Cerence Inc. (CRNC) Earnings Call Transcript & Summary

Great. Thank you, everybody, for joining us. My name is Mark Delaney, and I cover the autos and industrial tech sector at Goldman Sachs, and I'm very pleased to have with us today the Cerence team, including the CEO, Sanjay Dhawan; the CFO, Mark Gallenberger; and VP of IR, Rich Yerganian. Before we get going, I would like to remind everyone that this session is not intended for the media and is only intended for clients and is considered off the record in that regard. So Sanjay, Mark, Rich, I really appreciate you all joining us.

Thank you. Good morning.

So I thought to get going, I would love to talk through the company's key drivers as you think about the content per vehicle opportunity going forward. Currently, Cerence has mid-single digits, mid- to high single digits content per vehicle, selling, in particular, your voice AI software into the automotive end market. The company thinks that over the longer term that it could hit $15 to $20. Some of that is coming, in particular, from some of the new applications that are part of that 2024 target model that the company had talked about. So I thought, first, to start, we can talk through some of the features and what you think drives that increasing content per vehicle for the company over time.

Sure. Let me start, and my colleagues can jump in. So the growing content per vehicle is a key strategic angle that we took on from the first day that we became an independent company, and we're basically focused on 3 different areas. So one is enhancing and adding new functions and capabilities into our core conversational AI stack, which obviously gives us opportunity to increase value of our stack and, hence, add -- ask for more increased pricing. So this is sort of the core stack, which is ASR, NLU, TTS, SSE, et cetera, which is -- which provides the core voice AI functionality, and we are adding features like Just Talk to it. Just Talk removes the need for a wake-up word and keeps all the conversation local in the car, not in the cloud, but it makes the conversation very, very natural so that you don't have to use a wake-up word every time you want to interact with the system. We're adding -- expanding the language portfolio in terms of the number of languages that we are adding and so on. So that's basically enhancing the core functionality of our stack to increase revenue per car. Second element that we are focused on, Mark, is, what I call it, allowing basically creating cloud-based extenders to allow the consumer to bring their digital life in the car. So this includes basically coexisting with the big tech. This includes our cognitive arbitration product. This includes our products like Cerence Browse, Cerence Extend, Cerence Connect. These products basically allow the consumer or the driver to bring his or her digital life in the car. So these are all add-ons onto our core platform and obviously, gives new monetization opportunities for the company. And the third vector to add and increase the revenue per car is what we call apps. So these are connected services applications. We have launched 3 of them so far. We call it in Cerence Car Life, which improves the car ownership experience; Cerence Pay, which brings in-car commerce through voice; and the third is Cerence Tour Guide, which is basically providing tour guide experience -- virtual tour guide experience using voice in the car. So that's the portfolio, and that's the strategy that we are implementing right now.

And when you think about all those different new features and capabilities the company is deploying, is there any one in particular or a couple in particular that are a bigger proportion of that $90 million of revenue you're expecting by 2024 from those -- the pretty newer products collectively? Or is it pretty balanced?

That's like asking which of your kids you like more. It's -- I get excited about all of this, but in Cerence apps, Tour Guide is getting a lot of traction in the market. We'll go live SOP in January next month on one of the big OEMs with that product. It's just a -- it's a simple concept but has an amazing kind of experience for the consumers. I mean just think of having a tour guide in the car with you while you are driving through sort of new places and so on and so forth, you can create sort of like a Spotify where you can create a playlist. Now you can have a tour guide. You can create a tour list, and your car navigation plus voice basically takes you through that journey, tells you about your journey, helps you with booking museum tickets or whatever else you want to do in that tour and so on and so forth. You can publish your tours for others to enjoy where others can do the same thing and so on. So it's a very interesting, attractive product, and we're getting a lot of traction on that. We're really excited about Extend as well because Cerence Extend allows a consumer to bring his or her digital life in the car. As you know, Mark, the #1 device that has a digital life is the phone, right? And you can obviously integrate your phone in the car with CarPlay or Android Auto, right? But as soon as you do a CarPlay or Android Auto, firstly, from an OEM standpoint, they lose complete visibility and control. They have basically given the control to either Cupertino or Mountain View. They have no idea at that point what the consumer is doing in the car. So they lose their branding. They lose their control. They lose their visibility. Everything, right? Any monetization all goes to the big tech companies, and what we have done is we have created Extend, which basically allows to remote control your apps on your phone without touching the phone through voice, and the app developers on the phone, iOS or Android, they don't have to do any development at all, any special development for us. Our SDK goes under the phone through the companion app of an OEM, and that's it. And now the consumer can do whatever they want on the phone through voice and independent of the big tech, okay? So for example, I use iOS. I use Apple phone as a consumer, but I -- the messaging that I want to use is WhatsApp. I don't care about Apple messaging. I use WhatsApp, right, for -- or I use Teams. I use Teams for my business messaging. I use WhatsApp for my personal messaging, and I want WhatsApp and Teams messaging to be available and integrated to me in my car through voice. That's what I want as a consumer. If you ask Sanjay as a driver and as a consumer, that's what I want, and what we are trying to do is basically enable that through our products, independent of the big tech sort of closing the ecosystems in favor of their own apps.

Yes, that makes...

Mark Gallenberger, did I cover everything or anything else you want to add?

I think you covered it well, Sanjay.

Okay.

Yes. Can I ask another one? In particular, of interest to me is Cerence Pay, so as you're speaking and interacting with your voice to the car, if you want to purchase something, Cerence Pay can authenticate that it is, in fact, you are the person who is speaking and trying to place that purchase. Maybe you can talk a little bit on that application in particular, how much traction you're seeing?

Yes. So that application, you know what, again, the use case is very strong. OEMs like the use case. Bringing that application out on the OEM -- in the OEM ecosystems is a little bit more intrusive, meaning it takes a little bit more effort, right, to bring -- remember, we are not trying to be a wallet. We want to integrate with the payment gateways of an OEM that exist in the OEM ecosystem. And so although from a product standpoint, we're ready, ready with our product, with the integrations in the back end because there is a lot of integrations that have to happen as well, right? For example, for fuel. We want to make sure that a certain number of fuel pumps are covered or restaurant reservations. You want to make sure that we are integrated with OpenTable and Resy and others to make sure that a large number of restaurants are covered by our platform, et cetera. So -- but bringing that into the car definitely has a little bit more -- takes a little bit more effort, and we are going through that right now in terms of working with different OEMs on that, but the interest is definitely there because, I mean, the use case, Mark, makes sense, right? Why do I have to go park my car, take my wallet out, go to the pump, get my card authorized, blah, blah, blah, right, put the PIN codes or whatever, right? Why? Why can't I just say, "hey, Mercedes, pay $50 on 15," and that's it. My voice is my authentication, and the system does it in the background on its own, right?

Makes sense. And Mark, maybe one for you. When you think about all of these various applications, building up to that $90 million, can you talk about how much is maybe booked already, either quantitatively or even qualitatively, if you don't want to share an exact number?

No. That's fine. Actually, on our last earnings call, we did talk about for fiscal '21, we had about $120 million in new product bookings, and that goes beyond just what we've been talking about. It also includes some of those adjacent markets as well that I'm sure you're going to want to ask us about later on, but it's a good start. I mean we started talking about some of these applications only 18 months ago or so, and the development cycle has been pretty rapid, and now the customers are seeing the opportunities, and they're making commitments, and they're wanting to adopt some of these technologies into some of their next-generation designs. And so I think we're off to a pretty good start, and so I would expect that momentum to continue in this fiscal year and beyond.

That's helpful. And maybe just talking about how often Cerence technology broadly is showing up on cars. The company has talked about being in a little over half of all vehicles. Maybe talk about what you guys expect that to get to, perhaps what's built into the 2024 target model?

Yes. So I can take that, Sanjay. There is a secular tailwind that's been going on for many years, and the penetration rates of embedded technology continues to grow. What's even more exciting is on the connected side. The connected car is growing even more rapidly, albeit starting at a lower base. And so if you look at what IHS is forecasting, for example, on the embedded side, currently, today, it's about 69% of worldwide vehicles have this embedded AI technology designed into it. That's expected to grow to about 82% in 2024. On the connected side, those numbers are a little bit lower. Around 49% of worldwide vehicles have cloud or has connectivity designed into it from the manufacturer, and that's expected to grow at a steeper rate to about 68% in 2024. Those numbers from IHS, they were upped, I think it was last year they increased those numbers. So they're actually expecting faster adoption of these technologies, which, quite frankly, is no surprise to any of us, right? I mean digital technology gets trickled down into more and more vehicles faster than what people originally expect. And so today, like you said, we're a little bit higher than 1 out of every 2 vehicles, and if our market share holds, we would expect -- if you do the math on our current market share on the embedded, for example, which is around 80%, you would expect, in the 2024 time frame, we should be somewhere in the mid-60% range from where we are today, which is in the low 50s.

Got it. And we had a question come in from the audience that I think fits in well here is around your market share on the connected side and where is that now but also where do you see that going over the next 5 to 10 years?

Yes. We're not as dominant on the connected side. There's clearly more options on the connected side, and many of our customers actually will do some of their hosting on their own, but today, we're roughly 25%, maybe 30% market share today. Hard to get an exact number, but that's our best estimate today. We do expect that share to grow over time because we think that because we got such a high share on the embedded, as more and more cars get connected, [ for ] us, it feels like it should be a natural extension for the automakers to just try to combine those 2. Obviously, I don't need to, but we think that having our position already on the embedded side of the car, that gives us a pretty strong competitive position to win those connected opportunities.

Got it. And maybe either for you or for Sanjay. I mean when you think about your attach rates on EVs versus cars overall, right, I mean you shared the EV specific attach rate of over half of all cars have Cerence on it, but what would that look like if we just went and looked at the EV part of the industry?

Sanjay, do you want to cover that one?

Yes. Yes. No, Mark, we don't -- unfortunately, we don't -- to the best of my knowledge, we don't track EV -- exact EV attach rate, right, because we're indifferent to EV versus non-EV cars. We are more -- we track more the connectivity and kind of connected car, which can be an electric vehicle and non-electric vehicle, right, both. Having said that, we do work with many, many EV companies. Tesla is a customer of ours, so is NEO, so is Xpeng, so is a bunch of others, right? So we announced Volkswagen ID3, ID4 as a customer. So there are many, many EV cars that has our platform in it, but we don't -- to the best of my knowledge, Mark, I don't think we look at EV shares, right?

No. I don't think we have that hard data, but it's -- I think our view is a lot of the EV companies are -- a lot of them are start-ups, of course. They're not legacy auto OEMs, and as a result of the fact that they -- they're startups and they have the ability to design in right from the ground up a lot of this technology, my assumption would be that most of them are going to have much higher penetration rates just from the get-go just because of the sheer fact that many of these companies are start-ups. In the grand scheme of things, there's not a lot of volume relative to the more traditional OEMs, but they're clearly going to be strategically important to us. So even though they may not be driving much business from a sheer volume perspective, I think it's very important for us to get in there from a strategic point of view because a lot of these companies, I'm sure, are going to grow pretty rapidly over the next several years.

The EV that I'm most excited about, Mark, is EQE and EQS Mercedes that's coming out. I was in Stuttgart a few weeks back with -- now that things are opening up, I started to travel over the last few -- last couple of months. So I personally traveled to Detroit first to meet the big 3; then to Germany, met all the OEMs there; and just recently U.K., France for the OEMs in that countries -- in those countries. And when I was in Germany, I was with Mercedes-Benz Stuttgart, and they were kind enough to bring the new EQS, the new S-Class fully electric car, which has a lot of our technology in it, and they wanted me to experience it firsthand, and what a beautiful car. Oh, my God. I'm a fan. Just absolutely gorgeous, right, both inside out, and some of the functionality inside is just absolutely amazing, right? So that's -- and very good range as well. 700 kilometers was what they told me on full charge, which is quite good, right? I don't know. It's -- I haven't done the math. That's north of 400 miles, right, 450 miles or something like that.

Yes. No, it's always fun getting to see the latest and greatest that's coming out. I wanted to dig in a little bit more into some of the comments we had and also following up on some of the prior discussions we've had around how you compete with and also coexist with some of these big tech voice assistants, and Sanjay, you talked about being able to work and coexist with them, and I know you've had that view for a while. Last time we spoke, you said there's a number of examples you could think of where Cerence was on vehicles, along with offerings from Google or Apple, and there's limited situations, if any, that you know of that you were being displaced by them. Maybe if you can give us an update about where that stands. I mean is that still the case where you're coexisting with big tech? Or are you seeing any examples where OEMs are saying, "Look, we can take something for free for one of these big tech companies."?

Well, firstly, in life, Mark, nothing is free. If there is somebody saying something is free, there is a catch, right? So if Big Tech is saying it's free, please go and look at it. It's not free. The -- I think the co-exist story has sort of continued the same way over the last 1 year. The -- so I'll -- let me just take 3 Big Tech, right, Apple, Google, Amazon, right? For Apple, there is really no change. They -- Apple is only on the phone. CarPlay is the only way to integrate. To run CD in the car, there is -- it still goes through the Cerence SSE where we do kind of speed signal enhancement and all that stuff, and then we have an alternative solution in Extend that I talked about earlier in the car in the call. So Apple is not embedding themselves. They are just staying on the phone basically for now, right? The Amazon with Alexa, they clearly have a more cloud-first strategy, embedded experience at least. It's a heavy platform. So embedding it is usually heavy for the car and for the OEM in terms of resources and all the stuff. And many OEMs -- I'm sorry. I had an interruption here. My apologies. Can you hear me okay now?

Yes.

Yes. Sorry about that. So the -- there's an incoming call that's interrupting the session. I'm so sorry. So basically, for Amazon, we coexist on at least a dozen different OEMs, right? So you probably have seen the ad of BMW and Alexa. We are also in the same car in terms of BMW, and so Cerence and Alexa coexist very nicely where we do the wake-up word arbitration. Same thing is true. You may have seen an ad on Buick and Alexa. Same thing. We are on that Buick and Alexa. So we're kind of happy coexisting with each other in the car with Cerence providing the underlying technologies to do the arbitration and all that stuff, right? Google is a little bit more different story because Google is more -- wants to be -- their products for both on the embedded side and also on the cloud side. So with their droid automotive operating system and the GaaS offering for the car, they want to embed both in the car and in the cloud, and there are fewer OEMs who have looked at them, right? There was Polestar, right? That was the first car that was -- Volvo Polestar, that basically launched Google Android automotive and gas solution. A good product, but having said that, it's very limited to the Google ecosystem, and as you and I would know, Mark, no one Big Tech can own the complete digital ecosystem of our consumer. If I look at my digital ecosystem, I want -- I use Apple devices and services, Google devices and services, Amazon devices and services, Microsoft Calendar, Outlook, Teams and all that stuff. I want all of them. I don't want to be limited to only Google or only Apple, right? And so what we are trying to basically do is sort of provide that multi-Big Tech connectivity in the car, right? And Google has started to open up a few months -- I think about a year ago, in November, December last year, they opened up their Android automotive APIs to allow other assistants to coexist with them, and this just happened, like I said, in the last 1 year, and the OEMs are basically considering that, and they're obviously talking to us as well to coexist on the Google platform itself.

Got it. Maybe a good one to layer and hear from the audience around what kind of ASP do you get when you're just a voice arbitration as opposed to maybe selling a broader set of technologies.

Yes. So I don't think there is any case where we are just a voice our arbitration because when we are voice arbitration, we have the rest of our stack there as well. Obviously, if we are just a voice arbitration with the wake-up word and all that stuff, and somebody else is doing the real -- the other heavy lifting from an AI standpoint, obviously, the contribution to ASP will be lower, clearly. Having said that, I don't know even of a single case where we are only doing the arbitration. Mark?

No. I can't think of any case either, Sanjay.

Yes. And then just on a kind of a related topic in talking to with market share. I think it is coming in from the webcast audience, and so thank you for everyone who has been sending questions. We've got a question around, "OEMS are trying to take a lot of software intelligence themselves and hiring lots of software engineers building out a lot of those capabilities. To what extent do you see that as a potential threat where any OEM may try and in-source some of this sort of capability? Or in fact, are you able to licensing potential where you can license your technology?"

That's a very good question, and yes, that trend is real. Mercedes started that with setting up a shop in Berlin called Ambition. Volkswagen announced it, and they created car ID where they consolidated all their software resources under one roof, and then there are others also doing similar things like Syvantis and others, right? So it's a real trend, and the driver of the trend is that OEMs want to basically control the system in -- the software integration piece. They want to control the architecture and the software integration. So there -- with every new generation of vehicles, they don't have to go back to scratch to 1.0. The old model, Mark, was that every time a new card model was released, the software would go back to 1.0, and then you start building it up again. The software is getting more and more complex, and you have to separate hardware and software. Sort of what has happened in the PC, what has happened in the -- on the phone. Same thing is happening in the car. And the capabilities that the OEMs are putting together and many of them, Toyota has Toyota Connected, right? Same thing, right? Great, great team. The capabilities that they are putting together is more around defining the software architecture and also controlling the integration. Under no scenario an OEM can build every single LEGO block that goes inside the software. So they can't build their own maps. They cannot build their own voice AI. They cannot build their own operating system. They cannot build their own -- I can go on and on, right? So what they do is -- what they're doing is they're defining an architecture that, okay, my architecture will be Android based or my architecture is going to be Linux, RT based or my architecture is going to be QNX based, and then let's define the building pieces, and I control which pieces I buy, how I put it together, and if one piece is not working, sort of like a LEGO building block, I take it out, put something else, right, but that control is with me. So the business model that is being disrupted by doing this, Mark, is the Tier 1 business model, not our business model, right? So in my previous job is to be at HARMAN or if -- the other -- the Tier 1s, the traditional Tier 1 used to do a lot of system integration. That piece is being in-sourced by -- a portion of that piece is being in-sourced by the OEMs.

Got it. Okay. The company did just have its fourth quarter earnings call and gave the fiscal '22 outlook. So I had a few questions, and I think similar to a few things that we're sending from the audience as well, but maybe just to talk through the outlook for next fiscal year. One of the points the company made was that you're assuming flat auto production in fiscal '22, which is consistent with the IHS forecast. I'm curious, is that also what you're OEMS are telling you, and it kind of aligns with what IHS is also expecting? Or is there any kind of a disconnect there?

Well, generally, we do like to get different points of view. IHS is one critical component, and of course, our own sales force working directly with our customers is another critical component. At the end of the day, a lot of this information is coming from the same sources anyways, and so I don't -- I would not expect any material or significant differences. There's always going to be maybe different degrees of it. But I think some of the sentiment out there with customers right now is still a fair amount of uncertainty as it relates to the semiconductor supply chain, and that's why I think IHS took their numbers down last few months. I think they started taking it down pretty dramatically in the summer time frame, if I remember correctly. And so if you look at our fiscal year and line it up with the IHS forecast, it does forecast a flat year-over-year during our fiscal year, and so that's how we based our guidance off of was that flattish, if you will, sort of growth right now. Now with that said, even with that, we are expecting to grow above auto production rates because of what we've been talking about, the cars becoming more and more digital, the cars getting more and more connected, and that's what's going to enable Cerence to continue to grow even above auto rates even if it does end up being a flat year.

Got it. A question, I think, makes sense to insert here that came in from the webcast audience was you're talking not just about bookings but also backlog, and this relates to the new applications that we were talking about earlier in the session, and I understand bookings accelerated this year half-on-half. How much of the new applications are in backlog?

So let me -- go ahead, Mark.

No. No. Go ahead, Sanjay.

Please.

No, no, no. Go ahead.

Okay. So as Mark said, of the 600 -- $590 million in bookings for fiscal '21, $120 million came from new products and apps. We didn't break that down into separate. So that's basically feeding into the 2 line items, the $90 million and the $65 million, those 2 line items for the fiscal 2024 model. But I'll just kind of peel the onion a little bit on the fiscal 2024 model just for yourself and for the audience here, right? So the fiscal 2024 model that we put out basically has 4 revenue categories. The first one is edge AI license that we are forecasting to be about $300 million, and that basically is coming from about -- if you look at our backlog, the $2 billion backlog, which is Slide #17 on our -- in our deck -- earnings deck, you'll see a backlog of about $940 million or so, $950 million. I'm speaking from memory. I don't have the slides in front of me. Which basically is the backlog that's feeding the edge AI license revenue, okay, for the coming fiscal years. Clearly, we're adding more. We're also using from that backlog as the backlog converts to revenue. The second piece is the connected AI, which is connected services, and that is forecasted to be $225 million in fiscal 2024. Supporting that is basically 2 of the backlog items: the new connected services and the legacy. When you add those 2 up, that's about $940 million in backlog, and that backlog is speeding the $225 million in revenue in fiscal '24. I would like to see that backlog increase to about somewhere from between $1.3 billion to $1.5 billion from $940 million right now, and the reason is that if you do simple maths, you take the backlog for connected services and legacy, divide it up by if you want to be very conservative; 7, if you want to be on the other end of the spectrum, this is the number of years that backlog will be serving, and you'll get a number out, right, basically, which is all the backlog is not that linear, but I'm just doing a simple maths there. So for that reason, I would like that number to be higher. We mentioned on the earnings call that there was one large contract that slipped out from Q4 into Q1. We don't see any issues in that. It's one of our largest. The OEM is going through internal restructuring, which the one that we talked about earlier, the software sort of restructuring, and that was the reason why that contract slipped out. That probably will add, I don't know, a couple of hundred million or so, right? So -- and then obviously, we have a pipeline where our sales team is working hard to keep increasing the connected services backlog. So that's that. The third item is the new adjacencies. We're forecasting that to be about $65 million in fiscal '24. We, right now, have no backlog. We have not put anything in our backlog for that line item, and the reason is -- and we mentioned this on the earnings call, that there are 3 customers that we already have. One is signed. Second one, we have an award letter and so on, right? But we have the elevator company that we have signed with. We have the contract. Our sales team thinks it's a pretty good revenue opportunity, but Mark and I decided not to put anything in our backlog yet because we would just want to be conservative. It's new market. We don't have penetration data. We don't have penetration forecast from the customer. We are waiting for those things to mature as they launch the product and all that stuff, but we expect revenue this year, right? But we have not put anything around backlog for that or these other couple of companies where we have the award letters, we are marching towards a revenue shipment. And the last item on the backlog list is professional services. We are forecasting $110 million in fiscal '24 for that. We have a backlog of $120 million, but this business is -- 50% or more is book and ship. So we don't see any issues in sort of meeting that number for fiscal '24 model. So that's my bridge, Mark, from where we are right now from the backlog standpoint to fiscal '24 model. So some will work definitely to be done on the connected services side to increase the backlog. I would be -- I would feel good, like I said, at that $1.3-plus billion number for connected services.

Got it. That's very helpful. Just one last one, the 10-K is out. We get some questions come in around this from some of the investors, which was around the amount of revenue that's been recognized but unbilled. It's been increasing, and we did see that in the 10-K. So Mark, I don't know if you have anything you can share around -- are you guys changing your business practices or doing it? What's leading to that?

Yes. The biggest driver of that, Mark, is the fact that our fixed contract license revenue went up substantially this year versus last year, and with that, there's components of payment terms. Some of those payment terms, if they're not scheduled yet, then it becomes unbilled or if there's going to be some minimum commitment deals, those are obviously unbilled as well. So that's where you just see that natural movement up in that fixed license contract line from last year to this year. This year, we're forecasting the fixed contract revenue to actually come down. So I would expect by the end of next -- or this fiscal year, a year from now, call it, that, that number would probably start to come back down because of what we're forecasting this year.

Okay. Thank you, guys. We are unfortunately out of time. Really appreciate you joining us and doing all the questions that we had, and thank you for everyone in the audience who nicely submitted questions. Is there anything else that we can do to follow up, please don't hesitate to reach out. Thank you so much.

Thank you very much.