SoundHound AI Inc

NASDAQ:SOUN

Earnings Call Transcript 2023-Q2

[Operator]

Good day, and welcome to the SoundHound Second Quarter 2023 Earnings Conference Call. At this time, all participants are in a listen-only mode. After the speakers' presentation, there will be a question-and-answer session. [Operator Instructions] Please be advised that today's conference is being recorded.

I would now like to hand the conference over to your speaker, Mr. Scott Smith, Head of Investor Relations. The floor is yours, sir.

[Scott Smith, Head of Investor Relations]

Good afternoon and thank you for joining our second quarter 2023 conference call. With me today is our CEO, Keyvan Mohajer; and our CFO, Nitesh Sharan. We will begin with some short remarks before moving to Q&A.

We'd also like to remind everyone that we'll be making forward-looking statements on this call. Actual results could differ materially from those suggested by our forward-looking statements. Please refer to our filings with the SEC for a detailed discussion of the risks and uncertainties that could affect our business and for discussion statements that qualify as forward-looking statements.

In addition, we may discuss certain non-GAAP measures. Please refer to today's press release for more detailed financial results and further details on the definitions, limitations and uses of those measures and reconciliations from GAAP to non-GAAP. Also note that the forward-looking statements on this call are based on information available to us as of today's date. We undertake no obligation to update any forward-looking statements, except as required by law.

Finally, this call is being audio webcast in its entirety on our Investor Relations website. An audio replay will be available shortly following today's call.

With that, I'd like to turn the call over to our CEO, Keyvan Mohajer. Please go ahead, Keyvan.

[Keyvan Mohajer, Chief Executive Officer]

Thank you, Scott, and thank you to everyone for joining the call today. In Q2, we started our second year as a public company. With our first full year coming to an end, we are happy to have reported over 80% revenue growth on a trailing 12-month basis. We ramped up our innovation cycle and at the same time, gained efficiencies across the Company as seen in the 50% year-over-year adjusted EBITDA improvement in Q2.

Over the years, we've demonstrated our track record as a leading innovator in AI. This is one of the main reasons we've been able to attract strategic investors like NVIDIA, Samsung, Tencent, Hyundai, Naver, HTC, VIZIO, Foxconn, Oracle and others. While some organizations have spent billions of dollars trying to build voice and conversational AI technologies in-house, we were able to do that with a fraction of that investment with our creative innovation and focused dedication.

As a result, our voice AI-powered millions of cars, TVs, apps and IoT devices, generating billions of successful user interactions each year. Because of our root in deep science, over the years, we've created groundbreaking advancements in voice AI from Speech-to-Meaning, Deep Meaning Understanding, Collective AI, Dynamic Interaction and our CaiLANCaiNET framework and globalize them in dozens of languages.

As an example of another recent innovation, we introduced SoundHound Chat AI for automotives. This combines our proprietary voice AI with the latest generative AI built from large language models. In our view, SoundHound Chat AI is completely unique in the industry and no one else can claim this type of progress. Within a short time frame, several of our existing customers have started piloting the solution in their products and we've received proposal requests from new brands. We expect the increased performance resulting from SoundHound Chat AI will result in faster adoption, better user experience, more frequent usage and ultimately, more revenue per device from our Pillar 1 category of automotive and IoT customers.

In a recent pilot conducted by one of our key customers, the usage increased by an order of magnitude after enabling the features. We expect adoption of our Chat AI platform from some of our customers as early as this year. In Pillar 2, which offers Conversational AI for customer service solutions, our Smart Ordering continues to benefit from our unique technology advantage of handling complex and compound queries, which is especially important in food ordering conversations over the phone. In the last quarter, we were able to further improve our solution by applying our generative AI integration really taking it to the next level.

Now, I'd love to play a sample dialogue with our next generation of Smart Ordering AI that is rolling out in the coming weeks to our existing and new restaurant customers.

What can I get for you?

Hi, I'm calling to place an order. My daughter is three years old and she is a picky eater. Do you have any recommendations for her?

Of course, for your picky eater, I would recommend our Kids Mac and Cheese. It's a classic favorite among children. We also have Kids Chicken Fingers and Kids Grilled Cheese if she prefers those options. Let me know if you need any more information or if you'd like to place an order.

Yes. Can I get the Kids Mac and Cheese with apple juice and yogurt?

I've added a Kids Mac and Cheese with apple juice and yogurt to your order. Can I get you anything else?

Yes. I was there last week and had one of those cheese burgers that was really good. I don't remember what you call it, it was something like Traditional or actually Classic. Yes, Classic. Can I get one of those with Cheddar cheese?

Sure. I've added a Classic cheese burger with Cheddar to your order. What else can I get you?

That's all. Thank you so much.

Most recently, we are excited to have launched our low-touch out-of-the-box AI customer service solution, Smart Answering, which will be formally announced soon. With Smart Answering, we expand our AI customer service offering from just restaurants to any business, even a single location barbershop that can benefit from SoundHound's AI customer service. Here is a short demo of Smart Answering with the barbershop.

Thank you for calling Smith Brothers luxury grooming salon for men. This call is recorded. And if you need privacy info, just ask how can I help you today?

Yes. Hi. I wanted to surprise my daughter who loves doing Barbie movie by dying my hair blonde like Kent. [indiscernible] Can you guys do?

Yes. We offer hair coloring for men at Smith Brothers salon. The cost starts from $250. Is there anything else I can help you with today?

Well, honestly, I've never colored my hair. So, I don't really know how it would react. I was wondering if I could schedule something just to ask a few questions and get some advice.

Great. We offer professional consultations for hair coloring. You can easily make an appointment on our website. Would you like me to text you a link to make it easy for you.

Sure. That sounds great.

Okay. I have sent the link to you. Is there anything else I can help you with today?

Do you have any advice on how to get there? Should I drive or take public transit?

To get to Smith Brothers salon via public transit, you can take the Brown Line to Addison Station, which is a five-minute walk from our location. If you decide to drive, we don't have any dedicated parking, but there is usually street parking available on Lincoln Avenue. Is there anything else I can help you with today?

Well, actually, yes. Do you allow dogs?

Yes. We allow dogs in our salon. If you'd like to learn more about our pet policy, I'd be happy to text you a link to our website where you can read additional details. Would you like me to do that now?

Well, the problem is that my new dog is a rescue and she gets a bit rowdy when around other dogs. I wanted to get used to going out, but it's best that there aren't any other dogs in the salon at the same time. Is it possible to figure out a time when that would work?

Okay. I will pass along a message to owner, Jared Smith, so that he can be in touch with you via text, to discuss arranging a time when no other dogs will be present. Is there anything else I can help you with today?

No, that's it.

Thanks for thinking of us for your hair coloring needs. Have a great day. Good-bye.

Awesome. Thanks. Take care.

These were examples of our AI customer service solutions that are fully autonomous and a key differentiator for SoundHound, separating us from other vendors that use a human in the background when users engage with their AI offering. As compared to human-centric low-margin call center solutions, our fully autonomous solution is affordable and can attract brands of all sizes, whether the brand has a single location or thousands of locations. When a caller interacts with our AI, they're only interacting with AI and there is no human involvement. If a caller specifically asks to talk to a human, usually at the beginning of the call, our merchant have the option to receive the call directly or forward to an agent.

For example, during the last week of July, about 93% of orders placed with our Smart Ordering solution were handled by AI and only 7% were from transfers. We expect the rate of these transfers to decrease over time as callers become more comfortable and confident speaking to an AI. In fact, we expect the callers will ultimately prefer to speak to an AI instead of an human. Our AI is also always learning, both generically and specific to our merchants. After going live with the merchant, our AI capabilities and accuracy improve as it learns to interact with users specific to a particular business. With this technology, we believe we can unlock long-term value for businesses.

We've heard from restaurants loud and clear. They are hungry for this type of automation as it directly creates financial returns by addressing labor shortages, overall cost pressures and help them generate revenue. We are deeply engaged with thousands of brands and have already closed with hundreds of them, thanks to our direct sales and strong channel partner ecosystem, including Toast, Square, Oracle and Olo. Last week, we announced that we are significantly expanding our relationship with White Castle, where they're committed to provide AI-powered ordering to 100 drive-thru lanes by the end of 2024.

And because our products are designed to be scalable, we can address a wide vagary of restaurants across a diverse area of cuisines and different sizes. For example, we are also delighted to be providing voice AR technology to Beef 'O' Brady's, which is a part of FSC Franchise Co., is a restaurant company with multiple brands that span 25 states and more than 180 locations. And you're adding more customers every day. For instance, Kneaders bakeries, Hot Table, Naz's Halal, Slim & Husky's, Crust Pizza, Kumori Sushi, Noi Thai, Bozelli's Italian and Dialog Cafe, among hundreds of others, note especially the diversity of cuisine types and brand sizes, which give us access to a larger market share than vendors that due to technology and scale limitations need to focus on just enterprise customers or specific cuisines.

And with Smart Answering, we're just getting started, allowing any business to handle nearly any inbound customer inquiry using automated and customizable voice assistant. Using generative AI to automatically incorporate data from the business website in its AI knowledge base, Smart Answering provides quick and easy onboarding, allowing the customers to test it out within minutes and be live on the same day. The service can handle multiple calls at once 24/7 and continuously filters out spam calls, saving hours of wasted time for any business. It can also provide convenient SMS responses in addition to verbal answers and it can seamlessly transfer to any employee of the business, if necessary.

Smart Ordering and Smart Answering are complemented by our visually enhanced SoundHound Dynamic Interaction, which is our multimodal full-duplex interface for kiosks, drive-thru or any device with a screen. Dynamic Interaction is unmatched as a breakthrough and we believe will mark the next generation of digital assistance and AI interfaces, which allows input via speech or touch and provides real-time responses via visual and audio with no turn taking or wake words. I encourage everyone to visit our website and watch the demo of Dynamic Interaction. We consistently get a wow response anytime someone has introduced this product for the first time.

When we attended the National Restaurant Association Conference this year, Dynamic Interaction stole the show. Trade Media selected Dynamic Interaction as one of the top things to see on the shop floor among all vendors in all categories, so not just technology. A typical reaction from our target customers is not just if they're interested in SoundHound Dynamic Interaction, but how can we go live with this as fast as possible. We are already working with a number of potential customers, including a fast-growing large franchiser that operates in 30 states and we plan to start rolling out Dynamic Interaction in the coming weeks.

While we are excited about our Pillar 2 business, there is more to share in our Pillar 1 category where we power products such as cars, TVs and IoT devices. This has been a critical part of our business since launching our Voice AI platform in 2016. In particular, growth in the automotive industry requires long-term investment, global reach and is difficult to penetrate. Thanks to our track record and differentiated technology, we continue to expand into our large customer base of about 20 brands, including Hyundai and Stellantis as well as initiate production with new companies such as Togg, a transformational Turkish electric car manufacturer.

Overall, we are seeing expansion of usage in our categories, seeing more inquiries with the device of repower, which has increased our annual run rate to over $2.8 billion. This frequency of usage increases, the performance of our AI systems will continue to get even better and more intuitive as they learn from the new data and large amounts of existing data. Our extensive experience and data so far have already fine-tuned our machine learning tools to make our products highly valuable to businesses and more useful to users. We now predict faster adoption, new kinds of use cases and ultimately more revenue per device.

Our strategic channel partnerships are also generating momentum. For example, Qualcomm has integrated SoundHound Voice AI on its chipset with the Snapdragon Digital Chassis and is showcasing its on the Concept Vehicle. These types of channel partnerships are critical in developing growth in the automotive industry. I also wanted to specifically call out the traction you're seeing in TVs where we've increased the usage by over 80% compared to this time last year. Last quarter, we discussed the new revolutionary television manufacturer we signed. They entered the market with our technology in Q2 and you will hear more about them later this year. This should drive further usage of our technology in the space where we believe there is significant opportunity.

For IoTs, we have also either signed, extended or gone into production with multiple brands such as robots, printers, coffee machines and kiosks. And we are collaborating with a large home automation IoT company and one of the largest telcos in the U.S. As I mentioned earlier, with the launch of SoundHound Chat AI for automotive, we created a number of upsell opportunities and we are in conversations with the majority of our automotive customers as well as quickly gaining interest from new major brands.

We continue to see a surge in enthusiasm for AI, which has brought SoundHound significant interest from customers and AI followers. We believe the new AI disruption that is catalyzed by generative AI will have a larger impact in nearly all industries than the last mobile disruption. We believe SoundHound is uniquely positioned to achieve significant long-term success in the new AI era, given our deep voice and conversational AI expertise, robust patent portfolio, full technology stack and our years of collaborating with major brands within key industries, including automotive, smart devices and now customer service.

In line with our track record as a leading innovator in AI, SoundHound is training its own foundational model in generative AI. But our approach is different. Instead of yet another large language model, we are combining everything you've learned in nearly two decades to take it to the next level. Our foundational model will be multimodal, supporting audio and text inputs and audio and text output, which prompts based instant customization capabilities that used to take significant software engineering and machine learning to achieve.

We'll focus on features that we know will change the equation in the numerous applications we are building. We'll be able to use billions of real conversations and over a million hours of audio data in dozens of languages that we've so carefully accumulated over the years. The amount and highly relevant nature of this data give SoundHound a unique advantage. Our internal name for this foundational model is Polaris, matching our ambition to create the brightest star in the constellation of foundational models. We will have more to share about Polaris soon.

We have an amazing and passionate team that makes all this possible. I want to thank them for their tireless efforts to get us to where we are today and for playing a lead role in harnessing the technology at the heart of one of the biggest disruptions in history and making it practical and real. In closing, we are innovating faster than ever. We are growing. We are on our way to profitability and have significantly increased our cash position. We have the core voice and conversational AI technology built from the ground up, which gives us distinct competitive advantage to maintain our leadership position and further add to the AI disruption that will make the world a better place and create tremendous value.

With that, I'll now turn the call over to Nitesh to talk about our financial performance, key growth drivers and full year outlook in more detail.

[Nitesh Sharan, Chief Financial Officer]

Thank you, Keyvan, and good afternoon, everyone. We are pleased to report another strong quarter and solid first six months of the year with 48% year-over-year top line growth and a 75% gross margin, all while significantly reducing our operating loss. We have also vastly improved our cash position and closed the quarter with total cash of approximately $130 million. The accelerants and AI continue to expand our sales opportunities with existing customers and catalyze new customer interest. These new opportunities have only further validated our long-term vision to voice enable the world by bringing together billions of voice-enabled products and services and unleashing the flywheel of their interconnectivity and new monetization streams. This is the essence of the three-pillar business model we have described before; product royalties, service subscriptions and monetization.

This quarter, we ended with $339 million in cumulative bookings backlog, up 20% year-over-year. To reiterate, our bookings are derived from committed customer contracts and reflect expected revenue to be realized from those contracts over its life. These contracts can have minimum guarantees or include work completed, but amortized revenue not yet recognized due to ASC 606 rules. And in many case, they include customer estimates of volumes for auto or device shipments, for example, generally supported by historical benchmarks.

The cumulative bookings goes up as new deals are signed or if existing customers extend and goes down as revenue is recognized. As we have discussed previously, cumulative bookings backlog growth can be uneven given the scale of our enterprise customers relative to us. We've added customers across industries such as in telecommunications, printers and TVs to name a few. Including automotive, we have long-term contracts spanning up to 8.5 years with an average contract length of roughly 6.5 years. And as I've stated before, these are back-end weighted.

In Pillar 1, where we voice enable products, we continue to extend our offering across new units while adding more features. In autos, which represents the majority of this pillar, we increased our new units by 60% and our active cloud users increased by close to 70% versus prior year. Within Pillar 2, where we voice enable services, we are expanding our offering beyond restaurants to a much broader customer base with Smart Answering. Although, it's very early, because of the more streamlined onboarding, we are very encouraged with how quickly this can build. For restaurants, our Smart Ordering solution is beginning to solidify a strong subscription base. We have already closed with hundreds of brands, including the White Castle expansion Keyvan highlighted earlier. Aside from that, our current mix is predominantly SMB customers and we are expanding in regional and middle market segments.

Let me give you two quick examples of some of our traction. Naz's Halal has nearly 30 locations spread mainly over the East Coast and Mid-Atlantic region. A few months into onboarding two of their locations, our product surpassed performance expectations and they have now signed up to roll out across the entire remaining fleet in the next few months. Hot Table is selling delicious panini sandwiches in the New England region. And after piloting one location, just last week, we have activated four new locations in just one day and are on rapid pace to scale across all of their locations. These are tangible examples and when you envision the opportunity across the million food establishments in the U.S. alone, hopefully, you can understand our excitement.

Over time, we will provide more details of the key metrics we are watching here, but to give you a current sense of the near-term opportunities with solely the brands we have signed up to date at scale and full deployment across these customer groups, we would have over 1,500 locations and well over $10 million of ARR from this one specific business opportunity and our pipeline continues to strengthen.

Let me now get specific on our financial results for the second quarter. In Q2, we generated $8.8 million in revenue, up 42% year-over-year, driven by strong growth in our product royalties, primarily in automotive. The expansion is mainly driven by an increase in auto units overall as well as a slight increase in unit prices. And on a cumulative basis, we saw our IoT cloud units expand both sequentially and year-over-year. Additionally, a customer contract modification positively impacted product royalty revenue as the royalty increases from other international customers. These were partially offset by a decrease in customization services revenue versus the prior year Q2.

In June, we finalized an agreement to modify the scope of an existing hosting services contract without changing the contract price. This modification involved the addition of new professional services in lieu of a tail support obligation beyond the contractual period ending 12/31, 2023. Consequently, a portion of the professional services fee that was scheduled to be recognized after 2023 before the contract modification occurred of roughly $1.8 million will be fully recognized in 2023.

In Q2, our gross margin improved to 79%, up from 60% in the prior year quarter, which was largely driven by the expanding scale of our business and increased data center efficiency. Also, this quarter was -- there was a positive impact from the previously mentioned contract modification. Cost of revenue for the quarter was $1.8 million, a decrease of 26% year-over-year. This improvement is due to the continued efficiency improvements from on-premise activities to the cloud migration, helping us to drive further gross margin expansion as we scale the business. Additionally, in Q2 last year, our data center migration weighed on our gross margins.

Operating expenses saw another step function reduction in Q2 compared to the previous quarter as we had our first full quarter after executing our announced corporate restructuring program in Q1. R&D expenses were $11.7 million in Q2, a decrease of 38% year-over-year and 45% compared to Q4 last year. This decrease was largely due to the restructuring program we announced last quarter. However, our investment in key areas has not slowed and this level of R&D spend is where we need to be. We have the talent and expert engineering to support our growth while remaining at the forefront of innovation.

Sales and marketing expenses were $5.1 million in Q2, an increase of 16% year-over-year and a decrease of 25% compared to Q4 of last year. We saw a year-over-year increase since we went public in Q2 last year and we're still in the beginning stages of ramping up our sales and marketing efforts. We continue to streamline our focus in digital marketing, lead generation and customer acquisition, both direct and through channel partners.

G&A expenses were $6.4 million in Q2, a decrease of 32% year-over-year and a decrease of 12% compared to Q4 of last year. We experienced the majority of our going public costs in Q2 and Q3 last year and we saw this already start to taper off as we ended 2022. Across all operating expenses, noncash employee stock compensation was $5.6 million in Q2, excluding the minimal expenses associated with restructuring as we finalize that program. As a result, our Q2 operating loss was $16.4 million, an improvement of 43% year-over-year. Interest expenses, which include ongoing amortization impacts and quarterly interest costs were slightly lower than anticipated at $5.6 million in Q2. However, slightly higher than we expect going forward due to one-time transaction-related fees and other charges incurred as a result of our debt financing in Q2.

Net loss was $21.9 million in Q2, an improvement of 28% year-over-year. This led to a net loss per share in Q2 of $0.10 compared to $0.19 in the prior year. Adjusted EBITDA, which excludes noncash charges of stock compensation and depreciation and amortization as well as other nonoperating activities, including restructuring, was a loss of $9.9 million, which was a 50% year-over-year improvement. Net cash used in operating activities for the first six months was $33.7 million, improving roughly 28% year-over-year and a year-over-year improvement of 40% in the second quarter.

Now, on to our capital structure. Our cash position at quarter-end was approximately $130 million, up from $47 million at the end of Q1. In Q2, we raised $100 million of debt financing, which we partially used to pay off our existing debt of approximately $30 million and eliminate the associated principal amortization payment. Our previously announced committed equity line of credit gave us additional access to capital upon which we raised approximately $43 million in Q2.

In July, we filed a three-year shelf registration under Form S-3, which was not available to us previously. So, this is just standard capital management and good housekeeping. Part of normal ongoing capital programs that companies should have in place and we don't have any immediate needs to raise additional funding off of that shelf. That said, it does allow us to be opportunistic and agile if the right circumstances arise. Our capital position is now a source of strength and gives us the optionality we need to drive the business.

With that, let me discuss our outlook for the remainder of the year. The first half in aggregate has landed pretty much in line with our expectations despite massive organizational market shift. We are proud of this. As we look to the second half, there are a number of factors at play. There is tremendous interest in our solutions that are opening many new avenues that we did not fully foresee earlier this year, particularly in generative AI. That said, most of these are in large enterprise deals that can have longer deal cycles and can fluctuate.

The pace and vastness of these deals gives us greater conviction on the potential strength of momentum into 2024 and beyond, but we will comment on next year later in the year when we get closer to that time. Because of our unique market positioning, differentiated tech and massive addressable market, we continue to believe over the longer term, we have the foundations to deliver strong, sustained and meaningfully disruptive growth. In the near term, there continues to be a nuanced macro picture and volatility in the markets in which we participate with uncertainties in green shoots growing together, mainly attributable to the timing of our large enterprise opportunities.

Ultimately, though, we believe the longer-term tailwinds are largely in our favor. All that said, calibrating the puts and takes, we reiterate our 2023 revenue outlook of $43 million to $50 million. As we noted last quarter, our revenue builds throughout the year due to seasonality of Pillar 1 and the scaling of Pillar 2, and we continue to work towards becoming adjusted EBITDA positive in Q4 of this year.

Before closing, I'd like to note that last week, our auditor, Armanino LLP notified us that they have made a business decision to exit their public company SEC audit practice. As such, our Audit Committee has launched a process to find a new auditor. This is a business decision by Armanino that affects dozens of their clients and is not at all a reflection of our financial position or any previously audited financials.

To summarize, we are driving significant market differentiation, delivering on customer demand and catalyzing disruptive innovation with a high velocity. Each quarter on our journey brings its own unique dynamics, yet we are steadily piecing together the foundations for sustainable, long-term growth and profitability.

Thank you, and we will now move to Q&A.

[Operator]

Thank you. [Operator Instructions] And our first question will come from the line of Mike Latimore with Northland Capital Markets.

[Mike Latimore, Northland Capital Markets]

Congrats on the strong results here. In terms of the second half outlook, we should think about that as largely driven by license deals out of backlog? Is that still the key kind of second half driver?

[Nitesh Sharan, Chief Financial Officer]

Mike, I think it will come from Pillar 1 activity, a lot of backlog, but also some new deals, potentials that I mentioned are sort of in motion. So, we do continue to see great traction in the Pillar 2. As I articulated, there's -- we're building sort of our ARR pipe. But as I think you're alluding to, as those really end up being monthly recurring. So, those will build and be consistent over time. But as we keep adding more and more restaurants, that's going to be increasingly a 2024 and out story. So yes, I mean we have -- and I think I mentioned in the prepared remarks, there's been a seasonality that comes with our business, particularly with respect to autos and just the trending of vehicle shipments and so forth. So, it's a composition of a few different things, but I think that's kind of the mix.

[Mike Latimore, Northland Capital Markets]

And I might have missed it, but did you say whether OpEx should trend down a little bit from here? Or is this a good kind of run rate level?

[Nitesh Sharan, Chief Financial Officer]

It's -- yes, probably generally a good run rate level. What I'd say is we took most of those restructuring actions in Q1 itself. There was a little bit of restructuring expense in Q2 but largely complete. At this point, we feel like we're largely in the right place. We are being thoughtful about areas where things are moving around pretty quickly. So, where we need to be concerted about driving additional investments, we're not going to shy from doing that. Again, maintaining our commitment to getting to the right place on profitability. So, in a couple of pockets, we're making shifts, in a couple of places we're moving away from. So, I'd say, by and large, yes, it's appropriate to kind of think of it as this is a good level and it really kind of comes to scaling revenue.

[Mike Latimore, Northland Capital Markets]

And just last one. Great to see the White Castle expansion. Can you talk a little bit about the ROI they saw in initial launch? And then I guess it's going to roll out over 1.5 years. I'm assuming it was a good ROI. So, why wouldn't they roll out a little faster? I think you guys can kind of deploy faster than that, but...

[Keyvan Mohajer, Chief Executive Officer]

Yes, the 100 number is by the end of next year, but it's not -- it's actually happening already quite fast. So, we are in multiple locations and they have an aggressive target to add more and more. It's just that drive-thru requires some hardware upgrade and it's not just a swap out of software, so that takes a bit of time. On your earlier question, they had a key metric that they were looking for in terms of order completion rate. That is correct. And we by far exceeded the number that they were looking for and we surpassed their human order takers. So, they're very happy with the partnership.

[Nitesh Sharan, Chief Financial Officer]

And by the way, on return, Mike, there's obviously labor cost benefits of utilizing AI. There is the opportunity to have upsell, which could generate revenue. There's consistency of service. I think in our press release, we commented it's 24/7. So, coverage over for a place that does operate 24/7, that's meaningful. You don't get deterioration of performance because you're tired if you're AI as an example. So, multiple things from an ROI perspective, I think, are very attractive to these restaurants.

[Operator]

[Operator Instructions] And that will come from the line of Brett Knoblauch with Cantor Fitzgerald.

[Brett Knoblauch, Cantor Fitzgerald]

I guess the first one, I think we look at like other AI companies in the market who are kind of training models and they're spending a lot of money on CapEx in this build-out. And I guess when we look at kind of your guys CapEx, it's quite minimal. So, can you maybe remind us why maybe you don't need the capital intensity of some of the other AI businesses that are trying to acquire compute and whatnot?

[Keyvan Mohajer, Chief Executive Officer]

Well, we have been training models for over 10 years. So, some of that has accumulated. So, we have both AI models in dozens of languages. So, some of it is reflected in that. But yes, we have become very efficient in the way we train models and reducing cost. And also we have scale already with traffic that we have to serve with our real -- we process billions of queries and we are being smart about utilizing our capacity when the same hardware is not being used to serve the production traffic. So, we are trying to not have additional expenses to train large models beyond what we have already allocated going forward.

[Nitesh Sharan, Chief Financial Officer]

And Brett, as we've talked about before, we've migrated cloud providers. We utilize Oracle OCI and we have a great partnership there that gives us access to GPUs and all the kind of technologies that we need to be able to train our models. So, sort of in the business model already.

[Brett Knoblauch, Cantor Fitzgerald]

And then I guess now that you guys have like ample liquidity, what's going to be the investment strategy as it relates to building out your go-to-market strategy. This has historically been an area where you haven't invested much as you guys kind of been an R&D-focused or product-led company. So, I guess can you just maybe explain like what your vision is for what your go-to-market strategy would look like?

[Nitesh Sharan, Chief Financial Officer]

Yes, certainly. So, it depends on the pillar. On Pillar 1, we built a real great set of partnerships and customer base and with really just a handful of business development resources and that was kind of our history dating back many, many years, and it was on the foundation of great technology, differentiated technology in the marketplace, our capacity to work flexibly and with a lot of agility and scale. And so that was sort of, I think we'll continue to grow. We leverage channel partnerships. We mentioned Qualcomm example an opportunity to leverage a great partner, embed the voice AI with Snapdragon chip and get access. There's other partners.

When we talk to Pillar 2, it's a little bit of a mix between direct go-to-market motion as well as channel partners. So, we have our own sales team. We've been growing. They're very active in engaging with small, medium enterprises, but also regional mid-market and going to conferences, getting our Dynamic Interaction technology in front of people because as Keyvan mentioned in the prepared remarks, when we get that out there, people see it, they're wow-ed, they really want to -- they want to keep the conversations going.

But we supplement that and really complement that and get a lot of scale with channel partners. So, we've talked about how we can now integrate with Toast, Square, Oracle, Olo, we're adding others. And those are great because take Toast as an example. I believe they have 1,100 sales reps that we've been -- we trained the sales reps before to kind of showcase our technology, how we can integrate with them, and that just gives us greater reach. So that's sort of what we're doing on the go-to-market motion. I will say you're right, we are feeling great about where we are from the capital perspective and it's really around execution and going out and continuing to get scale and ways that we can just quickly get large enterprise franchises are interesting to us and we'll continue to look at both direct and indirect ways of accessing those customers.

[Brett Knoblauch, Cantor Fitzgerald]

And then for your full year guide, I guess, do you need any like one-time license or big one-time license deals for you to reach that midpoint? Or do you think that can come just from increased usage or inquiries from product royalties?

[Nitesh Sharan, Chief Financial Officer]

We do different -- in our licensing, there's sort of a royalty scaling. And then we do have one-time NREs from time we do various deals. We do professional services engagement. So that's all part of the contemplation. I don't know that I can break that up too much other than just scaling in units at the pace we're going, that gets you part of the way. But yes, I mean we need to do some other deals in that, but that's part of our overall motion as we work. If we have to do upfront custom development work, then we get that. And if we complete that within a certain period of time, we recognize it. If the performance obligation is longer, we amortize it. So, some of it is the rollout and some of it is we got to go do deals. We've got a lot of deals in motion.

And I think I mentioned, especially with the generative AI stuff, there's just a lot more interesting conversations going on. But those are large deals and the timing is not always perfect. So, we're going to keep hammering away at those and showcase to you guys in future quarters what we've been able to accomplish.

[Brett Knoblauch, Cantor Fitzgerald]

And then maybe if I could just ask one more. I guess you guys said the auto units were up 60% year-over-year. Pricing was also up year-over-year. So, I guess if I'm looking at product royalty revenue, which is $5.5 million last year, if I assume that grew 60% or just 60%, that gets me to a number that's more than the revenue delivered in the quarter. So I guess, what am I missing from that equation in terms of units and pricing that kind of gets me to like where you guys are at for this quarter's revenue?

[Nitesh Sharan, Chief Financial Officer]

Yes. When we -- what I conveyed in the prepared remarks was around units and unit activity, which generally on our revenue recognition. As we get a royalty report, we invoice off the royalty report, we recognize that revenue. Revenue is also make comprised of previously completed NRE work that might amortize over the life of the support agreement, could also include professional services.

So for example, on a year-over-year comp, that's included in the prior year number, last year, we had announced sort of we had one particular customer in Europe, an auto customer that we did some engineering work, and that was, I think, particularly related to like language deployments and some engineering work that we had to complete.

So, those aren't necessarily related to shipments. So, you have to sort of adjust for those things. That's why I try to give the double click of like what's actually going on the ground with units that are shipped because there can't be that type of engineering numbers that they're playing out and I don't want to say reeking. They're just changing the numbers. So that's a reflection in our prior year numbers.

[Operator]

Thank you. As I'm showing no further questions in the queue at this time, this concludes today's program. Thank you all for participating. You may now disconnect.