Artificial Intelligence: The New Electricity?

I think artificial intelligence has the potential to be the biggest productivity enhancer to the global economy since the rise of electricity.

When I think about themes and investing in global technology, I really focus on the part of my framework focused on finding companies innovating in secular growth markets. The most important theme to me today is the rise of artificial intelligence.

I think that in order to invest in the artificial intelligence stocks, you have to find the linchpin companies that are enabling AI in order to happen. These are primarily in the digital semiconductor space and the semi capital equipment ecosystem. The second trend that we're seeing in all of global technology all around the world is the continued transition from offline retail to online retail.

And there are great e-commerce companies all around the globe that are enabling that transition, whether it's in Latin America, Europe, the U.S., or Asia. And finally, we're always looking for the rise of cloud adoption. This is because every enterprise in the world is focused on having a strong software strategy and a strong data strategy. They need that in order to be competitive in this new software-intensive global economy.

Sometimes I get asked the question, “Why has AI hype really taken off recently? How has it captured the global zeitgeist as strongly as it has?” And there's really two innovations behind artificial intelligence that's resulted in a step function capability and what we've seen from the likes of chatGPT and these type of chat models. The first is on the hardware side. In semiconductors, we've seen the rise of the graphics processing unit, or the GPU, from companies like NVIDIA or AMD. The second is on the software side and the rise of large language models, specifically the transformer large language models. That's what the T stands for in chatGPT—transformer. Let's talk a little bit about the GPU. Why is it so powerful?

Well, if we took a typical GPU from the likes of NVIDIA or AMD and ran it against a typical CPU, or a central processing unit, we can see these differences. If I tasked a CPU with the problem of reading a book, say Charles Dickens’ A Tale of Two Cities, and find all the times the author says the word “the,” it would start on the first page—"It was the best of times, it was the worst of times"—and count all of the different “the’s” in the book. The GPU, on the other hand, would rip the book into 100 pages and read all those pages at the same time. As you can see, running those parallel processes that the GPU can do is dramatically more efficient.

On the other hand, we have large language models. What we've done with those is basically scour the entire internet, all of the written language, and thrown them into the GPUs and out pops almost humanlike interactions. That's why we've seen such a step function change in artificial intelligence capabilities.

I probably spend more time thinking about is AI hopeful or hyperbole than any other question today. That being said, there's a question of the long-term reality versus the short-term cycle, and that's why when we're investing in these stocks, we need to be careful what is being priced in to the market. That's why we always rely on our framework when we're trying to invest in this artificial intelligence space.

Probably the most important risk that I'm thinking about is the AI hype versus AI reality. I personally think that AI has the potential to be the biggest productivity enhancer to the global economy since electricity. All that being said, sometimes the stocks can get ahead of themselves, and that's why active management really matters.