The brick and mortar of AI

Ashleay: Welcome to the In Your Interest podcast. My name is Ashleay and this week I'm joined, as usual, by my chief strategist colleague Sébastien Mc Mahon, and we'll be talking about the growth of artificial intelligence and its impact on the stock market with Maxime Houde, who was our guest last week as well. Maxime is director and portfolio manager in thematic investments. So hello, gentlemen. How are you?

Maxime: Hey, Ashleay. Hi, Sébastien.

Sébastien: Hello, Maxime. It’s good to have you again. Last week we had a very interesting talk about the AI revolution, how we got to where we are today in 2024. This week, Ashleay and I, we thought it would be interesting to talk about, you know the “brick and mortar” elements of AI. Because before we have artificial intelligence everywhere in this world, we have to build the infrastructure for it. So, we thought maybe if you could shed some light on what's happening right now in real R&D development and how this is being built.

Maxime: It's a great subject and it's topical right now. I think we need to come back to Nvidia. Nvidia is the centre right now of AI. And the reason is simple: GPUs are at the core of this acceleration of advanced computing. And really what's happening is Nvidia last year, it was a monopoly. This year there's one competitor. But Nvidia still has a large competitive advantage over them. And so, just to put some numbers: last year, Nvidia was up more than 240%. This year – we're in March – it's already up close to 80%, and it's still trading above its five-year average multiple. So, there's a big acceleration of revenue growth for them. But when we think about it, what's really happening is that the data centre overall, they're moving from a normal computing architecture to an advanced computing one. So, GPUs are taking more share of the pie instead of data centres. And that's where Nvidia is benefiting.

Sébastien: And, you know, just to recap – we've been discussing that outside of this podcast for months now – so, Nvidia is a firm that is designing those chips. But those chips are actually being produced by TSMC, which is a firm that's based in Taiwan. So, there's a geopolitical element to this story.

Maxime: Yeah. So, when you think about Nvidia, Nvidia the business model is what we call fabless. So, they're really designing the chip and they're sending it to TSMC which is really the factory that's going to build it. What's interesting is: TSMC, like you said, it's based in Taiwan and it's right now one of the only fabs that is able to build chips for Nvidia. When we think about the level that we're going to right now, TSMC is able to do three nanometers to two nanometers in terms of the size of those transceivers. So, it's close to atomic level. And they're the only ones that are able to do it in the world right now. The US are really aware of that, and especially the Biden administration.

Ashleay: Right.

Maxime: They put in place what they call the Chip Act. And the goal was really to bring back this manufacturing process to the US because of the geopolitical risk behind that. So TSMC is supposed to… will build a fab in Arizona around ‘26, ‘28, depending on the time that it takes to build it. But yes, there is a risk. It's a global supply chain right now, and the goal is to bring it back to the US. But also what's interesting is that, yes, we've talked about Nvidia, but TSMC is benefitting. And when you need to sell equipment to TSMC, what we call semi-cap equipment, it's also a sector that has done really well. So, the overall semiconductor space has been on fire since the beginning of the year.

Sébastien: Yeah. And you also mentioned that TSMC they're looking to build factories in the US. And their ability to build smaller and smaller is really driving this revolution here. But if TSMC comes and builds factories elsewhere in the world, there's no guarantee that these factories will be as efficient as those that are in Taiwan.

Maxime: No, exactly. The Taiwan fab has something special all over it, I would say. In fact, there's a lot of questions about the moisture in the air, the temperature, like… will a fab in the US be able to be as efficient as what we have in Taiwan? And will we be able in this fab to build a two nanometer chip, for example? So that's a question, an unknown. But the US administration is doing everything that they can to bring back this piece of equipment to the US because it's the most sought-after thing in the world right now, and it's a geopolitical risk.

Sébastien: Yeah. So, there's engineering issues, there's geopolitical issues. And then, if you build those data centres, those data centres will have to be built. So, there's construction involved. It also needs a lot of energy so there's pressure on the grid. So, there's more economic factors at play here than what we can imagine.

Maxime: Absolutely. I think outside of Nvidia there's a big opportunity. And just to give you an idea, when we think about the impact on data centres, we need to rethink the architecture behind the data centre. When we think about GPUs, there's two interesting things: you need way more energy and it emits way more heat. So, you need to rethink the ventilation system inside of your data centres. So, do you go with HVAC? Do you go with liquid cooling? Which are two new technologies that were not really used in data centres before. So, there's a good opportunity there. When you think about the grid, we need more electricity. So, all the electrical equipment companies will also benefit from this increased demand. And as we built new data centres, it's going to be good for industrial companies overall because it's an increase in manufacturing spend. Now, to get back to your question about the grid, what's also interesting is that we know that the grid in the US is quite old and need to be revamped. There's also a big question about: we need to have renewable energy to power those data centres.

Ashleay: Right!

Maxime: So there's also a big opportunity in that front where we've seen something interesting in the last couple of weeks, where Amazon signed a deal with a utility-based company that is offering nuclear energy. So, they will build their data centre next to the nuclear plant. And so that's a new way of using the nuclear plants that we're getting, because it's a reliable source of energy, which is ideal for a data centre. So, we're seeing a bit of a revolution on that front, and that could propel just the theme of renewable energy. We know it's been involved in the electrification overall that we've seen in the US. And now we have this second layer of growth that could impact them.

Sébastien: So, electrification of transports plus this. So, you had an estimate for how many years it will take for the energy demand to double. Is it a few years?

Maxime: So, the estimate is that because of this increased demand from data centres, the energy demand will need to double in the next five years, roughly.

Ashleay: Wow!

Maxime: So that's a big demand. And so, we need to meet it with supply. And what we're seeing is that obviously renewable will participate. But now we're seeing nuclear as a big opportunity on that front.

Sébastien: So there's geopolitics, engineering, ESG… because if you want to produce all of this electricity, you know, in Canada, we have lots of renewable; in the US, of course there's some renewable, but there's also still a lot of gas and coal energy production. So, you're putting pressure on the theme of decarbonization of the economy. All of these themes are wrapped into it.  So, before we just have artificial intelligence everywhere and our life changes, we kind of have to move through these real-world elements, which are… there's multiple layers of challenges.

Maxime: Exactly. So it's really interesting because those two themes are building one over the other right now where AI adoption will bring new manufacturing. It will advance manufacturing in the US. We’ll need to build more fab, revamp the grids and use renewable energy. So, when you think about it, it's not just a situation where technology will do well. You can think about industrial materials, utilities also, real estate, because sometimes those data centres, there's a real estate company that owns them. So, it's really across the whole economy and not just technology that’s involved.

Ashleay: So when people are afraid that AI will “steal jobs” – not necessarily true. And, in fact, I believe we had been discussing a little bit earlier that they create jobs normally when there are big revolutions like this.

Maxime: Yeah and one interesting, and I know you've talked about this in the past, but one interesting piece of research, it's from Goldman Sachs, actually. They expect that roughly 300 million jobs could be automated from AI. Now when you look at past technological adoption cycles, what's interesting is when you had a situation where a technology eliminated jobs, the back of that is that they've created more jobs that they actually eliminated. So, I do think we're exactly in that same pattern where yes, the automated jobs could be replaced by an AI, but it may actually create new jobs that we're not aware of right now.

Sébastien: And that's what we discussed in the previous episode that, you know, in emerging countries, when they started automating supply chains, some jobs were destroyed, but more jobs were created. But it was different types of jobs, maybe more of a premium on university diplomas, let's say. So, it creates more inequalities. And also, how do you redistribute the economic gains from all of this? So, there's a redistribution element for governments to take into consideration. So, the social aspect of these revolutions also need not be underestimated. But very interesting times to be alive, to be honest, we're seeing a revolution in front of our eyes. I don't know how long it will take, but let's say our grandkids will live in a very different world than ours.

Ashleay: Yes!

Maxime: Things are moving fast.

Ashleay: Yep, yep, yep! All right, so that's all for today. Thank you so much, Maxime, for explaining artificial intelligence and how innovative it is and, you know, explaining the phenomenon to us. And thank you also to our listeners. Please don't hesitate to reach out if you have any questions. And we'll see you all next week. Loved this podcast? Want to know more about economic news? Follow our “In Your Interest!” podcast, available on all platforms, visit the economic news page on ia.ca or follow us on social media.