Celebrating a 20-Year milestone: navigating tomorrow's AI-driven markets

Every day is different, every month is different, every year is different, every cycle is different. What is the same is the greed and fear, and the behavioural mistakes that investors repeat again and again and again.

Q: Who will be the winners in the AI race?

There is a big debate right now between hyperscalers versus those companies that really provide the inputs, like semiconductor companies. So the question really is looking and evaluating who will be the structural winners going forward.

And when you look at it, it is really useful to look at prior disruptions. And by the way, we are not saying that AI is not real. We talked about this three-plus years ago and said AI is going to be very disruptive, and people are going to overestimate the number of winners and they are going to underestimate the impact — the magnitude of the impact.

So if you look at a company like Amazon and ask the question, which is spending 200 billion plus dollars, "Are they going to be the winner, or a company like Micron?" which is selling DRAMs to Amazon and all other hyperscalers, whether it be Meta, Microsoft, or Google.

In the last twelve months, the market is saying the winners will be those companies like Micron that are going through a very significant margin expansion, top-line growth, and returns versus their history. In the meantime, the market is penalising companies like Microsoft and Amazon because they are in the "spending camp" in capital.

So for us, our goal is not to find the winners in a 12-month window, but to find the winners over a decade or longer. Because that is, we believe, how you fundamentally, in the long term, generate better returns than the market.

Q: Do you think the AI boom and market disruption is different to other booms of the past?

Sir John Templeton said, "This time is different," is the most dangerous phrase in investing. And actually, I would like to remind people about Sir Isaac Newton, who I believe was an above-average IQ person. Famously, Sir Isaac Newton, back in 1720, invested in the South Sea Company and he made a 100% profit — around 7,000 pounds back then. And then he watched his friends making more and more money. There was a lot of momentum back then. With the fear of missing out, he reinvested in the company and lost roughly 20,000 pounds.

In today's money, it would be millions, actually, but it was 20,000 pounds back then. And then he famously said something along the lines of, "I can calculate the movement of the stars and the heavens, but not the madness of men."

Basically, every day is different, every month is different, every year is different, every cycle is different. What is the same is the greed and fear, and the behavioural mistakes that investors repeat again and again and again.

What is important is we are not saying AI is a bubble. What we are saying is AI is real, but it is creating a bubble in a narrow set of companies — around two hundred companies that disproportionately benefited. Out of those, I am guessing a very small, narrow selection — maybe two or three of them — in the long term will be winners. Most of them are really enjoying this extraordinary demand that is not structurally making these companies, all of a sudden, great companies. By the way, this is not the first time we are seeing this.

As a rational allocator of capital, our role, we believe, is to distinguish between structural and permanent versus temporary, and thereby bubbles.

Q: What types of companies do you think are the most defensible and what makes you confident in this view?

I always say we always follow the money. What I mean by that is we really talk to actual decision-makers in regards to any investment we make. Meaning, we go and talk to the very companies that are making these capital allocations to software companies and ask them a very simple question: "Are you ripping these businesses or companies out and replacing them or not?"

And the answer to that question really comes down to how sticky your product is. If you are a system of record company that is truly sticky in nature for users, then the answer is people are not ripping these out and replacing them with AI.

Yes, AI can write software, but the question is not just writing software. Remember, companies always, for the entire history of software, have had internal software that they developed and they have had external software that they purchased from other vendors. So I think companies will continue to do internal, company-specific software, and AI will definitely accelerate that process.

But we believe that businesses which have high circumstantial costs — where a company, by ripping off a product and trying to save 10, 20, or 30 cents on a dollar, can really risk their business operations and continuity — we do not think that these companies will be at risk.