U.S. Tech: The Future of Artificial Intelligence

As the advancement of generative AI takes off, how might this inflection point in technology impact markets, companies, and investors alike? Equity Analyst and Head of U.S. Internet Research Brian Nowak and Head of the U.S. Software Research Team Keith Weiss discuss.

----- Transcript -----

Brian Nowak: Welcome to Thoughts on the Market. I'm Brian Nowak, Equity Analyst and Head of U.S. Internet Research for Morgan Stanley. 

Keith Weiss: And I'm Keith Weiss, Head of the U.S. Software Research Team. 

Brian Nowak: Today, we're at Morgan Stanley's annual Tech, Media, and Telecom conference in downtown San Francisco. We've been here most of the week talking with industry leaders and emerging companies across the spectrum, and the topic on everyone's mind is clearly A.I. So today, we're going to share some of what we're hearing and our views on the rise of artificial intelligence tools. It's Thursday, March 9th at 2 p.m. here on the West Coast. 

Brian Nowak: All week, Keith and I have been meeting with companies and speaking with new companies that are developing technologies in artificial intelligence. We've written research about how we think that artificial intelligence is reaching somewhat of an iPhone inflection moment with new people using new tools, and businesses starting to realize artificial intelligence is here to stay and can drive real change. Keith, talk to us about how we reached this moment of inflection and how do you think about some of the big picture changes across technology? 

Keith Weiss: Well, thank you for having me, Brian. So we've been talking about artificial intelligence for some time now. Software companies have been infusing their solutions with machine learning driven type algorithms that optimize outcomes for quite some time. But I do think the iPhone analogy is apt, for two reasons. One, what we're talking about today with generative AI is more foundational technologies. You can almost think about that as the operating system on the mobile phone like the iOS operating system. And what we've heard all week long is companies are really seeing opportunity to create new apps on top of that operating system, new use cases for this generative AI. The other reason why this is such an apt analogy is, like the iPhone, this is really capturing the imagination of not just technology executives, not just investors like you and I, but everyday people. This is something that our kids are coming home from high school and saying, "Hey, dad, look at what I'm able to do or with chatGPT, isn't this incredible?" So you have that marketing moment of everybody realizes that this new capability, this new powerful technology is really available to everybody. 

Keith Weiss: So, Brian, what do you think are going to be the impacts of this technology on the consumer internet companies that you cover? 

Brian Nowak: We expect significant change. There is approximately $6 trillion of U.S. consumer expenditure that we think is going to be addressed by change. We see changes across search. We see more personalized search, more complete search. We see increasing uses of chatbots that can drive more accurate, personalized and complete answers in a faster manner across all types of categories. Think about improved e-commerce search helping you find products you would like to buy faster. Think about travel itinerary AI chatbots that create entire travel itineraries for your family. We see the capability for social media to change, better rank ordering and algorithms that determine what paid and organic content to show people at each moment. We see new creator tools, generative AI is going to enable people to make not only static images but more video based images across the entire economy. So people will be able to express themselves in more ways across social media, which will drive more engagement and ultimately more monetization for those social media platforms. We see e-commerce companies being able to better match inventory to people. Long tail inventory that previously perhaps could not find the right person or the right potential buyer will now better be able to be matched to buyers and to wallets. We see the shared economy across rideshare and food delivery also benefiting from this. Again, you're going to have more information to better match drivers to potential riders, restaurants to potential eaters. And down the line we go where we ultimately see artificial intelligence leading to an acceleration in digitization of consumers time, digitization of consumers wallets and all of that was going to bring more dollars online to the consumer internet companies. 

Brian Nowak: Now that's the consumer side, how do you think about artificial intelligence impacting enterprise in the B2B side? 

Keith Weiss: Yeah, I think there's a lot of commonalities into what you went through. On one level you talked about search, and what these generative AI technologies are able to do is put the questions that we're asking in context, and that enables a much better search functionality. And it's not just searching the Internet. Think about the searches that you do of your email inbox, and they're not very effective today and it's going to become a lot more effective. But that search can now extend across all the information within your organization that can be pretty powerful. When you talk about the generative capabilities in terms of writing content, we write content all day long, whether it's in emails, whether it's in text messages, and that can be automated and made more efficient and more effective. But also, the Excel formulas that we write in our Excel sheets, the reports that you and I write every day could be really augmented by this generative AI capability. And then there's a whole nother kind of class of capabilities that come in doing jobs better. So if we think about how this changes the landscape for software developers, one of the initial use cases we've seen of generative AI is making software developers much more productive by the models handling a lot of the rote software development, doing the easy stuff. So that software developer could focus his time on the hard problems to be solved in overall software development. So if you think about it holistically, what we've seen in technology trends really over the last two decades, we've seen the cost of computing coming way down, stuff like Public Cloud and the Hyperscalers have taken that compute cost down and that curve continues to come down. The cost of data is coming down, it's more accessible, there's more out of it because we've digitized so much of the economy. And then thirdly, now you're going to see the cost of software development come down as the software developers become more productive and the AI is doing more of that development. So those are all of your input cost in terms of what you do to automate business processes. And at the same time, the capabilities of the software is expanding. Fundamentally, that's what this AI is doing, is expanding the classes and types of work that can be automated with software. So if your input costs are coming way down and your capabilities are coming up, I think the amount of software that's being developed and where it's applied is really going to inflate a lot. It's going to accelerate and you're going to see an explosion of software development. I'm as bullish about the software industry right now as I've been over the past 20 years. 

Keith Weiss: So one of the things that investors ask me a lot about is the cost side of the equation. These new capabilities are a lot more compute intensive, and is this going to impact the gross margins and the operating margins of the companies that need to deploy this. So, how do you think about that part of the equation, Brian? 

Brian Nowak: There's likely to be some near-term impact, but we think the impacts are near-term in nature. It is true that the compute intensity and the capital intensity of a lot of these new models is higher than some of the current models that we're using across tech. The compute intensity of the large language models is higher than it is for search, it is higher than it is for a lot of the existing e-commerce or social media platforms that are used. So as we do think that the companies are going to need to invest more in capital expenditure, more in GPUs, which are some of the chips that enable a lot of these new large language models and capabilities to come. But these are more near-term cost headwinds because over the long term, as the companies work with the models, tune the models and train the models, we would expect these leading tech companies to put their efficiency teams in place and actually find ways to optimize the models to get the costs down over time. And when you layer that in with the new revenue opportunities, whether we're talking about incremental search revenue dollars, incremental e-commerce transactions, incremental B2B, SAS like revenue streams from some companies that will be paying more for these services that you spoke about, we think the ROI is going to be positive. So while there is going to likely be some near-term cost pressure across the space, we think it's near-term and to your point, this is a very exciting time within tech because these new capabilities are going to just expand the runway for top line growth for a lot of the companies across the space. 

Brian Nowak: And this is all very exciting on the consumer side and the business side, but Keith talk to us about sort of some of the uncertainties and sort of some of the factors that need to be ironed out as we continue to push more AI tools across the economy. 

Keith Weiss: Yeah, there's definitely uncertainties and definitely a risk out there when it comes to these technologies. So if we think about some of the broader risks that we see, these models are trained on the internet. So you have to think about all the data that's out there. Some of that data is good, some of that data is bad, some of that data could introduce biases into the search engines. And then the people using these search engines that are imbued with the AI, depending on how hard they're pushing on the search engines on the prompts, and that's the questions that they're asking the search engines, you could elicit some really strange behavior. And some of that behavior has elicited fears and scared some people, frankly, by what these search engines are bringing back to them. But there's also business model risk. From a software perspective, this is going to be the new user interface of how individual users access software functionality. If you're a software company that's not integrating this soon enough, you're going to be at a real disadvantage. So there's business has to be taken into account. And then there's broader economic risk. We're talking about all the capabilities that this generative AI can now do that these models can now take over. So for the software developer, does this mean there's job risk for software developers? For creative professionals who used to come up with the content on their own, does this mean less jobs for creative professionals? Or you and I? Are these models going to start writing our research reports on a go forward basis? So those are all kind of potential risks that we're thinking about on a go forward basis. 

Keith Weiss: So, Brian, maybe to wrap up, how do you think about the milestones and sort of the key indicators that you're keeping an eye on for who are going to be the winners and losers as this AI technology pervades everything more fully? 

Brian Nowak: It's a great question. I would break it into a couple different answers. First, because of the high compute intensity and costs of a lot of these models, we only see a handful of large tech companies likely being able to build these large language models and train them and fully deploy them. So the first thing I would say is look for new large language model applications from big tech being integrated into search, being integrated into e-commerce platforms, being integrated into social media platforms, being integrated into online video platforms. Watch for new large language tools to roll across all of big tech. Secondly, pay attention to your app stores because we expect developers to build a lot of new applications for both businesses and consumers using these large language models. And that is what we think is ultimately going to lead to a lot of these consumer behavior changes and spur a lot of the productivity that you talked about on the business side. 

Keith Weiss: Outstanding. 

Brian Nowak: Keith, thanks for taking the time. 

Keith Weiss: Great speaking with you, Brian. 

Brian Nowak: As a reminder, if you enjoy Thoughts on the Market, please take a moment to rate and review us on the Apple Podcasts app. It helps more people to find the show.