Top tech investor on how to profit from artificial intelligence

Artificial intelligence is the hottest investment theme in the world right now, but Ben Rogoff, manager of the Polar Capital Technology (LSE:PCT) Trust, has been investing in the sector for a long time. He sits down with interactive investor’s Sam Benstead to discuss how he invests there, including which shares he currently likes related to AI in the software and semiconductor industries. He also explains why despite a strong year, AI shares are unlikely to be in a bubble.

• Invest with ii: Invest in Investment Trusts | Top UK Shares| Interactive investor Offers

Sam Benstead, deputy collectives editor, interactive investor: Hello and welcome to the latest Insider Interview. Our guest today is Ben Rogoff, manager of the Polar Capital Technology trust. Ben, thank you very much for coming into the studio.

Ben Rogoff, manager of the Polar Capital Technology Trust:Thanks for having me.

Sam Benstead: Artificial Intelligence, AI, is the hottest investment theme in the world right now. You invest in it, but what exactly is it?

Ben Rogoff: Well, the term artificial intelligence really relates to the idea of computers being able to, I suppose, behave like humans, to be able to think like us, and to be able to perform tasks. The pursuit of what is known as artificial general intelligence has been the domain of science fiction that computers with enough power and data would be able to analyse how a human being behaves across multiple disciplines and replicate the way that our species behaves. And I think it's fair to say that it’s been disappointing progress in AI since there was a focus on it post-World War Two.

There have been moments where it feels that we've made progress, whether it be IBM beating, I think, it was one of the chess grandmasters in the 1980s. More recently, we've seen the team at Google DeepMind solve for very narrow fields like the game Go and others solving for things like poker. So, some exciting advances there, also in protein folding not that long ago. But the excitement right now is to do with something called generative AI, which is a different class of artificial intelligence. Based on this thing called a transformer or a foundation model, they're different names for similar things, which has profoundly changed the outlook for AI. We can go into greater depth about this, but the excitement right now is about generative AI, ChatGPT, which many of your viewers will have used, and what appears to be a very sudden advance in the pursuit of AI.

Sam Benstead: And who are the big companies winning in this space, and what are the sectors that you should own if you want to be invested in artificial intelligence?

Ben Rogoff: The first thing to say is that ChatGPT was born out of a not-for-profit, OpenAI, which Microsoft Corp (NASDAQ:MSFT) very deftly invested in and took a minority stake, a very large minority stake, in that company. So, Microsoft has, I suppose, pole position as it relates to software companies, certainly in cloud companies as it relates to AI because of that alliance with OpenAI.

But there are obviously other companies looking to compete there. Google [Alphabet (NASDAQ:GOOGL)] has its own product, Bard, which is its large language model equivalent to ChatGPT. And that is obviously looking at the opportunities that's associated with AI, but also thinking about the risk to its search business posed by it.

But if we get deeper than that, the ability to make, or allow, a computer to analyse all the information on the internet pre-2021, which is what ChatGPT was trained on, requires an awful lot of semiconductors to store, to process. The training process is the way that people describe how a computer is able to learn, that process can cost $100 million, and the primary spend is on chips. So, a long time ago, McKinsey talked about how 40-50% of the AI value chain could end up going to semiconductor companies.

The recent quarter from NVIDIA (NASDAQ:NVDA), which is a very long-term holding of ours, and roughly around 7% of the trust today in NAV terms. The guidance that company just gave was the largest upside surprise in the history of the semiconductor industry, which really gives you a very strong sense of the excitement around AI and what some people are describing as the AI's arms race, as companies across the software space, but also within cloud, and you'd like to think across most sectors in the world, are rushing to understand the risk posed by, but also the opportunities presented by, generative AI.

• Why tech has another great decade ahead of it
• Stockwatch: could Nvidia be stock of the century?

Sam Benstead: Let's go a bit deeper into chip companies because they are quite a hard area to get your head around. So Nvidia, what does that do? What does rival Advanced Micro Devices (NASDAQ:AMD) do? What about some of the companies which build the chips or sell the equipment used to make the chips? Can we break that down in simple terms?

Ben Rogoff: Yes, we can, and I'll do my best. The nature of the semiconductor food chain is highly complex. But in broad strokes, Nvidia dominates the graphic processor unit market, the GPU market, which is really the primary semiconductor chip used to train computers to ‘understand’, I should put that in inverted commas, the huge datasets that are put their way, and be able to find relationships between words in the case of generative AI. So, Nvidia dominates that market.

Right now, most of the spend associated with AI is in the training and of compute clusters. And so that's dominated by Nvidia today. Its rival in the GPU space is AMD. Again, a very sizeable position for us, one of our biggest active bets in the portfolio and something we've owned for a very long time because the company has been taking back market share from Intel (NASDAQ:INTC) in the CPU, which is an adjacent, much larger market today, as the world has gravitated towards the cloud, so AMD has been a big beneficiary of a few things, more stress, the move to chip design, but really taking market share from Intel is the kind of punchline in [the] cloud server.

So, AMD’s own aspirations for the training AI market, we'll find out more. They're due to release their new chip, the MI 300, at the beginning of next year, so that's the plan there. More broadly, once we've trained these AI clusters, once we've done the training process then, when you go to your desktop, your smartphone, and you ask a question of a large language model like ChatGPT, that's called inference. That market also requires a large amount semiconductor content. It's estimated, we don't know exactly, that it’s somewhere between an $80-150 billion market opportunity, much bigger than the training market, and there the competition is going to be slightly more intense. There will be CPUs, GPUs A6, FPGAs, a panoply of different ways to do that [are] part of the much larger process. So, overall, you can see that the opportunity set is very large, as it relates to what is already an exciting end market or semiconductor market.

Not everyone's a player, not everyone's a winner. This is stuff that's happening at the leading edge, so there will be chip companies that don't participate here that perhaps are lagging edge, they're analogue-type companies. So, we're most focused on the leading edge and what parts of the manufacturing process become more critical as AI becomes a greater focal point.

• Ian Cowie: how to buy AI stock winners on the cheap
• The funds and investment trusts profiting from AI excitement

Sam Benstead: Is all the hype about AI justified? Are we going to enter a new industrial revolution linked to artificial intelligence?

Ben Rogoff: Again, I've been doing this for 25 years or thereabouts, and the nature of the beast is that as a tech manager, the hype cycles are almost endemic to our industry. It's how the industry raises capital. A great idea explodes on to the scene and gets some amount of traction, but never quite makes it to mainstream. And in that capital-raising process, the sector tends to overpromise and under-deliver. And again, the pandemic has really heightened that dynamic. Lots of things that we, as investors, but mostly the world, was excited by, home learning or an exercise bike in your spare room, lots of that stuff hasn't really lasted the trip now that we're reopening.

So, we've had more than our fair share of hype cycles during the past two to three years. I think investors are naturally skeptical of another one and here comes AI. So, it's a ‘how big could it be’. I studied history at university [and] we try to apply historical, I suppose, lessons to the current day and make sense of where we're at.

There are three historians in our team of eight, for what it's worth. And we are all agreed that this is one of those major moments. If you went back to ancient history [and] the development of systems of writing. If you went back to ancient Greece and systems for mathematics, these are foundational systems that have allowed humans to codify and disseminate knowledge, rather than word of mouth or cave paintings. And more recently, scientific revolution; in our lifetimes the smartphone, the internet, the ability to share knowledge, YouTube videos, how to change a tire of a bike, or news that flows in seconds on Twitter, we’ve lived through a process like this.

But the ability for a computer to be able to ingest all of human knowledge and be able to answer a question and give a human-like response to a natural language question is incredible. And for me, it’s a theme that we have to have exposure to because I think it does represent a key moment in the technology sector akin to the internet in the 1990s and, more recently, the smartphone and the cloud.

• AI stars Nvidia and Palantir: buy, hold, or sell?
• Be cautious of ‘AI tech hype’ as seven US stocks dominate returns

Sam Benstead: You’ve been running this trust for nearly two decades. There’s been lots of hype cycles over that period. How do we know this isn’t one of those? And what about some of the valuations of Nvidia, for example? It’s extremely expensive versus its current earnings. Why do you still own it given its price?

Ben Rogoff: First, the unfortunate, uncomfortable answer to the question of ‘how can we know that this isn’t another hype cycle’, is we can't. I think that reminds us, again, [of] some of the negatives associated with technology, the overpromising and under-delivering, the fact that a lot of things have come and gone and not really left any mark. I don't think that's where we are with AI, but it is a good reminder of why investors should go down the diversified investment path.

And if we're wrong, we have a very liquid portfolio of 90-odd stocks, and we can make changes to that portfolio that will hopefully insulate our investors from the worst if AI fizzles out again, like it has in the past. I don't think that's what's going to happen here. As for Nvidia and some of the other chip makers that the market's caught on to as beneficiaries, I saw a statistic the other day that said roughly 75-80% of the building materials of an AI server are currently accruing to Nvidia.

Nvidia is the go-to name in AI right now because it dominates the training AI market. Its forward price-to-earnings ratio (P/E) is somewhere around the 50 point. Again, optically not cheap, I think we can say. We're not selling any. That's not to say I won't sell some tomorrow or the day after, I reserve the right to change my mind, of course, but to date we've only added to that stock. On the day the company provided this incredible upside to their Q3 guidance, we added, [and] it felt like a moment when a drug company gets a phase three trial through. Something changed. That was the day when the AI story was born, really, in an investment sense. So, we've been adding there.

We've been adding to other chip companies and we've been looking for new names in Asia. The Asian supply chain looks like a fertile place for us to find less well-trod AI opportunities. So, overall valuations have been certainly lifted by excitement around AI. That's true for Nvidia, it's true for AMD, it's true for some other companies. But is that unreasonable, given what we've just experienced, given that we've just seen generative AI happen? We've just had 100 million consumers use ChatGPT in two months. Instagram, which most people have used, had 100 million people in two and a half years. The rate of diffusion is incredible, the scope of what this technology is capable of. I think Goldman Sachs talks about 300 million jobs being potentially at risk, which again, isn’t a positive thing in a societal sense, but it gives you an idea of the scope and the potential size of disruption represented by AI.

• Artificial intelligence: is the hype real, and how to invest in the winners
• Why you may already be profiting from AI

Sam Benstead: You've got 5% in cash. Is that normal or are you particularly bullish or bearish at the moment?

Ben Rogoff: Well, candidly, that's about normal for me. We've tended to hold cash in the portfolio, which I think is probably not what one is supposed to do on a textbook investment approach. But I've tended to use cash particularly when I've felt that overall valuations were not cheap. I would, in hindsight, wish I'd been more fully invested through my career, but the returns we've generated have been done so without leverage primarily and with a little bit of cash drag. That cash is there to provide us with firepower on days or weeks where the market's weak. I think we also use it to ameliorate what is always a portfolio beta well above one. In other words, the way we construct portfolios, the focus on growth stocks, our view, ultimately, is that returns should be driven by the growth of our underlying investments, revenues, cash flow and earnings. And what that means, of course, is that we tend to stay away from companies where they're not growing so fast or they're value situations or special situations.

So, the portfolio, through a cycle, we expect to grow faster than the index. That's how we typically would like to be delivering outperformance. But that leaves us with a portfolio beta well above one at times, and I use the cash to try to take the sting out of that.

Right now, the tech sector's relative valuation is reasonably high. It has been a very strong start to the year. We're trading at the high end, if not slightly above, the post-great financial crisis relative range, i.e., the forward price-to-earnings (P/E) of the tech sector divided by the forward P/E of the S&P. And so, with that in mind and running into earnings season, I think it's not inappropriate to have a little bit of cash.

Sam Benstead: Finally, the question we ask all our guests, do you personally invest in the trust?

Ben Rogoff: I absolutely do. I joined Polar on 1 May 2003, and on that day I sold my [existing] holding - I'd been working at abrdn - and rotated it into PCT. I've bought obviously more since there's been 20 years of Polar, and I like to try and invest in either PCT, or one of the other technology funds in our stable. We have three altogether, as well as a range of other Polar funds. I have a reasonable amount of my net worth in Polar funds, including PCT.

Sam Benstead: Ben, thank you very much for coming into the studio.

Ben Rogoff: Thanks for having me.

Sam Benstead: And that's all we've got time for. You can check out more Insider Interviews on our YouTube channel where you can like, comment, and subscribe. See you next time.