Kinaxis Inc. (KXS) Earnings Call Transcript

Welcome to the National Bank of Canada Virtual Event. Doug Taylor, you may begin.

Thanks. Hi, everyone. I appreciate you joining us on Friday morning at the end of what has been an exceptionally noisy week in the financial market, I think you'll agree. As introduced, I'm Doug Taylor with National Bank Capital Markets Equity Research. I'm happy to have the opportunity here to host a fireside chat with the management team of Kinaxis, which has been one of our top ideas in Canadian technology. Here with us today, we've got Razat Gaurav, CEO of Kinaxis since earlier this year. And we've also got Herb Yeh, who is CFO since I believe it's this summer broadly. So relatively new faces for many of you here at the head of Kinaxis. I'll also point out that this comes on the back of the company's stellar Q3 results, which were released overnight on Wednesday, produced a positive reaction yesterday on a day when there was not a lot of green to be found on most of our screens. Now the goal here today is not to regurgitate the quarterly disclosure and that process. But perhaps those who didn't get a chance to go through the results, it was a busy day yesterday, and some of you are otherwise new to the story. So why don't I get the management team to sort of give us the highlights as you see them, just to orient everyone, and then we'll proceed with some bigger picture questions that I get from investors on the Kinaxis story. So Kinaxis, Gaurav or Razat, over to you. Just maybe provide us a quick recap and get us started here.

Yes. We had a very strong Q3 performance. The team did a great job. We posted 22% SaaS revenue growth. Our ARR grew by 19%, 20% on a constant currency basis. And on just about every metric, we were really pleased with our performance. We had a good balance between net new logos as well as existing expansions with existing customers. And as we look forward, given the performance we've had for 3 consecutive quarters, we felt really comfortable to further increase our guidance for the entire year. So we've increased our guidance in terms of total revenue, SaaS revenue and EBITDA margin for the year. And again, really tremendously pleased with the performance so far.

Okay. That's great. So we'll get into the Q&A portion, which is where we're going to spend the majority of our time here today. I'm very happy to lead the discussion. But for the audience, you can input questions into the portal, you're watching this through and it will hit my inbox here and I'll try to work those questions into the discussion. So please make use of that feature to try and make this event a little bit more interactive. So with that, let's get started. I think before we drill down into some of the nitty gritty, I want to step back a bit and talk about the market that you're servicing, supply chain as a function for many of the world's largest companies. Obviously, there's a ton going on in the world that's stressing these functions. So I wanted to start by talking about the why for Kinaxis because I think it's one of the key arguments for a story like this amongst the broader software universe that we've all choosing to invest within the budgets for these types of solutions, the appetite for change and adoption. So you've been in supply chain for some time, Razat. How would you characterize where we are in the cycle for supply chain automation and modernization?

Yes. I've spent most of my career in building supply chain software. So you're totally right. I've never seen a time like this, both in terms of opportunities for innovating given the enabling technologies, the compute infrastructure, all the new data architectures and advance of agentic AI. But also, I've never seen a time like this in terms of the appetite on the customer side for them to transform and truly adopt these new capabilities. So let me first start with anchoring everyone on like what do we do at Kinaxis? What does the Kinaxis solution do, where we're positioned, and then I'll talk about what I see as the evolution of that, right? So at Kinaxis, we are the leaders in supply chain planning and decisioning. We play in 7 verticals, and we are working with leading companies in every one of those verticals. Customers include the likes of Unilever, Qualcomm, Merck, Colgate, Ford Motor Company and several ExxonMobil, just all the leaders in every one of those verticals that we play in. And we help these customers in making decisions around their demand predictions, so how to forecast demand, how to plan their supplies, how to plan their production, how to plan for their inventory levels across their supply chain network. And the way we do that is by creating a digital representation, a digital model of their physical supply chain that incorporates all the nuances and all the gory details that encompass the physics of the supply chain. So what constraints exist on a shop floor, what constraints and configurations exist on a dock door in a distribution center. how to think about the interdependencies across that supply chain network. If you make a change to demand downstream, how does that impact your production plans upstream. All of those representations are contained within our semantic representation within the Kinaxis platform. And then with that representation, we then have curated decisioning engines for coming up with the best answers for these different use cases. So obviously, on the forecasting and demand planning side, we use a heavy dose of machine learning techniques. On the supply planning, production planning, inventory planning side, we use a heavy dose of very complicated and sophisticated optimization algorithms and heuristics algorithms. And of course, in the last few years, we've been leaning in on using the reasoning capabilities enabled by LLMs or generative AI. And more and more, we've been leaning in on developing agentic capabilities as well. So we've been heavy users of predictive, prescriptive and then more recently, generative and agentic AI. So that's an overarching picture. What we have done is we've created this platform, a single code base, a single platform. So it's not something that has just grown through a mishmash of acquisitions. We have a lot of proprietary technologies for a company of our size, we have a very rich patent portfolio and some deep tech surrounding this for how we represent concurrency in the supply chain, how we enable an in-memory architecture that is very foundational to what our customers do with us, how we enable scenario planning, which is super important, especially when things are changing so much in the supply chain and with all the volatility and the variability that has existed in the supply chain and that volatility keeps growing, by the way. And with all those technologies, we bring it to life with our teams, our field teams, our partner ecosystem. And that's what really sort of forms the shape of both our innovation cycles, but also our go-to-market and customer delivery cycles. So hopefully, that gives you an overarching view of where we are. The key trust we have is we are continuing to stay focused on our core of supply chain planning and decisioning. We feel there's plenty of room for us to continue growing in the 7 verticals we operate in. And then the second layer to that growth really is in expanding beyond just planning into how do you action those plans, how do you operationalize those plans. And that's where we've announced earlier this year our vision and our game plan for operational orchestration that is leaning in and harnessing the power of new semantic and oncology architectures, agentic AI and a composable platform, and we are bringing that to life with forward deployed engineers. And the reception and the interest level we are seeing from our customer base is very encouraging. And while we are early in that journey, we are definitely seeing a lot of need and demand and interest from our customer base. So that's where we're at. I'll stop there. And Herb, I don't know if you want to add to that.

No, I think it's well said.

Yes. Don't worry, Herb. I've got specific questions for you coming. Yes. I mean -- so you mentioned a lot about the product and the tech there. And I want to get to all of that in a little bit. But I'm just going to stick on the theme here of the demand and the appetite just for a little bit longer here. I think those of us who have covered Kinaxis for a while, longer than the 2 of you that have been at the helm here, we've seen in recent cycles where there's been major impacts on supply chain, some degree of disruption. I mean, whether it was tariffs or fuel costs, which are current, or things like COVID, that was certainly an example. You saw a bit of this period where the customers focused on putting out the fires and there was a little bit of a lag and then a wave of adoption of new tech in the wake of that. I think I'm characterizing that accurately. And the question to you here is, given everything that's going on and the timing of all this, where do you feel you are if you picture Kinaxis adoption in waves right now?

Look, I think we are still early in that journey, right? We have less than 450 customers. But most importantly, let's talk a little bit about what's happening in our customers' operating environments, right? And some of it is situational to industries. But in general, both the demand and the supply side is becoming more variable, right? So I'll give you some examples of that. Obviously, when it comes to the high-tech value chain, there's been a surge in the whole build-out of the data center, which is creating significant pressures and constraints in the entire value chain, right? Constraints shift all the way from wafer capacity on the semiconductor chip side to energy and utility companies being able to power them, and we're working with that entire value chain. On the other hand, you look at industries like pharmaceutical and life sciences, where the velocity and frequency with which new products are getting introduced are growing very rapidly. And that's causing for new supply chain models to come into play, right? For example, one of our customers has one of the GLP, Mounjaro offerings where perishable fluids, vials and syringes had to come together in a supply chain, which is seeing unprecedented demand, right? And we had to build a new optimization solve for how to bring these things together for this pharmaceutical company. That same company, by the way, also has a tablet version of that, that is going through FDA approval, right? Now all these shifts and changes create a bigger need for the Kinaxis platform. In the CPG industry, in FMCG, for example, we have some large global players where there's a generational shift happening in consumer behaviors. It's sort of like some -- a certain generation, like my mother's generation does grocery shopping once a week, once in 2 weeks. And my nephew, they shop for the next meal. And the supply chains required to service those different generations look very different from each other, right? And there are other examples that are caused with changes because of geopolitical shifts, the wars that are happening around the world with the different tariff regimes, the regulatory changes. So as the world is getting more and more variable, right? And as Ray Kurz will taught us, the pace of change is not linear. It's only getting exponential. The key thing supply chain professionals are trying to grapple with is -- I mean, change is an input. They don't control that change. But how do they deal with all this change is the key question. And how do they minimize the latency between picking up these change signals and making smart decisions and then from the latency between making those smart decisions and actioning them, right? That's the space that we are playing in. And we see the sense of urgency from our customers and the discussions I have with Chief Supply Chain Officers, with the digital transformation teams of our customers where we are continuing to see significant prioritization for solving for these problems because they're at the very core of what these companies do -- not do that, we have to make sure we have the best technology, the best platform, but we also have the smartest people to solve these really complex problems. And from an adoption perspective, we are still very early in that journey.

Yes. Let me dig in a little bit on that. I mean, yesterday, you talked about the new customer momentum was very strong, the 40 go-lives in the pipeline. I do get questions frequently in trying to understand the significant amount of white space that's still within your existing customer base as you open up new growth vectors and work through your product development initiatives. You talk about an NRR of over 100%, pretty big figure, if I could poke at you a little bit on that. I just want to generally just visit and ask about that white space within your customer base. And if you could maybe provide us with a sense of the amount of room left to go. I don't know if you've got any numbers or anything that you could share or range as you think about the penetration rate of your solution even within your existing customer base? Because when I go to your user conferences and other supply chain events, I'm often struck by how fragmented the deployments are within some of these organizations.

Yes, it's a good question. So I'm going to start with your last point first and then sort of talk a little bit more about the white space. Supply chain management is one sort of term, but it's made up of a lot of different subprocesses and a lot of different underlying systems that underpin that. There is no one single system of record for the entire supply chain. We are one of those systems of record for planning, but there's a whole plethora of other systems like transportation management systems, warehouse management systems, order management systems, sourcing systems, manufacturing execution systems that are also systems of record for various other subprocesses. So just keep that in mind. I mean it sounds very simplistic to say supply chain management, but there's a lot of different sub processes and systems that underpin that. When it comes to our customer base, your question is a good one, Doug, because we've been doing some really granular analysis on the white space available within our customer base. And the conclusion we've come to is the opportunity is significant for us to expand within the existing customer base. And that's obviously coming to bear as you've seen some of our results in Q2 and Q3, where expansions within existing customers have been a growing share of our overall net new ARR, right? And there are different flavors in terms of what those expansion opportunities are. In many cases, we may be working and we may just have license with one business unit. And now we can sort of expand to additional business units. So we may have been working with a subset of the product lines or we may be working with the aftermarket service repair parts of the organization. Now we are working with the finished goods part of the organization. So all kinds of expansion in terms of business units. The same kind of expansion may exist regionally, like we may start in North America or in part of Western Europe, and then we may be rolling out to other global parts of their organization. Sometimes those create significant expansion opportunities for us. And then third flavor for the white space is as we've been adding additional modules to our R&D and innovation cycles, we just have a lot more to take back to our existing customers and cross-sell to them. Examples of those include we've all within the same platform, but we've recently added a whole production scheduling capability that is net new, machine learning-based forecasting capability. We're in early adoption with our advanced inventory optimization capability that we're going to scale out as we go into 2027, all of our agent capabilities. These are all cross-selling opportunities within the existing customer base above and beyond what they may have deployed with us. And then the last thing we've done here is we've created a dedicated customer -- current customer-focused farming sales team, right? And that team is becoming super productive because there's all these significant expansion opportunities within our existing customer base. So we tend to agree with you that there's -- that's a significant growth opportunity for us. And then as we lean in and build out this operational orchestration footprint to operationalize the plans, again, that will even further expand the TAM within the existing customer base in a very profound way.

So if I could -- yes, I mean, lots to take into consideration there, but summarizing, significant TAM within your existing customer base that is yet untapped. You've done work on it. You're perhaps not going to give us any numbers or anything around it, but early in the evolution of what Kinaxis could be within any single customer. Is that fair?

Yes. What I'd like to tell my team is we can be double the size we are within our existing customer base, right? That's the size and scale of the opportunity.

And that customer base continues to grow every quarter.

That's right.

So all right. Let's get into the AI type questions. you can't avoid them these days, not that you want to. And I want to start, I guess, another question I find myself constantly fielding, I mean, really with every software company, but in Kinaxis specific, even as the term SaaSpocalypse fades from everyone's mind here a little bit in the last couple of months. A big picture question for the generalists out there, they're seeing all these unbelievable advances in these models and their abilities to quickly optimize outcomes. And they'll sit there and Kinaxis is not a tool that they use day-to-day and many of -- most of us don't. But it is a tool for optimizing outcomes as well. And so I constantly find myself yielding the question, even with all the growth and success you're continuing to show why a sophisticated customer can't jam the same data into some horizontal frontier model and a supercomputer and have it solve those. And I know that's a very crude way of asking the question. But honestly, as I say, I continue to get it. So I just want to attack that head on with you so that you can make it clearer for everyone why Kinaxis is necessary in the future -- agentic future.

Sure. Well, firstly, we love AI, right? And we love every advancement in AI, and we want to be at the forefront of applying the latest in AI to our domain of supply chain planning, decisioning and orchestration. So I'll start with that. The second thing is when it comes to -- I think what you're talking about are LLMs, so Claude, OpenAI, I don't think you're referring to optimization models, machine learning models, right?

That's what the investors see, right?

Yes. And look, we love harnessing the power, the reasoning power of LLMs. And these LLMs, which are built on these multi-tiered neural networks, they are creating new opportunities for us to provide a conversational interface to our platform. They are allowing us to expand and democratize the usage of our platform across a much broader set of stakeholders. And agentic AI is creating all kinds of new possibilities for us to create new automations that drive a lot of productivity improvements, but also to help reimagine the flow of work, the ways of working and decisioning for our customer base. So all that is really good. Now what these LLMs don't do is a few things, which will continue to be within the purview of what we have to invest in incremental to applying these LLM models and agentic models, right? The first part is what I talked about is representing the physics of the supply chain. That context is very important, right, for the LLMs or the agents to really be able to do anything smart, right? And building out that context and that semantic model of the physics of the supply chain network is something that we've spent decades honing in on, and it's complicated. And I want to make sure you understand why it's complicated, right? Supply chains are not in the ether. They're in the physical world, right? It's on shop floors. It's in warehouses, distribution centers, dog doors, pallets, containers. And there are constraints. There's constraints around material availability, parts, capacity of production sites. There are ships that need to operate that you need to understand. So representing those constraints at the minutia and granular level of detail is very critical before an LLM or an agent is able to really make sensible decisions. The other thing that's really important is the supply chain network, part of the physics is that you have to understand the interdependencies in that network, right? If you have a shortage in incoming parts into a manufacturing plant, what is the impact of that to a customer order that, that particular finished product is spent to? Or if you make a change to downstream demand, how does -- how should that change your upstream production plans? Those kinds of bidirectional network-wide interdependent kind of understanding, what we think of as the concurrency representation of that supply chain network is very critical for these agents to understand. And I've used those examples with some of the investors just to give you the scale of the complexity in the domain we play in. When Merck makes a change to the expiry date of a single ingredient, it results in over 18,000 recalculations in Maestro, right? That's how connected these things are. And that's also why we -- across our customer base, we're doing over 300 billion to 400 billion calculations a month, right? So this is not a simplistic domain. I remember a former executive, CEO telling me Raad, a SaaS CEO telling me in the Silicon Valley result, why are you in the supply chain domain? This is really hard stuff, right? Like all these hard problems. Why do you keep spending all their time and solving these hard problems. We are suckle for that at Kinaxis. We love solving hard problems. And thank God, we're solving hard complex problems because they don't just go away with wide coding or with using LLMs. I think LLMs have a very important role to play in that stack, right? And we are leaning in and embracing the use of LLMs and agents in our platform, but they don't replace the need for this model representation of the complexity and the minutia details of the supply chain. So that's the first part. The second part is with that representation, we are making decisions around what to manufacture in what quantities, in what locations, in what time frames for customers that may have anywhere from 5 to 100 different manufacturing sites. And they may be manufacturing anywhere from 2 or 3 products to tens of thousands of products, right? And so when you look at the explosion of the permutations and combinations of what an optimal plan looks like, it explodes into hundreds of millions, sometimes hundreds of billions of permutations and combinations. So you need advanced mathematical models where we are using linear programming and miss of programming techniques with solvers that providers like [ Gurobi ] and NVIDIA are providing or with machine learning constructs or models that Google DeepMind is providing to us to solve those in a very mathematical way because these decisions our customers are making, they are to the tune of -- they impact tens of millions to hundreds of billions of dollars of inventory decisions and product decisions. So they have to get it right, and they have to be very deterministic, right? And by the way, those calculations don't get done by LLMs that are designed for predicting and looking at patterns based on text and image data, right? That's why a lot of the things we're doing we are leveraging the models from Google DeepMind, Gurobi, NVIDIA and the open source world in addition to layering on top of that LLMs from a reasoning, from an inferencing and from simplification and explainability of what these models are doing. So it's an important part of the stack, but it doesn't replace the need for these advanced mathematical models and these decisioning engines, right? And if you don't believe us, go and talk to Google DeepMind and go and talk to NVIDIA because they're investing in these solver models, and we are applying those to our domain, right? So that's -- those are the things to keep in mind. So we love the advancements in LLMs. We are big users of these LLMs even within our own operations, and we've seen significant productivity in our coding velocity, in our engineering velocity and all kinds of internal functions. We're embedding that into our platform, and it's part of our agent harness as we're building out our composable architecture. But they don't replace the need for the supply chain context and the model representation with all the complexity and they don't replace these advanced mathematical models that need to underpin the decisioning engines for our customers for the use cases we service. So that's what's important to keep in mind. I know it's all very confusing for a lot of people because it's a technical discussion. But the way to think about it is that is in terms of a stack.

Yes. Well, I just -- I mean, maybe I want to simplify that a little bit just to make it maybe I'll talk about -- so there's the data that is customer proprietary to your customers, and that is -- I mean, not yours. And then you've got all this advancement on top, LLM, the compute power, the calculation models, all that, that's being built by others. And your position sits in between. And that is -- and I get --

Yes, can I say something? We have one of the systems of record as well. So for our customers, when they need to understand what is the forecast for a given product or SKU or a unit, that gets computed and derived from our platform. Or if they want to know what is the supply plan or what is the inventory policy in their network for different segments of products. The system of record is ours. That gets derived and computed through our platform. So we are at the system of record level as well. So just keep that in mind.

Yes. I just wanted to illustrate that. And then your 20, 30 years, whatever, of developing the sophisticated physics of the supply chain models is very proprietary to you and not replicated easily even if someone was trying to with some new technology that's coming out or coding it, as you say.

Yes. In fact, what we are seeing is in the world of agentic capabilities. A lot of those proprietary capabilities we have like our version database, like scenario planning, it's becoming more important, right? And so as we think about sort of the next 12, 18, 24 months, customers will be building agents on our platform, and we have an agent harness to help enable that and an agent studio to help enable that. I fully expect that there will be third-party agents that would be part of a different stack but that need to extract data or that need to ping our scenario planning engine or doing iterative replanning on our platform, right? And so -- and that's why that was one of the big reasons why we've evolved our pricing structure to the activity unit. So it gives us a fair and transparent way of monetizing those use cases and still being very relevant in that go-forward future where there will be multiple agents from multiple providers, all doing different bits and pieces of the way that an organization functions.