NVIDIA Corporation (NVDA) Earnings Call Transcript & Summary

Good afternoon. My name is Hayley Tabor. I am the Vice President of Global Industries at Dell Technologies. Welcome to NRF, and thank you for joining our session this afternoon. We're going to spend some time at the Big Ideas session talking about and focusing on how Target and Canadian Tire use GenAI for personalized shopping, driving revenue and improving productivity. So joining me today is Cari Covent. Cari is the Head of AI and Emerging Technologies at Canadian Tire. Canadian Tire celebrated their 100-year anniversary last year. From a single garage in Toronto to 1,700 stores nationwide, and all of those stores are within 15 minutes of 90% of the population. And if you haven't noticed by now, I do live in America, but I have a Canadian accent, and I am a consumer of Canadian Tire. And I will tell you that Canadian Tire retail is more than tires. Canadian Tire now has divisions in seasonal, home improvement, automotive, play and does a whole lot for the people living in Canada. So welcome Cari. Also joining me today from Target is Melissa Ludack. Melissa is the Vice President of Data Science at Target. Target for -- many of you know, those of you who are joining us from international regions, Target is headquartered in Minneapolis. Target serves many, many customers through their 2,000 stores and target.com. What is very special, too, about Target is target from 1946 until today donate 5% of their profits back to the community. So thank you very much, also, Melissa for joining us. And the last panelist I'd like to introduce is a friend, and I often sometimes think we've known each other for a long time, Azita Martin. Azita is the Vice President for AI Retail at NVIDIA. I don't know that I have to do a big introduction for NVIDIA right now, but I would definitely like to make a couple of comments. I think it's interesting and maybe some of you don't know this, but more than 40,000 companies and 5 million developers use NVIDIA's AI platform, driving AI transformation across markets. So this is your panelists this afternoon. We're going to spend 45 minutes to an hour talking about GenAI. And I'd like to get started by asking you, Cari, to share with our audience a little bit about the work you're doing at Canadian Tire.

So I -- my name is Cari Covent. As Hayley had said, I'm responsible for AI and emerging technology strategy at Canadian Tire. And what that entails is the development and implementation of a robust AI strategy. And the 4 sort of tenets of that AI strategy are really using AI as a catalyst to shift the mindset at Canadian Tire. And how we do that really is by creating a robust employee engagement program that's focused on skill development, on training and also by providing them very unique opportunities to drive innovation using AI. The second tenet is focused on boosting employee productivity. The third one is really around improving our customers' shopping experience. And then the fourth one is developing and operationalizing a robust, responsible AI framework for Canadian Tire.

Thanks, Cari. I'd also thank you, Melissa, for joining us. Maybe you can talk a little bit about your role at Target and your responsibilities.

Yes, absolutely. Thanks for having me. At Target, I have the privilege of getting to lead our data sciences teams that build out our automated decisioning models that support primarily our marketing and digital businesses. And so you can see this come to life through the algorithms that power our search engine, the personalized experiences we deliver to our guests across e-mail, their interactions with the site, our app, the relevancy of the offers that we provide to them. And then my team also supports Roundel, which is Target's ad tech media business. And so my teams do a lot of forecasting, optimization and simulation work to help support that business.

Thanks, Melissa. And Azita, I know you work across retail in so many different regions around the world. Can you share a little bit about the work that you're doing?

Yes. So I'm Azita Martin, I'm responsible for NVIDIA's retail business and consumer packaged goods around the world. And really, our team works with companies like Canadian Tire and Target to ensure that they're building the most high-performance, responsive AI applications and build it in a very, very cost-effective way. So we have the privilege of being an adviser to retailers about what are some of the best use cases that drive the greatest business value because we work across so many companies, and just be able to share also that knowledge with many of you here. So thanks for having me.

Thanks, Azita.

I think we all know that since the introduction of ChatGPT in November of 2022, every CEO and every board of director is involved in discussions on how to best leverage AI and GenAI, and what are the risks and complications. But one thing we believe is for certain is that to stay competitive, every company needs to figure this out. I heard, and maybe some of you have heard this, a note recently online ChatGPT was an AI iPhone moment for AI. No other technology has gained 100 million users in 2 months. And for many of you, it was probably our kids first. So with that, I want to start the conversation today with asking you about the journey. So maybe, Melissa, I can start with you, and you can share a little bit about the journey you and your team and Target has been on around GenAI.

Yes. So we started working around our approach for GenAI a little while ago. And one of the first things we started with was executive immersion. And so we knew that we had to help leaders across the company, business leaders, technology leaders, understand the power of this technology, what it was, what it wasn't and the problems that it could help us solve. We knew we needed to have them bought in because whatever we were going to build had to support their short-term and long-term goals. And so those executive immersion sessions that we did at the beginning were really key to launching us forward. The second thing is we've had a lot of conversations with some of our great partners and learning from them, understanding what had worked well, lessons that they had learned, use cases that they had seen be really applicable in the retail space. And then also getting -- using those relationships to be able to explore a lot of different large language models. There's a lot of large language models available. And one of the things we knew is that we wanted to experiment with more than just one. We wanted to test out a lot of them. And so we brought in a lot of those as we were doing our experiments. And we were really clear with our teams throughout that experimentation phase around what it was that we wanted to learn. And so documenting what those key learnings needed to be coming out of these experiment was really important. And it also allowed us to set really tight time lines for our teams, sometimes as short as 3 days. And by giving them those tight time lines, we prevented them from getting bogged down in additional details, it allowed us to pivot quickly and adjust for us if needed. And I think that was another really big benefit that helped us get there faster, was just that mindset of experimentation and pivoting quickly.

Excellent. And Cari, so -- Melissa's journey, and I'm sure your company's journey, the journeys are different. But can you share a little bit about the journey that Canadian Tire has been on over the last, what, year or so, right?

Sure. I think it's important to note that Canadian Tire and Target, and probably many of you, have been on the AI journey for quite some time. And so our approach to GenAI or ChatGPT didn't change much in the sense that we really put humans at the center of everything we do. We take a human-centered approach to how we solve our problems. And we believe that AI is really the opportunity, whether it's GenAI or traditional AI, to solve these problems in a different way. So I think that's important. When ChatGPT sort of became widely accessible in November of 2022, it obviously caught our attention. And I would say that month of December, immediately following that, we were playing with it. We were trying to understand what are the opportunities and what are the downfalls of it. And then shortly following that, we all sort of went off on holidays, came back, and we had a significant amount of executive support to really say to us we want you to use this in a meaningful way. We see there's opportunity. So the month of February, we -- and this is 2023, we spent that month working across all of our business units to really identify some high-value use cases. And those high-value use cases weren't just about achieving tangible benefits, but it was also around how can we use this technology to transform the way we work. It was focused on what are the learnings that we're going to take from it and how do we build up our muscle so that we can move fast with this technology. At the end of -- we identified the high-value use cases, we went into a very quick but extensive discovery process on those use cases to really understand the desirability. If we're going to use this technology to build an AI solution, will our stakeholders actually use it. Stakeholders could be our customers, our employees, our dealers, our suppliers. So that was number one. Number two was, is it viable? So will it achieve benefits that are tangible and also those benefits that I talked about from a learning perspective. And the third one, which is equally as important, is it feasible? Do we have the data and is the data clean? Can we use it to actually accelerate the solving of this problem? Do we have the integrations and the pipelines built? And do we have a cloud-based environment that will actually allow for the seamless and scalable consumption of the output of this AI model? So out of that, we came up with 2 use cases. One is focused on improving our shopping customer experience, and then the second one is focused on employee productivity. And if you recall, what I mentioned earlier, those line up very closely to our AI strategy as well as to our overall business strategy.

And so in your examples, there's internal use case, there's external use case. Melissa, can you share a little bit with us about the use cases at Target, like same kind of thing, internal, external? Maybe you can share with the audience what you're thinking about from a use case perspective.

Yes, absolutely. So we have been working on quite a bit on our product detail page. Doing a lot of things that you're seeing in the industry around review summaries. But the one that I'm really excited about is the work that we're doing to rewrite our product titles and product descriptions to be optimized for search engine performance. And the performance we get out of that, but also the scale that we're able to achieve is just phenomenal within that space. And so that is one that we're currently working on that I am really excited about. Second one that we are working on actually comes out of hackathon that we did. And so we -- when I talked about we had done an experiment where we gave our teams only 3 days, that was a hackathon, where we took our technology teams that are supporting our ad business and we ask them to go after a couple of key business problems to see what they could solve during that time. And we gave some access to some large language models and -- to take a look into and see what they could come up with. They actually were able to build a performance insight engine that can actually pull out insights on how well our ads are performing and then provide that back to our media analysts a lot faster than some of our traditional techniques have done historically in the past. And that's another one that I am really excited about. And then lastly, I'll say, we have had chatbots at Target for a really long time. They're primarily theoristics-based, and they're a clinical tool for our internal team members. Large language models, this is their bread and butter, and so this is a natural place to apply them. And so we're working on improving a lot of our chatbots that we have for our internal tools.

So Azita, I mean, obviously, you're talking to retailers and probably have a ton of experience and visibility on many use cases. And I know when myself and my team are meeting with customers, everyone wants to hear about the use cases. Can you kind of share what your thoughts are and what you're seeing?

Yes. I mean it's really summarizing what Cari and Melissa said, right? I think a year ago and very beginning, most companies said they wanted to start with employee-facing use cases, right? Because they didn't know how generative AI was going to perform, is it going to hallucinate and so forth. And so many, many companies started with using generative AI for employee productivity and any use cases like cogeneration. I mean, in many ways, generative AI has democratized building AI solutions. The second one, which I'm equally as excited and Melissa mentioned, is helping content managers write more descriptive product descriptions that leverages meta and attributes and all of that, and that does improve SEO optimization, and we want to help all of our customers achieve that. And so those are kind of examples. And a third one that I'm super excited about is around creating marketing and ad campaigns using, not only large language models but a lot of multimodal capabilities, which is videos and 3D rendering of your products, especially if you have luxury products and so forth and videos, how do apply makeup. So those are kind of some really interesting use cases. And then, of course, the customer-facing examples that drives revenue. And number one is, again, what both of you talked about, is shopping adviser, right? Imagine having the smartest, the best sales person, sales associate in your company talking and helping every single one of your customers, every minute find the right products that they're looking for. That's going to have an unbelievable impact on revenue. And most new retailers, their revenue is in the billions, right? And expectation is that actually, if you're using generative AI, you can actually increase revenue by 10%. That's a study that was recently done by Accenture. So these are all just very, very exciting use cases that many, many companies are using. And I want to encourage all of you to make sure that you're leveraging this kind of technology.

Yes. We at Dell, we see GenAI have a big impact on our business. We're building AI capabilities into our products. We're building platforms for our internal use. And we're also working with a very broad partner ecosystem so that we can bring partners together that can deliver on the business outcome for those GenAI workloads. So we kind of also -- while we're looking at the improvement of productivity on our development and inside of our organization and sales optimization, we also spend a lot of time looking at how do we help our customers by bringing these examples forward. And it kind of aligns -- everything we're doing is aligning to our overall business objectives and transformation. So when you think about Canadian Tire and Canadian Tire's business goals, how are you aligning your GenAI initiatives? I mean you touched on it a bit, but maybe to give a little more about how that alignment is working would be great.

Sure. So in terms of kind of improving the overall shopping experience that's obviously foundational to Canadian Tire, and our customers are at the forefront of what we do. And so as I had indicated earlier, one of the early use cases that we've done is to build that shopping assistant. And the value -- it's going to be launched in the next few months, but the value that we expect to create is really by giving our customers an opportunity to interact with us in a very different way and get the information that they're looking for much quicker and also in a different way. And so we're very excited about that. And then the second piece of our business strategy is really around changing the way we work. And as I mentioned, AI kind of falls nicely into that. So the second use case that we built is what we're calling ChatCTC. And that is truly changing the way we work, where we have thousands of employees that are now using it, not just at an individual level to do some code creation or understanding errors in their code, or not just to summarize reports from a communications perspective or get insights, but what we're also doing is we're now giving teams an opportunity to use it. And the value that we're seeing out of that team usage is, for example, teams that look at online reviews, they can now use our ChatCTC to respond to that, obviously, with the right human oversight. Or we're giving our technical service managers an opportunity to use ChatCTC to understand incidents and to be able to respond to them more quickly based on incidents that had similar patterns or symptoms. And the third area that we're really excited about is using ChatCTC for our data governance and our data classification group, where they're using ChatCTC to classify data into the data fields, into domains and sub-domains and then be able to then use that to write description on that data. So in terms of time saved from a data steward perspective, which we have mainly across the organization, we're starting to see the real benefit. So the goal really is everything that we do lines back to our business strategy. We're not doing anything sort of off the side of our desk or ad hoc.

And I think that's really, really important, especially as Melissa, you had said at the beginning when you talked about your journey, is getting that executive alignment and buy-in, right? So flip side of that, I guess, looks like the piece that has to tie back to that, is that these GenAI projects and initiatives are tied to the goals of the business. Maybe you can share a little bit about how, at Target, you're aligning to that too.

Yes. So for us with AI, we have been building out AI for a long time, similar to Canadian Tire. And we have traditional ML models that are powering all parts of our business. And we will -- as part of our strategy, we will continue to refine those. We'll continue to build new ones with the right technique for solving the problem that's in front of us. And so GenAI became another tool that we had in our AI tool kit for how we wanted to solve our problems. And the way that we thought about it is really kind of on a spectrum, if you think about it from a strategy standpoint of, we could be a visionary leader out in front, we could be a fast follower on the flip side or we can lay somewhere in the middle. And ultimately, where we landed is that playing that place in the middle is where we wanted to be and works really well for us. And we call it being a strategic planner. And it comes down to the use cases that we're going after, ensuring that they are tied to the business objectives that we have short term and long term, that they're also grounded in our community as we want to be in and putting our team at the center of them. And also that we know that they're going to have an impact, and that we're going to be able to measure and quantify that impact, as Cari talked about a little bit, too.

So kind of building on this a little bit. Azita, beyond large language models and -- at GenAI, what other GenAI? Like can you share a little bit about in your experience, what else are you seeing.

Yes. I mean I think I've mentioned it before, multimodal is really important, right? And I think what's incredible about GenAI is that it's progressing so rapidly, right? It started with large language models, text in, text out. And now you're able to actually interact with it by feeding it, let's say, a picture of your living room while you're chatting with a shopping adviser that's recommended a couch, and now give it a picture of your living room and ask, "Hey, how is this couch going to look in my living room," right, and being able to actually see that. And that's really when you're trying to make a purchase, so how does this dress look on me and upload a picture of yourself. Or when you're buying makeup, how do I apply this makeup and have a video that's explaining it to you. So those are kind of some really important multimodal capabilities. We have customers in the athletic business, in the luxury business that are actually designing their products in 3D in NVIDIA's Omniverse solution, and then taking those photorealistic rendering of their products and actually making that part of their generative AI. And being able to actually show your customers a full rendering of the product as it rotates with the right shading and so forth. And that's also incredibly valuable for your marketing and advertising teams, because these type of models can actually generate ask for you, right? Because they're trained on images, millions of images out there. We have a partnership with Getty Images and they've taken one of these models and trained it with millions of images on Getty and they provide that as a service that actually is fully licensed. So you have, within compliance, that you are basically using these Getty Images to create these beautiful marketing campaigns in less than 10 seconds. Something that takes literally photographers going to a specific area and actually taking pictures, and now being able to do that within a few seconds. And then just makes your creative team, your marketing and media team a lot more efficient and gives them more creative ideas.

And when I hear what you're talking about, it's data, right? Like there's so much data. And I think many people in this room, you probably consider the data management piece of this. Like let's talk about data management and how that fits into your models. Because I think we all know GenAI uses a lot of data. So Cari, can you talk a little bit about that and how that fits in all of this.