Incumbents Under Threat

Speaker 3

Come on! I've got sunshine On a cloudy day When it's cold outside I've got the Monday morning Everybody sing! I guess

Speaker 3

Hello, everybody.

Speaker 2

It is Friday, March the 8th, 2024. If it's Friday, it must be that was the week. And if it's that was the week, it must be some terrible AI art, which Keith always leads with. Problem with this AI art for people watching, two things, Keith. Firstly, and the subject is a very interesting subject, incumbents under threat, governments, big tech and startups, the inevitable rise of the new. Your photo in it looks suspiciously young. Have you told the AI to make you look younger?

Speaker 1

Actually, that's an overlay on top. That's a real photo taken, I don't know, two years ago.

Speaker 2

30 years ago, was it?

Speaker 1

30, no, but.

Speaker 2

29 and a half and the other thing about this ai art it's got big tech in a roman roman uh amphitheater and then uh governments and the eu somehow symbolically situated that the ai has managed to misspell governments they've spelled it g-o-v-e-r-m-e-n-t-s which doesn't say much for the spelling of ai

Speaker 1

Well, you know, AI is terrible at words. Have you not noticed that?

Speaker 2

Whenever it's doing images... That's the headline. AI is terrible at words. So what's it good at?

Speaker 1

Actually, it's good at words when you're doing text. It's terrible at putting words into graphics. So the graphic engine doesn't understand words.

Speaker 2

So why do you persist with this terrible AI? Are you trying to undermine AI somehow?

Speaker 1

Look, let's firstly just agree it's terrible.

Speaker 2

You are. I've already said that.

Speaker 1

And I did take the first attempt. I was busy. This was like at 9 o'clock last night after I'd written my editorial. So there's no doubt I was lazy and unconcerned about how bad it was just because it kind of did a half-decent job. Do you know what I did, Andrew? Here's the good thing. I gave it my full editorial. And I said, illustrate this. And it came up with that illustration. And it's a fairly decent synthesis of all the different points in there.

Speaker 2

Maybe it's because your editorial is complicated and interesting. And the AI couldn't manage to take that complicated, interesting editorial and turn it into complicated and interesting uh vision anyway so let's we've done the critique keith let's focus on the editorial which is very interesting you talk in the editorial about there being three dominant players in the world today um the government uh startups and big tech companies are they fighting each other

Speaker 1

Well, you'd have to do a three-dimensional grid to show all the fights. Governments are disagreeing with governments. Big tech is resisting government, and government is fighting big tech. But big tech is also fighting big tech. There's inter-scene wars between them. And then startups are... know being described by commentators as unlikely to succeed due to the dominance of big tech on the one hand yet they are actually threatened by by by startups so there's no single pattern there's there's a lot going on and so the editorial's job really is to disentangle all of that and try to understand it so you use this term incumbents under threat the incumbents are

Speaker 2

government and big tech, or is it just government that's the incumbent?

Speaker 1

The incumbent really describes big tech. It comes from Kyle Harrison's essay, which is the first one in Essays of the Week.

Speaker 2

Yeah, it's an interesting essay. The age of incumbents. Distribution plus data become concentration functions, according to Kyle Harrison. What does that mean?

Speaker 1

So he's really talking about the bar for a startup to be even in the game has got super high and only big companies can afford it. So they concentrate data, computing power, storage in ways that startups can't approach. That's the thesis. And Kyle's an investor. So he's, in a way, being quite pessimistic about investing, saying that the chances of success are much smaller because of the power of incumbents.

Speaker 2

So are we, and Keith, you and I have talked about this before, and you see everything because you were and you remain an entrepreneur, and you're also an observer of investment. Are the 2020s different, and especially the age of AI, different from the 1990s or even the 2000s, when you could just start something in your garage, in your dorm room, a la Zuckerberg or Princeton, Page and Brin turn your company into a trillion-dollar monolith, Leviathan. These days, you need to have a few billion to even start, don't you?

Speaker 1

No, you don't. That's the myth. I mean, if you look at some of the very successful AI companies, Perplexity would be a good example. But I think a lot of the open-source AI companies as well,

Speaker 1

There's a lot of relatively cheap, if not free, resources out there to build AI. But you have to focus on applications that generic AI isn't focused on. So OpenAI, Microsoft, Google, they're all focused on artificial general intelligence. And they're investing very large sums, billions. and yes startups probably can't go there anymore but most of the money from ai isn't going to be made from the infrastructure any more than it was in the web the money is made by clever uses of ai on specific problems where it is a solution so um you know if if i tried to reinvent venture capital i'd fail but if i apply ai to scoring venture capital and selecting companies that score high i i can succeed because that's a niche no one else is focused on so i think it's it's really all to do with what you focus on one of the stories this week is this ongoing feud which has now hit the courts between

Speaker 2

elon musk and open ai it's really a feud between it seems musk and altman about who stole what from whom doesn't that reflect though the foundations of open ai uh are also the foundations of big tech it's not really a startup environment it isn't a startup environment if you're looking to build you know the future brain that's good at everything

Speaker 1

But it is a startup environment if you're looking to apply AI to specific problems. Like perplexity AI focuses on web search. And that's all it does.

Speaker 2

Which is pretty ambitious. I mean, they're directly going up against Google and everyone over the last 20 years that have tried to challenge Google's dominance in search have failed. So, well, perplexity. And they just got valued this week at, what, a billion dollars? So they're not a tiny startup either.

Speaker 1

Well, their revenue is only $10 million, I read. $10 million. So they're quite small. So they're valued at a very large percentage of their revenue. And that's because of the promise. And you've been using it, right, I think? Yeah. What did you think?

Speaker 2

I think it's interesting. I think it definitely offers an alternative to Google. But

Speaker 2

I'm still not profoundly dissatisfied with Google, I think.

Speaker 1

Well, it's a different thing. The other thing that came out this week is Anthropic launched the next version of what they call it Claude, Claude 3.

Speaker 2

Yeah, I've been playing around with that. But they have hundreds of billions of dollars of investment, Anthropic.

Speaker 1

They do, but they started as a startup. That's the key. It didn't start in Google or Amazon.

Speaker 2

But it started in OpenAI, so it's a schismatic company that came out of OpenAI.

Speaker 1

Well, the individual started in OpenAI, but OpenAI didn't fund him.

Speaker 1

So... That's what happens in the Valley. You know a lot of these people, but... smart people get together they sometimes are in the same company and then new ideas emerge because every time you you push against reality um you open up thinking that didn't exist before and then new things come out and the valley is very good at that in fact the elon musk thing you pointed to the fight with altman is so contrary to silicon valley's normal way of behaving. Normally, Musk and Altman would have a friendly difference of opinion and go their different ways and both build something and still be somewhat friendly. The fact that Musk is choosing to sue OpenAI and then OpenAI exposed all of the emails and messages that went between the two of them, which make Musk look really bad, by the way, I think.

Speaker 2

Um, but if he makes it bad in your eyes, then he must look really bad. Cause you're one of his few defenders.

Speaker 1

He certainly comes out as being completely contrived, not really having any principles and just trying to get his own way, like a spoiled teenager. Surprise, surprise.

Speaker 2

You thought otherwise, what did you imagine?

Speaker 1

Yeah. And you know, that, that probably produces the adrenaline. um that gets him and all the dopamine that gets him motivated for whatever he's doing next he i think he needs that um ron conway yesterday published an open letter from his angel investing fund silicon valley angel more or less championing the potential of ai and asking everybody to focus on that and not fight each other right keith

Speaker 2

Is what's happening now, this fight over AI, is history just repeating itself? Back in the late 90s, early 2000s, the incumbent was Microsoft. Google was the rebellious company, the disruptor. Is now perplexity challenging Google? Is Google really just Microsoft and perplexity the new Google? Is everything basically just replaying itself?

Speaker 1

Well, there's definitely a constant replay in technology because incumbents almost never survive. And that's why Kyle's Age of Incumbents is interesting.

Speaker 2

I mean, one could bring up IBM. One could bring up Microsoft. I mean, they've survived.

Speaker 1

Microsoft went to NC. Microsoft went through a near-death experience only to survive once it dumped Office and even Windows and moved to the cloud, which was more aligned with the times. And it still is a failure on mobile. And arguably, it's a failure on AI.

Speaker 2

I think it's now the most, I mean, what is it, a $3 trillion company? So the markets wouldn't agree with you.

Speaker 1

Well, the $3 trillion is all about the cloud success, which is considerable.

Speaker 2

But is Google... I mean, one of the other interesting pieces you linked to today in the newsletter is one by David Kifferman about how Google blew up its open culture and compromised its products. History is repeating itself here in the sense that Google has increasingly become a bureaucratized... company, not very spirited, the staff always in rebellion, weak leadership, bad products, basically like Microsoft, isn't it?

Speaker 1

It's interesting. Sergey Brin was videoed doing an hour-long talk last Saturday to an internal team working on AI. And honestly, I thought he was great. He was relaxed, thoughtful, honest.

Speaker 2

But he's not an officer of the company.

Speaker 1

But he's not an officer of the company. So I think what this is, it's the modern version of the fall of the Roman Empire, if you want to talk about history repeating itself. It's aging entities, especially successful aging entities. You could argue the United States is one of those.

Speaker 1

The leadership get complacent. They fear being unpopular, so they take steps to remain aligned with the public consciousness or the direction of public thought. And they lose any real vision or direction formed from that vision. And I think that's just age. It's age.

Speaker 2

Is that why you, on the art, you made yourself look younger? Exactly. Old people are not original, are not spiritual.

Speaker 1

Because I'm so comfortable with how I look, obviously.

Speaker 2

So we've got the problem of the startups now challenging incumbents, which have historically been themselves startups. What about the role of government? You cite a piece from the New York Times about how tech giants are now being forced to bow to the global onslaught of rules. This is different, isn't it? This didn't really happen, certainly not in the 1990s and probably not in the 2000s.

Speaker 1

Yeah, it's really the success of technology in becoming global has led to multiple authorities trying to regulate it. And the opportunity in tech is global. So it would not be smart for a startup to focus only on a single market. The world is your oyster. And so that comes with regulation because we live in a nation state organized.

Speaker 2

And you're a big fan of nation states, Keith, aren't you?

Speaker 1

You know, nation states are... I'm teasing you.

Speaker 2

I know you don't really like them.

Speaker 1

uh it's not about liking them they're the status quo organizing principle of the world due to history they are increasingly also showing their age because they want to slow everything down and control it and and whenever authority wants to slow things down and control it the authority usually falls into disrepute And history is not kind to people who try to slow history down. So I hate nation states. I think nation states will ultimately not survive as the best way for the human race to organize the world.

Speaker 2

So say that again, history not kind to people who try to slow things down. Yeah. Like the Roman Catholic Church? Anyone, anyone, anyone. They were pretty strong because one of your arguments in the, um, against government in the editorial is you say, um, government's attempt, and I'm quoting you, government attempts to get technology under control cannot succeed because they do not have global reach. The end result with many governments all trying to control the new technologies, like the Italians with AI, is inevitable. The historical parallel that comes to mind is the government of Florence trying to hold back the formation of Italy. So that got me thinking about And when you look at the government of Florence, Florence lasted between 1115 and 1569, and an independent Italy didn't appear for another 400 years. So there's nothing inevitable about this, is there?

Speaker 1

It was way less than 400 years, Andrew. No, it wasn't. The Republic of Florence... No, I'm saying Italy didn't arise in the 1900s. Yes, it did.

Speaker 1

No.

Speaker 2

Well, this is not a show on Italian history, but the formation, an Italian nation state appeared, I think, in 1860.

Speaker 1

Yeah, exactly, 1860.

Speaker 2

Okay, so 500 years after the collapse of Florence. So my point is not to turn this into a debate on Italian history. Well, yeah. Firstly, that it's not inevitable, and secondly, it takes an awful long time.