Mark Odendahl
Welcome back to Industries In Motion podcast from RBC Capital Markets. Reminder, this is where we'll be exploring what's new and what's next in today's fast moving markets and industries to help you stay ahead of the curve. Please listen to the end of this podcast for some important disclaimers. My name is Mark Odendahl, and I'm Co-Head of Global Research here at RBC Capital Markets, and I'm really excited to have Matt Hedberg on the podcast today. Matt is RBC Head of Global TIMT research. Matt, welcome back.
Matt Hedberg
Thanks, Mark, happy to be here.
Mark Odendahl
So the reason we wanted Matt on the podcast today was Matt led a team of RBC analysts to publish an RBC imagine report on generative AI across many of the sectors that we cover in research, and that was a follow up to an RBC imagine piece that we did a number of years that had a theme of AI activated, We're now, obviously now seeing that… so we're really looking forward to hitting on some points that Matt and the team detailed in that report. Also, RBC as a franchise, has had a busy period of investor conferences, one of which being our private technology conference that we held a few months ago. That's where we had 40 private technology companies, primarily in the software and internet space, and AI was a big topic at that conference. So we have this multi sector report. We had this conference that had a lot of cutting edge private technology companies attending, and we have Matt Hedberg here to talk about it. So we're really excited. Let's, let's start with the report, Matt. Tell us a little bit kind of big picture, why you did the report, why the team did the report, the follow up, and then let's dig into quantifying AI in terms of some of the disruptive potential, specifically in TIMT.
Matt Hedberg
It's always fun to watch technology trends over the years. And generative AI, we think, is one of these next mega trends. You know, really the impetus of this report, it really started from our RBC Imagine report from 2021 that had the AI activated theme in there. And you know, really, from there, you know, Gen AI has taken on a whole new perspective, I think, in the world, whether you're an enterprise or you're a consumer, everybody knows what it is these days. And so, you know, really, why this report came about is, you know, we were giving a lot of these generative AI discussions, and obviously we have a focus on TMT, but a lot of times the questions were, ‘Yeah, but like, how is generative AI going to impact the broader economy?’ Whether we're talking about financial services, consumer brands, obviously, within TIMT, but within oil and gas, and that really sparked this idea for this cross sector view of generative AI, where we brought a number of research analysts together to offer their unique perspectives on how we think generative AI will shape the industry over the next several years. And just to put it sort of succinctly, we think generative AI has the potential to disrupt the entire ecosystem. We think it will significantly change the way businesses, communities, people, interact with one another and machines. And while we do think this will take several years to play out, we think the proliferation of generative AI will reshape generations ahead, similar to the way the PC, internet, mobile phone, cloud computing, social media, all these things have impacted us today, and so we are, I would say we are bullish on the technology change, and we think there's both positive changes coming from revenue generation or maybe improved margins. We also think there's the opportunity for disruption in terms of really changing the way sectors interact. We do think this is going to take time. And within the report, we also outline a number of the limitations which I know we're going to get into.
Mark Odendahl
I think I'll go right to where I think a lot of people are thinking right now, where are we seeing specific examples of generative AI driving revenue?
Matt Hedberg
Sure, yeah, that is, you know, it is the number one question that we get in terms of who's in a better position to monetize things, who is potentially at risk for disruption. And maybe to kind of think about this in terms of numbers. When we think about, the broader implications of AI, we're looking at about a 42% CAGR for generative AI revenue growth for through 2032 to over $1.3 trillion. Within that, we think 280 billion is from software, and that's growing 69% through 2032 so, we certainly think there will be revenue generating opportunities. And we think we're on the very, very early innings of this. But we think just given the rate and pace of innovation, the proliferation of GPUs, and I know we're going to get to this but, but the innovation coming out of private tech is really, we think, remarkable these days. We think that opportunity to generate revenue for these software companies will be significant. Within the report, we outline a number of companies that are that are showing real generative AI products today. Obviously it goes without saying, Microsoft, with ChatGPT, you know, with Copilot. So that's the obvious example, but there's others that are also, maybe less obvious to people, but things like within cyber CrowdStrike has a… an offering called Charlotte AI, which is effectively a Copilot for a security engineer to help them better execute their day to day activities. And as a security engineer, it's a real challenge to sift through noise and leveraging large language models to help these professionals do their job. We think is a really unique opportunity to monetize generative AI. There's others like ServiceNow, which is a name that we've known for a long time. They are also using large language models to improve the workflow. We also see it within companies like Adobe, that's at least that's leveraging generative AI to help creative professionals create content quicker. And so, you know, it's very early, Mark. And I would say that that the rate of adoption is still early, the rate of disruption is still early. But we think as these GPUs will proliferate, as organizations build their broader generative AI strategy. We think these revenue opportunities will proliferate and really change, I think the course of a lot of these companies futures in terms of being able to really monetize and provide value to consumers. So it's very early, but within this report, we outline a number of the use cases, specifically within TIMT. But some of the early green shoots are exciting for us,
Mark Odendahl
So that's great. So there's three public examples, right? CrowdStrike, ServiceNow, Adobe, all large, public software companies. Let's pivot now to our event, the private technology event. I had the benefit of attending and watching you host like 15 firesides in a day that I didn't know it was possible, but you survived it. What'd you learn?
Matt Hedberg
I learned a lot at that conference. And we learned that one executive races cars, semiprofessionally, we learned others are hang gliders, wine makers. I happen to be a Pearl Jam fan, which, which, many of you know, but so we did learn that. But we also, I think we saw firsthand, we hosted a number of vertical software companies the power of some of these models, and really when it comes to AI, the opportunity here is for these vertical software companies to leverage unique sets of data that are very different From horizontal providers like an Amazon or a Google, and it's because of this unique data advantage we think they have the power to train their large language models to offer very, very specific insight, whether we're talking about the transportation industry, whether we're talking about edtech, whether we're talking about the insurance agency or within the medical field. It really, really is exciting to see, I think some of the power of these vertical software's company emerge as they think about what is that really unique app that will set them apart? And we think that is a really, really exciting opportunity for investors to look at some of these disruptive ideas that you know are providing a really unique value to the market that others can't do, and that's a really exciting thing. And I think generative AI is turbocharging that to some extent.
Mark Odendahl
Yeah, it was obvious that some of these vertical experts have built in advantages for generative AI because of their data in the auto sector or the wealth management sector or, you know, the various verticals we had representative. But key to that is structured data. And there's, as you know, there's a lot of unstructured data out there. How are they going to help these vertical software companies turn that unstructured data into structured data so they can build AI models around it?
Matt Hedberg
It's a great question. Obviously, we all understand that the power of unstructured data, and that could be tweets that could be social media posts that could be text or audio, there's an untapped opportunity to leverage unstructured data and give customers and companies the power to provide unique insights into whatever it is they're trying to solve. And so this really gets into a whole next wave of technology. Effectively it comes down to this whole idea of data lakes and giving customers the opportunity to pull insights from some of this unstructured data. But ultimately, yeah, it's understanding that, you know, not all answers come out of rows and columns, but sometimes it's, you know, what's the sentiment of a call? What are some key words that that they're, that they're bringing up, that could, that could offer insights into, you know, either converting that customer to an upsell or saving the customer if there's some frustration, and that's all unstructured data. And so yeah, the power of these models are extremely interesting to us, and a lot of them are being built on the back of GPUs. And that we talk about in the report, that's one of the limiting factors. As we all know, the scarcity of GPUs is still one of the bigger issues out there.
Mark Odendahl
Now, let's pivot a little bit and talk about one of your specialties, and that's security. Let's dig into a little bit more. How are developers rethinking security fabrics to protect very sensitive, generative AI data?
Matt Hedberg
Yeah, it's a great question. And you know, when I talk to CTOs or software companies, and we ask what some of the bigger limitations are, to generative AI adoption, obviously, scarcity to talent or limited access to GPUs is a big part of it. But, increasingly, we're hearing from many CTOs and companies that it's just protecting sensitive data. Data governance, data privacy is of utmost importance. I've seen examples firsthand of LLMs being asked a question, and, you know, all of a sudden, the LLM is just bringing back sensitive data. Being able to lock that data down to understand that there's parameters around data retrieval is a huge concern for a lot of organizations, and it's causing. A lot of these companies to have to kind of pause a little bit and understand that we have to think are through our security fabric a bit differently as these LLMs open up data that maybe wasn't even visible previously. We talked about CrowdStrike earlier, offering a Copilot to a security engineers. But there engineers. But there's other companies, a small cap company called Veronis, is offering data governance and data privacy solutions that really help companies figure out where sensitive data sits and how that sensitive data should interact with these large language models. And so we think as time goes on, as we rethink the security fabric for this new world, we think that that could be one of the primary value unlocking opportunities to really then unleash the full power of these models. So, yeah, we're still early, and we're talking about more generative AI from us, from the customer lens, but also cybersecurity vendors are also building tools to better combat malware and cyber criminals, and so there's a whole like offensive angle to generative AI that can help these companies, become more agile in stopping some of these cyber threats even sooner.
Mark Odendahl
I can touch base on something you just said, what's it a specific example of kind of a software security software company going on the offensive.
Matt Hedberg
Yeah, you know, one of the key benefits of generative AI is code suggestion, so non-coders being able to write code is one of those aha moments that I think a lot of a lot of us realized early on. So we do think that some of these bad actors are also leveraging generative AI tools for their own code suggestion uses, and that, that, that we think that could proliferate a malware at, candidly, an unprecedented rate. And so, from an offensive perspective, you know, we, there's companies like Palo Alto Networks that's seeing this, increase in bad actors writing malicious code, maybe at a more rapid pace, and they are fine tuning their own AI algorithms to find where some of this new malicious code may be residing, and some of it is pattern recognition. Some of it is combining multiple data points across network, endpoint and cloud. It's another emerging threat. And so when we look at cyber not only could it be, helping to stop, some of these bad actors from an offensive perspective, but it also offers, companies a way to better utilize their own staff and perhaps not grow their cybersecurity staff as rapidly because of things like Charlotte AI that CrowdStrike offers. So yeah, it's multiple factors, and we think we're still in the very, very early innings of realizing some of the benefits here from a cyber perspective.
Mark Odendahl
If we could, if you could talk a little bit more about some of the other limitations for companies achieving some of their Gen AI, vision.
Matt Hedberg
We started the podcast out saying we're bullish on generative AI, we're, you know, a bit more, I think, realistic around the time to value. Data privacy is one of the, areas that could limit the growth of generative AI, but there's other things, you know, some of these things will just take time… like some of these models are hallucinating, and, you know, that's just going to require more data and more training. Some of these models cost a half a billion dollars to train. And so, you know, some of the the issues here is that, you know, some of the bigger vendors could get bigger over time. Potentially, from a regulatory perspective, you're seeing, you know, the top four or five tech companies capture a large part of the S&P market cap weighting. And so when we think about, you know, the power accruing at the top, it's something that I think a lot of customers think about as well. I think a lot of you know companies could get concerned, you know, if, if one vendor, or two vendors controls a lot of their IT fabric. And so we do think that some of these, these concerns, work their way out over time. But it's something that you know, we think, you know, is a big question mark for IT executives. One of the other limitations, is finding dollars. You know, a lot of these companies are working with relatively flat IT budgets, and trying to find dollars to fund these AI projects is also a challenge. And so that means de-emphasizing other projects, non-mission critical aspects of their tech spend get de-prioritized as well. And so we think some of these limitations will work their way out over time, but it's going to take time. And I think as that continues to work its way out, I think there's going to be a lot more comfort in unleashing the full power of generative AI in an organization.
Mark Odendahl
What about regulation?
Matt Hedberg
Yeah, yeah. I mean, it's a huge, it's a huge question. There's this question around, you know, the DOJ and potentially have to, you know, kind of watch some of these platform vendors continue to grow, but also start to, kind of question the accrual of power. Because it becomes really a race to get data, and those vendors that have the most data, ultimately, we think, win. And, you know, it begs a question, some of these smaller vendors, you know, how are they able to compete with some of the bigger vendors if they don't have the quality of data? Now, we've spent a bit of time talking about how vertical software is unique set of data that the bigger platform vendors don't have. But we do think that as time progresses, I think a lot of regulatory bodies countries will think hard, long and hard, about how where this power is accruing.. and does anything need to happen there to kind of level the playing field? Some of it could change, you know, from an M&A perspective, you know, there's been a lot of questions around, you know, M&A and you know, is the DOJ watching M&A more closely, because it could fuel, you know, further market share gains. But, it's one that we think is currently being debated.
Mark Odendahl
Matt, this has been fantastic. We're very fortunate to have you and the rest of the tech team so involved in this mega theme across all sectors that we cover at RBC. So really appreciate your time today. I want to end it with something you touched on earlier in the podcast, that you love Pearl Jam. Pearl Jam's, kicking off their summer tour. Where's the best venue to see Pearl Jam play?
Matt Hedberg
Now we're talking, Mark . For anybody who's seen a concert over the last several years, would there? Wrigley Field in Chicago is an absolute must. The other fun venue is Madison Square Garden. And, you know, they, they always put on a heck of a show there. So yeah, we're looking forward to a few of those.
Mark Odendahl
Well, this is this has been fantastic. Thanks, Matt and I really appreciate our partnership, and look forward to having you on many times in this venue to help investors navigate this opportunity.
Matt Hedberg
Thanks, Mark.
Mark Odendahl
What else lies ahead in today's ever evolving markets and industries? We'll keep track right here on RBCs Industries In Motion. Until then. Thank you for joining on this episode recorded July 8, 2024 Make sure you subscribe to Industries In Motion. Wherever you listen to your podcasts, if you'd like to continue this conversation or interested in more information, please contact your RBC representative directly, or visit our website to get more info on industries in motion for further insights. Thank you very much.