Show Transcript
The so-called AI trade has been a key driver for markets, especially among the biggest names in tech. But concerns about overvaluation have some investors looking elsewhere for opportunities. Justin Flowerday, Managing Director and Head of Public Equities at TD Asset Management, tells Greg Bonnell there are other ways to play the AI long game beyond big tech.
Print Transcript
[AUDIO LOGO]
* Concerns about overvaluation in the so-called AI trade have some investors looking elsewhere for opportunity. Well, my next guest says there are other ways to play the AI long game if you know where to look. Justin Flowerday, Managing Director and Head of Public Equities at TD Asset Management, joins us. Justin, welcome back.
* Great to be here, Greg.
* All right, so, obviously, the AI play up to this point has largely been concentrated on those people making the chips that power all of this. It is a long-term growth story. Let's start with productivity. I mean, if this is to be believed, then it is long term.
* 100%. And I think there is a story to be told where productivity gains will come back for the economy. The problem is, it won't happen immediately. And we've seen this phenomenon play out over time where you get this delay between a revolutionary technology and the productivity gains that come from it versus society and industry.
* And people refer to it as the productivity paradox or the productivity j-curve. And it's the gap. Productivity can go down, actually, a little bit before it ramps back up. Saw it with the steam engine 300 years ago, right? Economists would say it took 100 years for productivity gains to stem from that.
* Electricity during the industrial revolution. Managers and companies had to redesign their manufacturing footprint to use the distributed power. The PC in the '70s came out in, what, '73, '75. We saw 3% productivity in the '60s, 1% in the '70s, 1% in the '80s. It kind of rose back up to 2% in the '90s. So it takes time.
* For artificial intelligence, I would argue that it's going to happen sooner. I think that time period will be compressed just because of the speed everything is being done at and the investment that's going into the infrastructure to make use of artificial intelligence.
* All right. That leads into the next question, because that's a longer-term play. In the shorter term, if the chipmakers have had their moment in the sun, who steps in to take the spotlight?
* Right. Yeah, there'll be waves of beneficiaries from AI. And the next beneficiary could come from a whole bunch of different areas. I mean, one of the areas that's probably benefiting now and will continue to benefit is the utilities space. This has been talked about a whole bunch. I don't think it's new. It's not sexy.
* It's just the reality that we went through a 10-year period where power demand in the United States was flat. And we now have seen globally demand for power from data centers has tripled since the pandemic. It's up, I think, to about 20 gigawatts in the US today. It's predicted to go to 50 gigawatts by the end of this decade.
* And so adding to that, if you think about everything that's going on in EVs and electrification of industry and decarbonization of industry, I think there's a pathway for stronger demand. And it's not going to 10% growth, but it could be 2%, 3%, 4%, 5%. And that would be a real solid demand picture for utilities.
* You think about the other backdrop for utilities today, and the economy is slowing. Interest rates are coming down. The utilities have been flat for two years. The market's up-- US is up 40%. Canada is probably up 25% in the last two years. Tech's up 80%. Utilities are flat. And so there's an interesting story developing there.
* Watched the segment on the news last night-- I'm still in that cohort that watches the evening news to try to figure out what's going on in the world. They had a piece about a hospital using artificial intelligence to improve health outcomes for patients.
* Yeah. Look, this is one that I would put into, potentially, the short-term bucket but also the longer-term bucket. And an easy way to think about the benefits from AI are to put them into a couple of categories. You're going to have companies who are going to be able to reduce costs over time by tweaking processes. And you're going to have companies that are going to use AI to create new products and services that will help generate revenue. So those two buckets.
* And if you can find an industry or companies that are going to be able to tap into both, it's a really kind of positive thing. And I think healthcare is an industry that can do that. And it's related to drug discovery and the notion that we have the potential to improve the success rate of clinical trials.
* And so if you think about this, clinical trials for drugs, 9 out of 10 fail. So 9 out of 10 drugs that go into clinical trials fail. And they fail because we still have a very human approach to drug discovery. And it's very linear.
* And so you have a hypothesis and a scientist who predicts that some molecule is going to reduce some protein, which is going to increase a protein, reduce a protein. And we have a whole body of work that generally supports that. And we do a whole bunch of experiments. Eventually, we decide, yeah, we'll put that drug into clinical trials. 90% fail.
* And they fail because human biology is so incredibly complex. And there's all these feedback loops and all of these interconnections between proteins. And Chris Gibson, who's the CEO of Recursion, explains this much better. So listen to the podcast there.
* But there is the ability to increase the productivity or the success rate of clinical trials because of artificial intelligence, non-linear approach to analyzing the data, and the huge amount of data that's been available through the Human Genome Project, through experiments and trials that have been successful or unsuccessful, and all kinds of different databases.
* Over time, companies that can build data and a data structure and an architecture and apply AI in their R&D process are going to be very successful. And we could see, again, costs actually going down and new products rise in that sector.
[AUDIO LOGO]
[MUSIC PLAYING]
* Concerns about overvaluation in the so-called AI trade have some investors looking elsewhere for opportunity. Well, my next guest says there are other ways to play the AI long game if you know where to look. Justin Flowerday, Managing Director and Head of Public Equities at TD Asset Management, joins us. Justin, welcome back.
* Great to be here, Greg.
* All right, so, obviously, the AI play up to this point has largely been concentrated on those people making the chips that power all of this. It is a long-term growth story. Let's start with productivity. I mean, if this is to be believed, then it is long term.
* 100%. And I think there is a story to be told where productivity gains will come back for the economy. The problem is, it won't happen immediately. And we've seen this phenomenon play out over time where you get this delay between a revolutionary technology and the productivity gains that come from it versus society and industry.
* And people refer to it as the productivity paradox or the productivity j-curve. And it's the gap. Productivity can go down, actually, a little bit before it ramps back up. Saw it with the steam engine 300 years ago, right? Economists would say it took 100 years for productivity gains to stem from that.
* Electricity during the industrial revolution. Managers and companies had to redesign their manufacturing footprint to use the distributed power. The PC in the '70s came out in, what, '73, '75. We saw 3% productivity in the '60s, 1% in the '70s, 1% in the '80s. It kind of rose back up to 2% in the '90s. So it takes time.
* For artificial intelligence, I would argue that it's going to happen sooner. I think that time period will be compressed just because of the speed everything is being done at and the investment that's going into the infrastructure to make use of artificial intelligence.
* All right. That leads into the next question, because that's a longer-term play. In the shorter term, if the chipmakers have had their moment in the sun, who steps in to take the spotlight?
* Right. Yeah, there'll be waves of beneficiaries from AI. And the next beneficiary could come from a whole bunch of different areas. I mean, one of the areas that's probably benefiting now and will continue to benefit is the utilities space. This has been talked about a whole bunch. I don't think it's new. It's not sexy.
* It's just the reality that we went through a 10-year period where power demand in the United States was flat. And we now have seen globally demand for power from data centers has tripled since the pandemic. It's up, I think, to about 20 gigawatts in the US today. It's predicted to go to 50 gigawatts by the end of this decade.
* And so adding to that, if you think about everything that's going on in EVs and electrification of industry and decarbonization of industry, I think there's a pathway for stronger demand. And it's not going to 10% growth, but it could be 2%, 3%, 4%, 5%. And that would be a real solid demand picture for utilities.
* You think about the other backdrop for utilities today, and the economy is slowing. Interest rates are coming down. The utilities have been flat for two years. The market's up-- US is up 40%. Canada is probably up 25% in the last two years. Tech's up 80%. Utilities are flat. And so there's an interesting story developing there.
* Watched the segment on the news last night-- I'm still in that cohort that watches the evening news to try to figure out what's going on in the world. They had a piece about a hospital using artificial intelligence to improve health outcomes for patients.
* Yeah. Look, this is one that I would put into, potentially, the short-term bucket but also the longer-term bucket. And an easy way to think about the benefits from AI are to put them into a couple of categories. You're going to have companies who are going to be able to reduce costs over time by tweaking processes. And you're going to have companies that are going to use AI to create new products and services that will help generate revenue. So those two buckets.
* And if you can find an industry or companies that are going to be able to tap into both, it's a really kind of positive thing. And I think healthcare is an industry that can do that. And it's related to drug discovery and the notion that we have the potential to improve the success rate of clinical trials.
* And so if you think about this, clinical trials for drugs, 9 out of 10 fail. So 9 out of 10 drugs that go into clinical trials fail. And they fail because we still have a very human approach to drug discovery. And it's very linear.
* And so you have a hypothesis and a scientist who predicts that some molecule is going to reduce some protein, which is going to increase a protein, reduce a protein. And we have a whole body of work that generally supports that. And we do a whole bunch of experiments. Eventually, we decide, yeah, we'll put that drug into clinical trials. 90% fail.
* And they fail because human biology is so incredibly complex. And there's all these feedback loops and all of these interconnections between proteins. And Chris Gibson, who's the CEO of Recursion, explains this much better. So listen to the podcast there.
* But there is the ability to increase the productivity or the success rate of clinical trials because of artificial intelligence, non-linear approach to analyzing the data, and the huge amount of data that's been available through the Human Genome Project, through experiments and trials that have been successful or unsuccessful, and all kinds of different databases.
* Over time, companies that can build data and a data structure and an architecture and apply AI in their R&D process are going to be very successful. And we could see, again, costs actually going down and new products rise in that sector.
[AUDIO LOGO]
[MUSIC PLAYING]
