Good evening. Okay, this works well. Very quick little stop through PCT, PH Tech Trust, where I've been the manager since 2006. Let's tell you a little bit about PCT. So this is the team at Polar, I suppose a boutique, a bit bigger than that now, fund manager built by a technology team actually from the '90s that built a business around the franchise that I now find myself managing. That's PCT, that's PRW 500. But around we also manage two other products, and between them you can see it's a sizable, sizable business, which is lovely. It means we can invest in the opportunity set. We have one of the largest teams in Europe looking exclusively at tech.
Although I should say, about seven and a half years ago we launched an AI fund in anticipation of what I'm going to talk to you about today, which of course is central to our portfolio. But our excitement around AI, and we launched that, by the way, because we felt back then that at some point AI would happen, and then when it did, we wanted a vehicle that we could invest not just within the technology sector but also beyond it. And so you'll find companies in that portfolio like Walmart or Publicis, or companies that we wouldn't normally hold in a technology portfolio, for companies that we think would be beneficiaries of technology and AI.
The way we invest, this is SW, so it has to be very brief. We leave very early stage investing to our people, usually venture capitalists, who focus really exclusively on that part of the life cycle. And it's great because if you can do that well, you can find a Google or a Wiz or something that gets bought for a big premium. But the hit rate will tend to be lower, big return profile will be higher. We don't really enjoy our part of the market. We equally try to avoid companies at the late stage of their life cycle or themes at the end of their life cycle because they typically look cheap but almost always old. And like empires, technology companies and technology themes are either rising or they're falling, and tail growth rates, as we call them in the industry, almost always negative at the end of a life cycle. So we focus in this dark blue bit where things are happening, they're happening fast, people struggle to model them, which gives us opportunity as investors.
Where we are today is a very unusual moment, we think, and I always talk about AI now, as you could imagine. We liken where the AI story is today to something called a discontinuous technology change or discontinuous technology progress. Difficult to bring something like this to life, so we focus on skyscrapers, which for the whole of human history, people lived in buildings of about five or six stories, unless you lived in Yemen, which probably you didn't. But everywhere else you lived in five or six stories until steel and elevator came along, and very, very quickly the cities were transformed, and we lived in buildings that would have been once, would have been the conserve of fairy tales. And the people of 1885 couldn't possibly have understood quite how significant this moment was. In fact, in Chicago, where the first skyscraper was built, it's said that people would spend evenings watching them move around in the wind, anticipating, I assumed, that they were going to fall down. Nobody could prove at the time that they weren't going to fall down, but they didn't. And again, cities transformed. We don't think about skyscrapers; they're just part of the world that we live in today.
What's happening in tech and AI is not dissimilar. We've been making progress actually over time from, you know, winning at chess against humans and so on. But the AI industry didn't start yesterday; it started in the 1950s. But this real progress that we've seen has happened really since 2017 with a new AI architecture called Transformer that Google came up with and then freed to the world. And that's the architecture upon which things like ChatGPT have been built. And the transformer model has changed the parameter count, the trajectory of AI improvement very significantly. And you can't really tell on this chart because this is a log scale chart, but GPT-4, we don't know for a fact, but the parameter count, the sophistication of the AI model itself, is said to be a trillion, a trillion parameters, which is unheard of, literally off the chart based on history. There's lots more to come.
I don't know about any of you, maybe I'll just take a quick pause. How many people are using AI in a chatbot or something regularly? Oh, well, that's a good effort. More than I've seen at some other AJ Bell events. So congratulations. The pace of this is very, very fast and furious. It's quite hard to articulate that or to really demonstrate that. My colleague right now is in San Jose listening to the CEO of Nvidia, Jensen Huang. I heard him in January. Colleagues of mine were listening to Sam Altman and others just a couple of weeks ago. The pace of model improvement is furious. It's very hard to put into numbers, but if you think about the pace of Moore's law, which was really the underpinning of the semiconductor industry for the last 50 years, and we're moving at five times that pace. And the scaling of AI models is very difficult to grasp. Humans don't really like non-linear change, but unfortunately that's what you're faced with with AI. We're scaling in lots of different ways. It's hard to do that in this session, but lots of models that are coming from the likes of OpenAI, from Elon Musk's xAI, Anthropic, obviously Meta and Google are in there, the Chinese that we'll talk about in a second. But we're moving very fast.
And well, if you've used AI, you all know that. I've been using AI since the beginning, and recently have been using the full AI models from OpenAI, I think Research, where you're literally getting PhD-level help in a handful of minutes. Again, moving very, very fast, and lots of things that we believe will continue that should keep us moving at a very uncomfortable pace.
The market opportunity for AI is the biggest market opportunity because in the end, 60% of work in the Western world is knowledge work. And so AI will cut across all knowledge work. I mean, it will cut across all of it in ways that an agricultural revolution would have cut across most of the economy at that time, and the industrial revolution that followed. This is a knowledge economy revolution, of course. Now, I guess you can spell out some very negative things about AI. It's going to cost jobs and all that stuff. But the things that we do today mostly didn't exist in 1945 anyway. I'm pretty confident that humans are going to find better things to do with that time. We're going to probably enter a much more productive period for humans, and surprise people for a very long time.
And there's a few examples here where we expect AI to be felt. But if you look at the education vertical alone, if AI was to capture just 2% of the money that's spent on education, that's a $6 trillion market that we can capture unavoidably. That's just one example. A tiny little bit of that market comes to tech, and that's a new market opportunity for white companies. And this is going to happen much faster than people think. OpenAI already has 400 million people using ChatGPT every day. It probably has 700 or 800 million people using it every month. That's faster than any company that we're aware of. It took Uber 70 months to get to 100 million MAU. It took OpenAI two. This is a history of technology diffusion, and I hope you can see that each successive generation of ground-breaking technology takes less time to happen. And of course, this is going to happen much faster. ChatGPT for most people is going to be accessed by smartphone. There are six billion smartphones, and everyone now is a data point. So very much sitting on the shoulders of the giants: data that's already been digitalized, models that have already been trained, and a very easy conduit in the form of a smartphone. So this is going to happen much faster than people think, much faster than the internet.
And if you listen to companies, this is not, you know, at the end of last year, people said that it's probably past peak, people have stopped talking about AI, it's becoming old hat. And yet in the last quarter of the earnings, we saw more companies than ever talking about their AI efforts. There's some phenomenal data points from companies like Walmart that talked about how it would take 100 times longer to do what they're using AI for if they had to rely on humans, or 100 times more humans if they didn't have AI. And lots of quotes here that you can talk to, or we can, are supportive to this idea that this AI story is a very unusual moment for civilization.
This is a really interesting use case. We're invested in a company, Axon Enterprise, the company that makes tasers, but it also makes the cameras that you see everywhere now, tube workers and policemen have. And this is a thing that's designed to really change the behavior of people. If you know someone's recording your exchange, people tend to behave differently. These devices also capture audio and video. And the number one reason that police officers retire in America is said to be that they don't like writing police reports. They take up a lot of their time. They didn't join the police to write reports, and frankly, based on how good their reports are, we can see that. And it's why they retire. Now, you can take the audio content from a camera and in six minutes generate 80% of the police report required. That is one of the templates, one of the ways that AI is going to change the way we work and enable us to do high-value work, which in the case of police is catch the bad guy, do the timeline.
All this really means, of course, isn't about capturing police reports. It's about an ability or capability of AI to increasingly tend towards human-level capability. And we don't know exactly when that's going to happen. People talk about something called artificial general intelligence. So we already have AI that's more capable than humans. We already have AI in certain categories that, you know, really are at the bleeding edge. But we don't have AI today that's better than all humans, and we certainly don't have AI that's better than all the things that humans can do. But we're making progress. And if you listen to people like Elon Musk, which with xAI, probably 2030 seems like a reasonable time frame for that kind of capability. A year ago, we thought it was 2030. While they're in for that, it was 2030. The timeline for AI to reach human capability is shortening. Again, that's very good news, I think, from an investment perspective.
In the portfolios, I want to just quickly touch on this. PCT was, I've been running this since 2006, as I say, and we've obviously been hugely excited about the technology opportunity, have been really for many cycles. And we've been proud of the fact that we've been able to build portfolios that are diversified such that you've had exposure to the most important companies at any time, but also built in a way that is designed at least to withstand the buffeting that you get in stock markets. And you tend to get bigger buffeting in places where you're concentrated. So we are obviously one industry, this is the tech industry, and so there will be periods where we get more buffeting, like we're in right now. We still try to build portfolios, despite being hugely excited about AI, we're still trying to build them in diversified ways. So here we're trying to lay that out. You can see 30% of the portfolio are in AI-enabling companies, so companies like Nvidia, which is our biggest position in the portfolio today. We've owned it for, I don't know, 15 years probably. But also companies in the cloud area, companies like Amazon, like Google, that for most people at least, you're going to access AI through the cloud. Unlikely, unless you're J.P. Morgan or maybe OpenAI, but you're unlikely to be building your own AI clusters anytime soon. You're going to be accessing AI through these AI factories. You're going to be accessing that through the cloud, so we can capture it there.
We can capture also in beneficiaries, companies in SaaS, I don't know, the cybersecurity space. We know that the earliest adopters of AI are the bad guys. Our guys aren't encumbered by compliance departments. That's a challenge for the people at the front. It's hard for us to change business practices, but this is like a golden age for the bad guys. The ability to be able to take your voice and be able to fake it, the ability to create email that looks real but isn't, phishing attacks, ransomware, these things are... be careful. Do what you do in the stock market, but be super careful because the bad guys are AI-enabled. So we can play that theme via cybersecurity. And then we've got some other stories in the portfolio that aren't AI-related, but we hope on the right side of history, perhaps the Shopify or the Netflix or Spotify that will use AI to make their products or to enrich their products, and hopefully that makes a better user experience and for us as investors result in higher prices.
I think that's all I had for a presentation. I'm hoping we have AI questions, technology questions even.
The AJ Bell... but I mean, I'm comfortable with that. This is all on the internet. You can go to our website. There's risk warnings, of course, there. But you absolutely can take a picture from my perspective. This one, yeah. Yes, this is the life cycle of not just the sector but also a theme. The characteristics of what we're looking for, in the end, we're growth investors. So the technology sector, I think, and again, I've done this for more than two decades and had good periods and bad periods, of course, but I think there tended to require a different investment toolkit to other industries. There's very little mean reversion. Why? Which I mean, I guess if you're a steelmaker, there will be periods in an industry cycle where you'll do well and there'll be less good periods. And so, have steel is a bad example because that's not been a great industry, but there are cycles and what have you. But in tech, when you tend to be doing well, you do well and you attract more talent and you can invest in the right areas. And then you tend to, success begets success. Money doing poorly, value investors all over you, and you tend to cut your R&D and having people that don't want to work there anymore, and the business starts to go into reverse. And this is what I mean by not mean reverting. So, for example, my wife's email address is a Yahoo email address. Some people in the room may still use Yahoo, but most people don't. Google won the search war, but Yahoo would have looked like a very interesting value investment for some people, but it wasn't.
But we're growth investors. We look for where the largest market opportunities exist. We look to find the best ways to harness that opportunity. So where we can invest in companies that may look optically expensive but have the highest growth rates, but also have the most exposure thematically to what you're excited about, but in the shower by the investments that are the best ways to access that investment theme. And in terms of overall profile, we don't do early stage. We only invest in companies really where we can see break-even within two years, and that's rare. This is a very, very cash-rich sector, high margins. So typically we focus on those companies. And again, it's a very sort of, seems very obvious to say these things, you know, focusing on the best companies is sort of a truism, right? But that's what we do try to do. We're not trying to be smart. This isn't a business that, but we don't care about going to the pub and boasting about the outsider that we want. We're looking to find the very best names and the very best areas to make sure we capture more than our fair share of the returns to the sector.
