Cerence Inc. (CRNC) Earnings Call Transcript & Summary

Okay, great. Thank you, everybody, for joining us. My name is Mark Delaney, and I cover Cerence for Goldman Sachs. As many of you know, Cerence is a leading provider of AI-powered voice recognition and speech software technology, and many years ago, was the automotive business at Nuance. I'm very pleased to have with us today, Stefan Ortmanns, the CEO and Director; and Richard Yerganian, VP of Investor Relations. Thanks for coming.

Thanks, Mark.

Thank you, Mark.

I thought, Stefan, to start, you could maybe give a general overview of Cerence's product offering with voice technology in auto, and what does that mean in terms of content per car and attach rates.

Yes. So first of all, Mark, thank you for having us here. We have a very compelling comprehensive product offering, product solutions, technologies and services. We have also transformed ourselves from a component supplier to a more innovative partner for all OEMs across the globe. Our core product is related to conversational AI. We have launched a couple of quarters ago, the so-called Cerence Assistant. And Cerence Assistant is a conversational AI solution, white label. So, we are allowing the OEMs to create their own branding to own the business logic and finally also to own the data. The embedded part is very important so that we have really a very fast response time behavior, but also cloud. And overall, I would describe our solution as a multisensory multi-model and multi-assistant solution. Multi-assistant is very important also to us because in the future, but we will have various different assistance in the car available but through our solutions owned by the OEM. And we have to do this not just for North America or for Europe or for the western world, but also for China, where we have different players like Baidu, Alibaba and Tencent. Multi-sensory is also important to us because when we have direct access to all the sensors in the car. And we can also then provide to the consumers to the drivers, co-drivers and passengers a kind of proactive AI. That's very important. So this is our core product, and we can also leverage the core product for other adjacent markets like 2-wheelers and trucks. 2-wheelers are also a very interesting opportunity for us. It's a bit lower in terms of volumes compared to passenger vehicles. It's in the range of 40 million to 50 million. But nevertheless, it has also a different complexity. Just think about, it's not a cabin, it's open, right, and we are providing also this with our audio AI capabilities. So overall, I would say, across the globe, we are a global player. We have all of the different technologies under lease like audio AI, text to speech, for example, for creating custom voices, all elements for conversational AI from speech recognition up to natural language processing, even large language models. We're doing this for plenty of years now. And then we are focusing also on new opportunities like emotional AI, communication, exterior communication with passengers inside the car. I'll stop here.

Yes. Lots of different products and capabilities that Cerence has, when we think about what that means in terms of price per vehicle or price per unit, PPU as Cerence refers to, the company at the last Investor Day had guided to about 380 PPU for some of your key programs in fiscal '23 growing to about 480 in fiscal '24. Maybe talk about how you're tracking toward that and perhaps where it'll progress longer term when you think about all these different innovations?

So when talking about price per unit, the 380 are only related to embedded products. At the Analyst Day in November of last year, we said, okay, there's a growth of roughly 26%, 27% in general, right? We are seeing that we are moving towards this increased PPU. Why? Because now there are actually 3 flavors. One, of course, is new programs, and we have a strong win rate here, right? Also, some nice win back here. And just looking at the Cerence Assistant, right, we are offering much more features than in the past. Secondly, some of the OEMs are now focusing also on so-called fresh apps. Meaning that we are doing, on a regular basis, some upgrades of their current software stack on the same underlying platform and then, of course, also some OEMs leveraging now OTA services, over-the-air updates.

In terms of your attach rates, I think last quarter, you said trailing 12 months. Cerence Technology was on 54% of vehicles. Where do you see that going over time?

So, overall, I think last quarter, we mentioned during the earnings call an attach rate or penetration rate of roughly, exactly 54%, which is really great. We see also a nice growth in hybrid units. Meaning hybrid, they have actually connectivity module in the car. I think year-over-year, the growth was roughly 50%, which is really good for us. We see also an increase of monthly active users in the range of year-over-year of 30%. So, also here, the growth driver for us in the future is more or less connected services.

In your last Investor Day, you talked about some new innovations Cerence was working on, things like wellness sensing, exterior vehicle communication, face biometrics. Can you dig into some of these innovations in a bit more detail and what they might mean for your revenue and profitability?

Yes. Also here, I think we are doing pretty well with our innovation roadmap, right? So exterior communication is that driver, a co-driver can talk to passengers outside the car, right, with closed windows right? This is all about also safety, very important when looking at India, the markets there. We see also a trend towards more emotional AI, which goes actually a bit towards wellness sensing here. We have these technologies available. Then, of course, when it comes to face biometrics, also exterior communication, we have some dependencies on the hardware, right, because we are just a software supplier, right? We're not producing cameras, for example, when thinking about driver monitoring systems, but we're leveraging also the hardware in the car with our new technologies.

So, Mark, a lot of these things that Stefan was talking about, what that does is these are add-on capabilities, right? So, things like car knowledge, which is taking the car manual that sits in your car, turning that into a virtual assistant that allows you to get any information about the car simply by asking it, or emergency vehicle detection, all these things just are add-ons that help us continue to build up that price per unit. So it's incumbent upon us to continue to be innovative, right, in terms of developing these technologies to help with that.

And where do you stand on developing some of these other related technologies? Things like face biometric wellness sensing doesn't necessarily seem immediately tied to voice. So do you need to do more R&D or M&A to be...

So we are doing a lot of R&D in the specific areas, right? And our approach from the beginning is also having a tight relationship with some of the key OEMs. We have defined some innovation OEMs across the globe. So in North America, in Europe, Japan, Korea and also in China, right? And then our road map is fully aligned also with their thinking, with their road map, right? And they're doing actually a couple of prototyping together with those big OEMs.

You spoke a little bit or alluded to this a little bit is your win rates and some of the win backs the company has been able to announce from big tech over the last couple of years. Maybe talk a little bit more on your market share, what's the competitive landscape look like, and are you seeing more traction with the competitive landscape against even some of these big tech giants?

Yes. So actually, automotive, the digital car is actually now a hot market to be in. It's a great market to be in. If you deliver what you have promised to, right, then you can easily secure all the recurring business. Over the last 2 years, we have significantly improved our core performance, embedded and also cloud. As also Rich said, right, we have added more and more new features, new functionalities. I think our solution is really competitive also when it comes to benchmarks. You have to understand that all of the OEMs are doing benchmarks before making a final decision here and we're doing pretty well. We won back some big business here, for example, from Google. When looking at the competitive landscape here, you have a couple of players here, of course, in North America. So Amazon, Google plus some niche players like SoundHound. When looking to China, it's quite fragmented here. You have the big players like Baidu, Alibaba. But they are not a big player in voice control for cars here. And then you have also a couple of players like -- niche players like AI speech, Unisound, or iFlytek, right? But overall, we are doing pretty well here. For example, we are the preferred partner of the #1 in the EV market, namely BYD.

To the extent you're on a vehicle, along with big tech, right, it's not necessarily an either/or, this can be multiple solutions can bring more functionality for the end user. What does that mean for your pricing? Does it change at all? Is it lower or do you still get the same content?

Yes. It's pretty much the same. There's not a tremendous amount of difference from one onto the other. The reality of it is, like we said from the very beginning, it's a co-existence story, right, where we're not trying to put this position out there that a car should only have our technology and no other access to any of the virtual assistants because those are part of a person's digital life just like many of them are part of their digital life. So our goal is to provide our customers or the OEMs with the ability to have that full digital life implemented inside the car. And that means working with and alongside some of these other virtual assistants, but being the primary one because that's in the OEM's best interest is to have that branded solution to them so that their customers identify the car to the OEM and not some third party.

Got it. One of the things the company has spoken about has been trying to limit the number of upfront sales in given your kind of volume sales, so to speak, to auto OEMs, and that perhaps could help with pricing. Double-click a little bit on that and maybe just more broadly too speak around what you're seeing in terms of pricing on a like-for-like basis. Is there any pricing pressure in parts of the business?

Yes. So you're referring to fixed contracts there, which are related to the license business only. And that business -- those fixed contracts are primarily or mostly with Tier I suppliers who are the outsourced manufacturers of infotainment systems for various carmakers. And it's primarily ones that are based in Asia, Japan and Korea. And historically, that's been a modus operandi for how they've done their business where their procurement teams go through this negotiation for getting an additional discount by buying an inventory of licenses in advance of them actually using it. So, there is a discount associated with that. That's not the primary reason that the company wanted to limit the ceiling of those contracts on an annual basis. The reason the company wanted to limit contracts on an annual basis was to be able to get to a point where the addition of new fixed contracts was normalized to the consumption during the fiscal year so that the license business would be more -- you'd be able to model it, if you will, based off of what auto production is expected to do price per unit, some of the variables that you would typically want to model a business by. And what had happened was because the fixed contracts became at elevated levels for a couple of fiscal years, it created a very difficult situation to be able to model that license business. So, this year, this fiscal year, we're going to probably do under $40 million of fixed licenses. And like I said, that's the ceiling that we have moving forward. That should normalize as we get out to fiscal '25 in terms of getting to the point where those new fixed licenses essentially are normalized to the consumption.

And just to make sure I understand what you're guiding for or anticipating at this point in time, so in 2025, do you think sell-in and sell-out are about equal in '25 or...

Well, there will always be a difference because the discount level has to be accounted for, right? So there'll be that difference in the discount level. And then the other factor in '25 and I think '26 is still a small tail to the minimum commitment deals that were done several years ago. Those are bleeding off at a slower rate because they cover a 4- to 5-year period versus the pre-paid ones cover about a 6-quarter period. So those are, like I said, they have a longer tail, but they're declining. They're not having much of an impact as we get out to '25 and '26.

Okay. Maybe we can talk about your market share. I mean you've been doing well, you had some win backs around 54% of cars. I mean what does that mean for your market share? And can you talk a little bit more on your market share and how that might vary by geographic region?

So, when doing the contracting, right, so we are working also with the headquarters of the OEMs. So for us, it's hard to get all the visibility when, for example, looking at a German OEM. They're doing pretty well still in China, right? But we have only this total amount of sold cars, right? And they're getting from us maybe the solution in 30 or 40 languages. And then we're collecting the revenue here. So it's hard to predict. But nevertheless, we can say China is, for us, a very important market segment, especially when looking at the big players in China. They have high ambitions here when it comes to overseas. I mean, this week there's actually a very popular automotive show in Germany, right, and all the big China players are over there. And they're trying to go through the Western world, Europe first, and then maybe there will be here some days, right? What I can tell here is that they are building really excellent cars here and leveraging our solution stack. So for example, EV makers like BYD, like Nio, right, they have already this multi-seat intelligence. You're sitting in the car, you can speak from every seat in your car to the system here. They are bringing also fancy new technologies with Avatar. Nio invented a physical Avatar. Renault is focusing on a more digital Avatar here, right, and they're leveraging our technology, which is great for us, right? So the European OEMs are struggling a bit also with the pricing scheme. But nevertheless, I mean, we have a very diversified customer network. So we are benefiting actually from this progress made in China and everywhere else.

I realize your technology is powertrain agnostic, but some of the electric vehicles tend to be more digitally enabled. Same thing with more advanced ADAS and even AVs, right? The most advanced electronic architectures are in those vehicles. So are you seeing any overexposure to EVs and AVs and what your content looked like…

So EV makers -- the new EV makers driving a lot of new innovations. What I just said multi-seat intelligence and so on and so forth. Nio was a perfect example, for example but looking at the traditional OEMs like Mercedes, they also just launched here in the States, the EQS, a couple of months ago and leveraging also our complete technology stack, right, and they have launched also a feature called Just Talk. Just Talk means there is no need for saying, Hi, Mercedes or Hi GM or high Volvo, right? You just speak to your system like talking to Rich, I'm not saying, Hi, Rich, can you listen to me? Hi, Rich, can you do this now or Hi Mark, right? And you understand -- also this has also a huge complexity here, right? Is it a discussion between the people in the car or do I like to have a discussion or interaction with the car, with some car functions, whatever, right? And we have provided this new functionality, and this drives also PPU.

We're in San Francisco. So, I think a number of us have probably seen Waymo and Cruise EVs driving around. Maybe talk about what sort of opportunity there is for voice technology in a true robo-taxi type of experience, things like emergency vehicle detection and what will be the price per unit opportunity…

So, good point, good catch. Also here, for example, for EVD, emergency vehicle detection, right, it's also a feature or an application actually, stand-alone application, going far beyond conversational AI to detect emergency cars. And they have roughly 1,500, 2,000 different signals worldwide. And we have this intelligence within our Audio AI suite for detecting those cars with the same infrastructure. Meaning the same microphone for a voice conversation, right, and then music is playing, and we can easily detect whether a car, emergency car, is coming from the left or the right. Some OEMs, they want to go now for exterior microphone. And with exterior microphone, we can also detect the distance to the emergency cars. So that's one of the big advantages we can offer to the OEMs for the ADAS solutions.

We talked about how you have both the connected business and embedded business. When you think about the more connected types of products, I think you can measure usage rates and how typically your products are being used. Anything you can share on the actual uptake of drivers and how often they're engaging with your products?

So when doing this for actually for monthly active users, right, we see a similar behavior like people talking to their Amazon device at home, right, 4 or 5 transactions per day here. I think overall, we see a huge uptick actually in monthly active users here was a growth of 50% year-over-year. I think we have also opportunities via OTA or fresh apps to create more domains for the users, right? We are focusing also now on generative AI, large language models, right, because the young generation, they are talking casually more naturally to the system here, right? And I think what we also presented at our last earnings call that was the so-called Cerence Assistant with co-pilot capabilities.

Can you speak to the opportunity to sell to the installed base once the license has expired? And are consumers or the OEMs the ones that are going to be paying Cerence for that?

So, the license never expires, right? So the license is applied to the car. From our perspective, it's one and done, except for now the opportunities to upsell as cars have now this OTA capability, the ability to enhance that initial installed capable system. So that does provide some revenue opportunity for us. But at the same time, the over-the-air updates is something that's just starting to really permeate the market. For a lot of cars in the road, they don't have that capability. But it is something that we do see that opportunity moving forward.

Some news this week was in addition to the senior management team, Christian Mentz, joining Cerence and Stefan. I was hoping you could maybe share a little bit about his background and what role he'll be taking on.

Yes. So Christian Mentz, he was GM at Amazon, and he was running sales and marketing for the smart vehicle division of Amazon. He joined us officially on Monday. He will be -- or he is now our new Chief Revenue Officer. Christian is well connected across the globe in North America and Europe. His wife is from Canada. Now he's living in Munich, is struggling quite a lot, and he brings in a lot of experience and he is well expected by all OEMs across the globe.

That's great. We look forward to seeing him at the investor events, and hopefully, he can have a nice impact for the company. We've spoken on AI a few times. I know it was a big focus on your earnings call a handful of weeks ago. When you start selling some of these more AI-capable products, and what does that mean for the PPU? I mean we talked about 480 as a target in fiscal '24? I mean can it be high single-digit dollars, double-digit dollar if you execute on some of the things that AI could bring?

So, we have also here high expectations on our new applications, for example, car knowledge. I think we have currently 18 POCs prototypes with OEMs across the globe. The feedback is really quite positive here. I know some of the big OEMs, they tried also ChatGPT to create their own car manual. We're going far beyond car manual here, right, and our solution is more accurate, more credible. And what's also very important is it's extremely fast in the response and behavior.

And given all of the investment that is taking place in AI development, large language models, you're using up lots of GPUs and compute resources, I'm curious are you seeing any impact to Cerence when you think about the R&D or CapEx intensity of your business?

I think in general, we see also an increase in OpEx here, but we will be within our cost envelope. What's important for us is that we are leveraging third-party language model for generic stuff, general stuff here, right? And then we're advancing those language models with our vertical data and vertical expertise.

And, Richard, I think you and I have talked about your ability to hire engineers beyond just the Bay Area and in some of the international talent where you found some good capabilities, but it's also been pretty efficient in terms of the overall expenses.

Yes. So it's interesting. We're a U.S.-based company based in the Boston area. But from an R&D perspective, the U.S. is probably the smallest region from an engineering perspective. So, we have about 300-plus people in China, which is a key R&D center for us in a few different locations there. We have about 300 people in India, another important area for us. Another large contingent in Montreal, Canada, which is very important in terms of our technology. And then, of course, over in Europe in several locations in Germany and Italy and the like. So, we're pretty diversified from an engineering talent perspective. But certainly...

And then we have various PS hubs, professional services hubs, very close to the OEMs that are also very important.

You are ultimately tied to what auto production does to some extent. So, could you elaborate a little bit more on what Cerence has seen in terms of auto production trends by region?

Yes. By region, as Stefan mentioned, right, we don't get reports from our customers. When we get a royalty report that says how many cars they shipped, they don't tell us where those cars were shipped to. So we don't really have that insight per se in terms of how that's going. But I think in general, like we said, 54% of new car production has some degree of our technology [indiscernible]. So I think it would be fair to say that wherever there are puts and takes in car production, we're still kind of following that trend to a large degree. And that 54% is a trailing 12-month number. It's actually -- if you look at it on a quarterly basis, it certainly bounces around a little bit, but it can be higher than that on a per quarter basis. So, I think we generally, as we've looked at the data, we are generally in line with IHS production numbers in terms of where our license business is going, if not slightly ahead of what IHS production is doing. So overall, we're not going to deviate that far from the market dynamics in terms of what the production forecast is going to be.

Got it. It makes sense. The company at the last Investor Day did give a 2024 revenue target. I think $385 million was the scenario. Maybe remind us what was assumed as sort of the underlying input? Global auto production, was it $85 million, $90 million penetration rates and some of those key inputs that you would have used to derive that.

Yes. I mean 2 of the auto production certainly is one. I think we used about 1.5% growth target in terms of production. But what we've talked about are 2 factors that will impact that $385 million number for fiscal '24. One of them was the consumption period for the fixed contracts, some of it pushing into -- a higher number pushing into fiscal '24 by about $10 million. So, we said, whatever you do, take that $385 million, lower it by about $10 million because of the dynamics. So, what that meant was as we typically model a fixed contract to be consumed over a 6-quarter period and we had 1 contract early in the year that was an 8-quarter contract and another one late in the year that was a 4-quarter one, right? And so that consumption pushed about $10 million into -- the net-net of that was $10 million into fiscal '24. The other thing is while we see the increasing price per unit for a number of different programs, the timing of when those programs will go in to start production, in some cases, have been delayed. Now we typically will have the older program. So it's not like we're missing out entirely on that revenue, but we are getting impacted by the delay of these programs on that incremental contribution. We haven't quantified what that is yet, because honestly, we're going through that analysis now. And we'll provide our fiscal year guidance on our conference call in November. And then we'll also update the '25 and '26 outlooks as well. The objective of the company remains the same, which, over the next couple of years, to get to that double-digit grower with north of 30% EBITDA margins. That what the goal of the company was when we laid out that plan last year and it remains the goal of the company.

Yes, that's very helpful context. In terms of the production timing starts of some of these newer programs that have more content, any common reasons why some of them may be taking place a little bit later? Is it lingering supply chain impacts or something else that may be at work?

It's actually, in my view, it's also an issue of some of the OEM software houses, right? Just to give you a crisp example here, it's CARIAD it's a software house of the VW Group. We are the preferred supplier for VW, including Audi, including Porsche, including SKODA and so on and so forth and Fiat here, right? And there was an announcement a couple of months ago that they have laid off their complete management team because due to these delays here, right? And of course, we're suffering from those delays here. We see it also for some other software houses, [indiscernible] I believe that they're still struggling from COVID. You have to bring back to people somehow that they can work together. I mean, automotive is really complex here. We have to deal with a lot of hardware components, right, software components, right, and everything has to work seamlessly together. That's one of the biggest challenge in automotive.

Understood. The pro services business has been a nice contributor to the company. Maybe talk about what that means in terms of lead generation and program generation for the license in connected business. And even on a standalone basis, how large can such services become?

So professional services for us is very important, right? So, they're working closely with the teams of the OEMs. They are doing all the integration and customization work. Professional services for us is actually not a margin driver to be honest, but they asked for building the right products because of the OEM.

From a revenue perspective, we said the pro services, if anything, is flat to down over time. And the reason for that is in the past, where we would be working with a customer, each solution would be somewhat of a snowflake, right? You have to put a lot of work into create it from scratch. And what we've done is we've taken the technology and create more of a base functionality system in our Cerence Assistant so that there is a much higher level starting point that we then take and extend and customize. Well, having that more better starting point, which is a benefit to our customers and to us means that it doesn't drive as much professional services because you're not, again, creating a snowflake every time you're working with a customer.

And maybe one additional comment to it, right? So our new software stack Cerence Assistant is platform-agnostic here. We can do a port to a specific head unit within 1 to 2 weeks.

We've got time for 2 or 3 more questions. I can ask them, but I did want to give the audience an opportunity if anyone has a question for the Cerence team. Well, if anyone wants to ask a question as we go through, just raise your hand. But one of the one I wanted to dig more into was on some of these adjacent end markets. You mentioned a little bit already about 2-wheelers, cruise ships, elevators. How much revenue is Cerence getting from some of these end markets? And as you think out 2, 3 years, how much could that grow to?

So let me go first. So over the last few quarters, we have signed 7 2-wheelers, big ones in China and in India. Also here, a big one here in North America, in Europe and also in Japan; 4 went live last quarter. I mean, there's typically a ramp-up phase, right? We're not expecting a lot of revenue for this fiscal year, but we have also high expectations for next fiscal year here. On the truck side, we are also doing pretty well. Also here, we have a couple of design wins. They will go live next year. We have also one deal with a recreational vehicle here. So overall, we are doing pretty well. We have also some discussions when it comes to [ automotive ], but it will take some time. So overall, we are doing, in my view, pretty well also for adjacent transportation market. The other big opportunity we have, of course, is looking beyond next fiscal year. The field of use restriction expires next year in October. And then we have all the flexibility offering our solutions to other non-automotive segments here. Just think about industrial application, health application, even home. We have also established a team with business development sales and their own R&D team. They're focusing on paving the way for this new opportunity in FY '24.

I'll go ahead and ask the last question. Just how to think about the progression of cash flow? I think the company guided to have positive operating cash flow this year. How should that track relative to EBITDA as you think about fiscal '24, fiscal '25?

I think one of the things certainly we laid out in the Analyst Day that cash flow should be on a really positive track. Part of what drives that is that -- you may be familiar, we have this legacy revenue from Toyota that we're reporting on a quarterly basis that is just running out that has no cash associated with it that was collected mainly by Nuance prior to the spin, but we're reporting out on all the -- reporting the revenue because it's amortized over the duration period. But what we're doing now is we're continuing to ship more and more connected cars, we're getting the opposite effect moving forward, which is we collect the cash upfront for that subscription period and then we amortize it. So as we're seeing the good increases in connected cars that we're shipping, that's going to help drive that cash flow positive moving forward.

Unfortunately, we are out of time. Rich, Stefan, thank you for joining us.

Thank you, Mark.