I'm Brian Gesuale, senior analyst covering industrial technology at Raymond James. Welcome to the 3:25 slot. Delighted to have Cerence here to take us through the story. This has been a stock that was absolutely on fire from $20 to $130. It's moderated substantially, and the business is really stabilized and appears to be turning. Delighted to have the company's Chief Financial Officer and IR here to take us through the story. We're gonna do this in a fireside chat format. If there are any questions from the audience, raise your hand, and we'll work them in as we move through the presentation. But Tom, Rich, thank you so much for being here. Thanks, Brian.
Thanks very much.
Why don't we just start off and level set the audience, if you will. Maybe take us through a very brief history of the company, your IP, why your customers choose Cerence, and maybe some installed base type of metrics out there as you look at the overall business to show your success in the industry.
Sure. Thank you, Brian, and thanks, everybody, for joining today. So Cerence is a spinoff from Nuance Communications. We spun off the auto business from Nuance in October of 2019. The auto business within Nuance had built a pretty strong foundation of being kind of the premium supplier to most of the OEMs around the world. Kinda starting off with kinda just command and control, the little button on the screen, to now selling really advanced Cerence Assistant technology. At the time of the spin, there was a separation of technical talent plus IP plus patents, so Cerence maintains a pretty strong patent portfolio around a number of technologies associated with the products that we sell, from natural language to...
We have the largest language portfolio in the auto industry. I think we support over 70 languages around the world. We have some voice biometrics technology, some unique auto technologies—I mean, audio technologies that we use inside the cabin. It's also allowed us to enter the two-wheeler market, which we can talk about later. So from that spin out, you know, the company's continued to invest and to grow on innovation across the auto industry. We announced last week kind of a slight pivot to our technical and product strategy, really embracing the advances in generative AI and large language models, which we have a long history in, going back with the company, but there's been a number of advancements in that.
What we believe is that we can take that core foundation, and we are the premium supplier to the auto industry around the infotainment, around, you know, speech and audio technologies. What the auto companies like about Cerence, versus some of the big tech companies, is that we have a large professional services organization which allows us to take our solutions, tailor them into a customizable solution for each OEM. And therefore, in simple terms, you can say, "Hey, Mercedes," "Hey, BMW," instead of, "Hey, Google." The other advantage is that the big tech companies are in because they want the data. They want to monetize that data. In our model, the OEMs own that data, and they're all working on programs that allow them to monetize that over time.
As we've rolled out this pivot in the technology, I think the OEMs are really excited about how we can take that core foundation and really enhance it, because we have more data. I think we have something like 60 billion tokens that we can apply to the technology and really create a very strong vertical automotive transportation solution that has the advantages over just a generic, ChatGPT or other open solutions, in that it would be tailored to the auto and the transportation. It'll have lower latency because it'll be condensed. It'll run at lower cost because we can shrink that footprint down. We're going to also have it such that it'll work embedded in the head unit and through the cloud, because a car moves around, and there's not always 100% connectivity associated with that.
But we can still and we feel like we can probably answer two-thirds of most queries that a driver will have in the automobile, and then we can still pass more generic questions to a ChatGPT or something else to answer in a seamless type of environment. That's our core auto business. We have also two opportunities. We won a number of deals in what we call the transportation adjacency market. Think two-wheelers and trucks. The two-wheeler market is about half the size of the auto market, so about 40 million units. We've won, I think, over nine deals in that market. We've really never lost a deal, and I think that's because we have the unique ability to combine our speech and assistant technologies with our audio technologies. And of course, a two-wheeler is in an open cockpit.
It doesn't always work. It can work through a helmet or anything, but you're still the person's outside. We've had 5 of those go into productions, but we're just on the beginning of the in-production curves around that. So we'll see how that market develops. Then the third area is what we call non-transportation. So it's the ability to take our technologies and apply them to industrial or consumer-type products. We are under a field of use restriction with Nuance/Microsoft, but that ends October first of this calendar year. There are certain areas that we can sell today and still stay within the framework of that. So we've been building out the capabilities for that over the last year or so. We've won 4 or 5 deals. We've got some market research done.
The beauty of those two segments, the transportation adjacency and the non-transportation, is it leverages our core technologies. We do a little bit of work to shrink the footprint because some of these applications are very small, but fundamentally, it's the same technology stack, and therefore, provides strong leverage against our core innovation and R&D investments across the company. Brian talked a little bit about the crazy volatility in the stock over the last few years. I joined—I actually led the spin for Nuance, joined the board, and then joined as the CFO just about six quarters ago. And the CEO became the CEO. He ran the business under Nuance and then was appointed the CEO when the previous CEO left.
Him and I have been on a path over the last few quarters to kinda really strengthen the revenue and the financial model, make it more predictable. We had what we called prepaid fixed contracts that had risen, not to go into a lot of detail, but to a pretty high level. We've brought those down. We're gonna bring them down again this year. We laid out our multi-year plan a week ago, Monday. I think we've built a plan for our fiscal year 2024, which we're on a September end fiscal year. That's very predictable.
We've had five quarters in a row of hitting the high end of our guidance, and then we revised our multi-year plan from last year that I think lays out a very strong proposition for our employees and our customers, and more importantly, our investors, that shows us getting in the medium term, I think fiscal 2026, 2027, of double-digit growth, 30% EBITDA, and extremely strong cash flows. The cash flows are generated because our connected services business is, is growing and will grow quite substantially over the next few quarters based on new technologies that are going into production at the end of last year and at the beginning of this year.
The way it works is when a car is produced, we get the value of the, the license, so the perpetual license that's in the embedded unit, plus if there's a connected solution on top of that embedded, that will usually have a 3-5-year service period. We collect all of that cash at the time of production, and then we amortize for the connected, that service revenue over the, the service period. So although we're not technically an ARR software company, we have a very strong recurring revenue base, 'cause when you get built on a platform, it's very difficult to take you off, and you just follow the production curves of the, of the auto industries.
I think the model that we laid out a week ago Monday, that we have pretty high confidence based on our backlog and deals that we've won, that we can deliver, shows, I think, a pretty strong investment opportunity for most of you.
I want to tug on a couple of the threads you mentioned there, but can you just maybe talk a little bit about market share and maybe the mix of business, particularly in the connected services versus licenses, before we dig into the model details?
Oh, sure. Yeah. Thanks, Brian. So today, the core auto business is really our, our biggest piece of the business and with opportunities in the other two areas. But in our core auto business, there's three revenue streams. There's the embedded license, and today we're on 54% on a trailing twelve-month basis of every car produced in the world, has some form of Cerence technology on it. On that 54%, 'cause we always sell embedded, and then we can sell connected services on top of that when you need to go to the cloud to get additional information to support, you know, a query or whatever. We're on about 21%-22% of that 54%. We add connected services to that.
So therefore, we're getting an upfront license on the 54 percent, and then we're getting a revenue stream over the service period for the connected. So 54 percent of all cars produced, then 21 percent on top of that. Our monthly active users that we track through our connected services, I think, was up 30% this year. Our billings were up 6% year over year, which shows kind of the strong growth in connected services.
Let's maybe go back in towards artificial intelligence and the large language models. Can you talk about the opportunities that that creates, and maybe help people understand a little bit more of the use cases? And as I understand, you've been putting a lot of work into this, and have some products on the horizon as well. So-
Yeah, I'm gonna give-
Maybe just help us, help us understand.
I'm gonna give Rich a little opportunity to speak here.
Sure
And give me time to catch my breath.
Yeah. Sure.
I'll pass that one to Rich.
So in that area, we have, like, a two-phased approach to it. First of all, we have products that we're enhancing with access to those large language models and generative AI. You know, it's an expensive proposition. There are several factors that go into play when you want to use those kind of generic large language models. Number one, they're expensive to use or access. Number two, there's latency issues, meaning because they're so big, a request can take longer than you would like to see in a car. Number three, they're not always necessarily accurate. They have, you know, hallucinations and so forth. So there's a number of different factors that can affect, you know, how well they actually work in a vertical such as the automotive space.
What we've decided to do is take several of our products today and enhance them with that capability. One of them, for example, is Car Knowledge, where we're taking that manual that sits inside the glove compartment of a car and turning it into a virtual assistant. Now, we're getting all the data from the car maker about what's going on in that car, our ability to read the sensors for getting information there. Like, that red light comes up on your dashboard, and you wonder what it means. We'll be able to interpret that and respond to the consumer's inquiry. But there's maybe 20%-30% of the questions someone may ask about a car that wouldn't be covered in that specific manual or that framework.
In that case, then we go out, and we access the generic, you know, large language model to be able to get that information, so that you, as a consumer, if you're asking a question, don't get the response from the system, "I don't understand what you're asking," which can be very frustrating. So we're taking several products and enhancing it with that type of capability. We've said on the conference call that we're currently in 15 proof of concept programs, meaning, you know, a car manufacturer never just takes anything on someone's word. We have to put it in a car and show it to them and prove it to them that it actually functions appropriately. So we're in the process of 15 of those now. We expect the first actual implementation in the first calendar quarter.
So that's phase one, which is taking our technology, some products, and enhancing it, and having it as an upsell opportunity in it, even for existing cars. And then concurrently, what we're doing is we're working on a brand-new platform or evolution of our existing platform that fully integrates, both on the embedded side and in the cloud side, the ability to supplement our large language models and generative AI, which we've developed, you know, 60 billion tokens through all our experience through the years, with access to some of these larger language models. But again, taking into account, latency, accuracy, cost, and so forth. So that's the approach that we're doing. We're actually. We'll be showing at Consumer Electronics Show. We'll be showing our new technology, the new platform, in terms of demonstrating it.
It'll probably be ready for deployment later in the calendar year. We'll also be doing a press event with a major car manufacturer just prior to the show that is adopting a lot of what we're talking about here today. The exciting thing is that a lot of the car manufacturers that we've had these discussions with have been very excited about our approach to this, so we're really excited about how this is gonna play out.
Sounds really exciting. How do we think about it? 'Cause when I think about your, your financial model, it's, it's a volume times price per car.
Mm-hmm.
Is there an appetite for significant price increases with the AI? Is it an additive price? How do you think about the bundle ultimately impacting that cost per vehicle?
Well, I mean, it all comes down to value. If you're providing more value, then you can get more price per car, right?
Yeah.
Our goal is to continue to innovate in a lot of different ways, right? It's not only with this approach from the AI, the large language models and generative AI, but it's also, you know, the audio capability in the car. For example, what we can do is set up the technology such that we can detect from each person sitting in the car as an individual bubble, if you will, of sound and be able to filter out the other three bubbles, maybe from the fourth person sitting in the car, or, you know, having someone in the back seat being able to access the system, and the system knowing that that someone sitting in the back seat wanting to have their temperature raised or wanting to get access to some kind of information.
So it goes beyond, you know, just this generic solution to adding more capability via the AI, but into all the other innovations that we bring to the market. And the idea is to continue to grow that, you know, price per unit over time.
Yeah. So there's kind of the core assistant technology, and then there's other solutions that can be bundled around that. We have a wake-up word technology, we have the audio technologies, we have a voice biometrics solution. And OEMs will go through that portfolio and decide how they wanna deploy that across their models.
Yep.
And so that'll drive higher PPUs, the more of that content. And then with the new tech, you know, with the advanced technology that we're developing, brings deeper solutions to that, that brings up the PPU. So you bring up the PPU on the license side, but then more and more, we're using the connected services to supplement that, that uses our cloud-based, LLM models, because you can fit more capability than you can squeeze into a chips et or on a head unit. So you get higher prices on the embedded license, which is a drop-down permanent license for that particular car, not for the whole models, but to that particular car. Then you add the connected services.
As I said, today, we're at about 21%-22% on top of the 54, and we see good growth opportunities for that, you know, to go to 25 or even higher. And then you increase the PPU on that, then you build up this really strong deferred revenue and cash flow model, and that's what we laid out.
Mm
... a week ago in the plan. Right?
That's right. Let's actually peel into that guidance a little bit. I've had a lot of questions kind of coming out over the last week. You brought down your out year revenue in fiscal 2024 a bit. And then you kind of laid out a multi-year, I wouldn't call it guidance, as I don't think you'd call it-
Yeah.
Guide rails or whatever we wanna think about. Can you maybe bridge the delta from where you had been maybe a year ago to where we sit today?
There are a number of pieces in FY 2024. We've had a long-standing, what we call legacy contract, that goes back to when auto was part of Nuance. So there was a company that did a very large deal with Toyota in 2011. It was the early days of connected services. It had a lot of content in it, and it had an extremely high price per unit. And Toyota shipped that solution on a number of cars between 2011 and they really started winding that down 2018, 2019, to the point where by the time we did the spin, they were only shipping that on a few cars, and I think they stopped probably 2020, somewhere around there.
The cash was collected when auto was part of Nuance, but at the time of the spin, there was a very large deferred revenue balance that sat on the balance sheet that came with the spin. We've been amortizing that over... That had a, what I call a goalpost, where that service was gonna end 12/31/2025. And so a very predictable roll-off of that amortization. At the time of the spin, I think the revenues were over $60 million a year-
Yep
... associated with that deferred. It went down to $34 million in 2023. It was supposed to be $34 million this year and $34 million next year, and then there was one quarter in our fiscal 2026 of about $8 million. Toyota made a decision this quarter that they were gonna end that service with their customers. This was a decision they made by 12/31. Therefore, the accounting rules say you take all that deferred revenue, and you accelerate it because you have no more obligation. So we will have $76 million of deferred revenue in Q1. If you deduct it from what we thought was gonna be $34 million, it's +$43 million.
Then we had some consumption delay between 2023 and 2024, so that would have reduced our numbers from what we said last November by about $10 million, and then we made the decision to reduce the level of our prepaid contracts. We've. I've been trying to take those down since I became the CFO. They're on models that are already shipping, and it's only done with some Tier 1s in Asia, and they will ask to buy an inventory of licenses. So you take the revenue, but you have to give a discount. I'm trying to get us off these discounts, right? The good news is they pay, but with the cash we're gonna generate from the connected services, it's not that critical for us anymore.
Right.
So I took it down to 40. In 2023, we actually did less than that. We did $36.5 million, but it did have a $20 million from what we thought we were gonna do a year ago. And you do all of that, and the guidance that we put out last Monday was slightly less than if you do all the math, what we said we were going to do a year ago in November. Now, in the out years, we took the numbers down quite considerably. A piece of it was that deferred revenue for 2025 and 2026. The $20 million is every year-
Right.
going out. I lowered the numbers slightly for the two segments that I talked about, just 'cause I was trying to be a little more conservative, and I was looking at the pipeline that they had, and I kind of refreshed the numbers. And then there were two other areas. One, as we pivoted to really focusing on the new platform associated with our core technology, particularly giving all the feedback we were getting from the OEMs around, "You guys got to-- This is really cool. We really want you to focus on that." There were a couple of peripheral-type solutions that we had thought we were gonna do last year that we've defocused. Things like Tour Guide, and there were a couple others. So I took those revenues out.
The last piece was, I introduced this five-year backlog metric last year in November, 'cause I think it's a very good metric for the company and for our investors. We had always reported bookings. Quite honestly, it's a terrible number, 'cause there's twelve-year deals in there, ten-year deals. You know, you have no idea how the revenue is going to play out. With a five-year backlog number, and we give it how it rolls out by year, is a much better indicator of the business that we're winning and how it's going to translate into revenue, at least over the medium... short and medium term.
As we went through that analysis and prepared, 'cause my goal was to be able to report that twice a year, we just got much better insight into how all of our deals that we've won, that are going into production, how they play out. And honestly, the numbers that we presented last year are a little too high.
Yeah.
So I adjusted it for all this new data and analysis that I had, which led us to being able to provide last Monday a couple of charts on the visibility we have to our two biggest pieces of revenue, which is the core license and the connected. So I think it's just I think we have a stronger, more predictable, more confident set of numbers that we released last Monday. And of course, it's updated for our new plans and strategy-
Yeah.
- going into FY 2024.
Brian, I just want to add one thing-
Yeah, please.
... 'cause there's been a little bit of confusion on this. The Toyota contract, right, was a, as we refer to, a legacy contract for a program that was bought a long time ago, as Tom explained. We still do business with Toyota in a lot of other areas.
That was my next question.
Yes.
I'm glad you brought that up.
We do that. We continue to do business with them in a lot of different programs, including their newest ones. That was just a one-time business that was bought by Nuance back then, and it was on a number of cars. That hasn't been on new cars for several years now, but all their new, you know, a lot of the-
Yeah, just think of that as one program that had a-
Yeah
... a curve that ended a long time ago. In the interim, as Rich said, we've sold Toyota multiple new programs that are shipping on cars today and will ship.
So still in, still connected and still licensed?
Yes.
Still doing both of those.
Yes.
Yes.
Absolutely.
Okay, fantastic. I wanna dig in now to the 2024 number a little bit. As I kind of looked at it and we ran through our numbers, there's a significant acceleration that happens fiscal Q2 through Q4, particularly on the variable license side.
Yep.
I know there's several factors that support that. Would you take me and everyone else through that, please?
Yeah, sure. A lot of it is associated with the new programs that have gone in, in the back half of 2023 and a couple that are coming in, into play in Q1 of 2024, that carry those higher price per units, both on the license side and also have a significant connected component to them. And also, starting, we've been building out the deferred revenues on connected, 'cause it has started to grow in the back half of 2023. So on the license side, it's more new platforms shipping with the OEMs, with the higher price per units, which we had talked about last November, but some of them got delayed.
Right.
It didn't show so much up in 2023, but it's starting to show up in 2024. We have pretty good insight to that because we know they've gone into production. We model how they ramp. We've got a lot of history on how, you know, platform programs ramp. But it's really those two factors, Brian, that come into play.
Fantastic. I wanna end with two, two more. One, maybe just... You've kind of touched on it periodically with some of your answers, but maybe give people a flavor for the long-term financial model, revenue, gross margins, EBITDA profits. And then, I'm gonna give you the last word here after you do that.
Sure. So quickly, you know, our license business, which is, which is growing, has a very high gross margin, high 90s%. Our connected services has been in the 70s%. This Toyota contract has very high margins, very small cost. So you'll see in 2023, very high connected margins. They'll come down a bit in 2025, and then they'll get back to kind of the 70s% in the out years. Our professional services is about mid-20s%. Our two segments are licensed business mostly, so they'll, they will have high margins. Our R&D, we continue to invest, we move around resources, we reduce some spending on legacy, so not a lot of incremental spending. And then we're a B2B model, so sales, marketing, and G&A, we get a lot of leverage, too.
So that our 2026 and 2027 model has us going to double-digit growth, and by then, a lot of the noise is out of the system-
Right
so it's pretty clean, and high growth margins and 30% EBITDA with extremely positive cash flow. So we've been in a EBITDA-to-cash flow relationship well below 100 the last couple of years, mainly driven by the legacy. When you look out, we're gonna be well above 100 due to the growth in the connected services.
Fantastic. Maybe in 30 seconds, the drop-the-mic moment, tell these folks out here why they should own Cerence.
I think all you have to do is look at the plan that we laid out last Monday. I'd ask you to not focus so much on what we took down from last year, but look at the plan that we laid out, look at the recurring revenue, look at the path to this 10% growth, 30% EBITDA, really strong cash flows, and say, "Gee, this looks like a company I'd really like to invest in.
The returns speak for themselves. Thanks so much, everyone.