a bit different there were there were less uh prominent attempts like the the whole narrative that said the internet was porn and um it was a place where child molesters can hide there was all that that didn't end up changing anything um and i doubt this will change anything either but i also Andrew, don't think super intelligence is anywhere close. I do believe LLMs are fantastic in what they can achieve, but I don't believe that LLMs alone architecturally are good enough to be the end game.
Yeah, and I agree. I think even the term super intelligent, I think it reflects the less than super intelligence of the people using it. I would advise viewers and listeners, anytime anyone utters the word super intelligence, it generally means they don't know what they're talking about, doesn't it?
Yeah, I thought Andre Carpathy, which is in this week's newsletter, is somebody worth listening to on this topic. And he puts it 10 years away, and he doesn't think LLM is out there.
Yeah, 10 years in Silicon Valley terms means you have no idea. I mean, you might as well say 10 centuries. Yeah. Finally, your startup of week's an interesting one. Sometimes I don't like your startups of the week because they always seem uninteresting in the broader sense of things. But Polymarket is a really interesting company in the sense that it speaks to a couple of weeks ago, we did a show on the age of speculation. And Polymarket, this supposedly $15 billion company, What would you call polymarket? The Wikipedia definition is an American cryptocurrency-based prediction market. When I went to its front page, it looked like a kind of modern version of a bookie.
Yeah, it is. You bet on outcomes of real-world events, and it doesn't run the house. So there's no house, in quotes. Who are you betting against? Other people who are taking the opposite bet.
So you and I could have a bet about whether or not OpenAI is going to be a $15 trillion company. Yeah, we'd have to. How would that get? I would put it up. So what's the Polymarket business model? Will they take a piece of it? They take a transaction fee, yeah. But it's a sort of a peer-to-peer betting service.
It definitely speaks to what young people are up to, which is doing these parlay bets on all kinds of sports, which kind of connects to when you pick teams. What's it called? You pick a team, your virtual team in football and things like that.
Yeah, you and I support teams that are already fantasy. Well, finally, we got your post of the week. I didn't even know who this guy was until you told me. And it's Sam Altman's brother. He's about as un-Sam Altman-like as possible. Low-key. He's not selling me anything. What's Sam Altman's brother up to? And why is this YouTube clip your post of the week?
So Jack Altman runs a venture fund. And he runs a podcast called Uncapped And this is an interview with one of the founders of Koto Ventures Sorry, I'm playing it here, so it's getting in the way And he talked about, he was asked the question by Jack Altman How do you know a good company versus a bad company? And Thomas used to work at the talent agency, CAA.
And he basically made the point that startup founders are like talent. You know one when you see one. And because... you know, startups are so unpredictable. It really is all about the people. And this assumes, by the way, that the talent is kind of quirky in some way.
Yeah, it's a good interview. And he mentions Tom Cruise, suggesting that when he first met Tom Cruise, Tom Cruise owned him in three minutes and he was a forever fan. So I guess what it suggests to startup entrepreneurs, Keith, and this is an area you're both part of a new consultant, is that we all need to find our Tom Cruise if we're going to raise money. Is that fair?
The opposite of Tom Cruise is probably... Who's that?
Words and timings
TheoppositeofTomCruiseisprobably...Who'sthat?
Andrew Keen
Or maybe Prince Harry, someone who's... Nigel Farage.
Words and timings
OrmaybePrinceHarry,someonewho's...NigelFarage.
Keith Teare
Nigel Farage is the opposite of Tom Cruise.
Words and timings
NigelFarageistheoppositeofTomCruise.
Andrew Keen
Oh, you want to go political, Keith? Well, interesting conversation. Not a browser. We will learn what it is in the next... few months or years, lots to talk about. I'll talk next week, Keith. Keep well and keep blowing bubbles. Bye.