Serve Robotics Inc. (SERV) Q3 2026 Earnings Call Transcript

Operator: Hello, and thank you for standing by. My name is Temi and I'll be your conference operator today. At this time, I would like to welcome everyone to Serve Robotics, Inc. Third quarter 2026 Financial Results and Conference Call. [Operator Instructions]. I would now like to turn the conference over to Steve Webb, you may begin.

Steve Webb: Thank you, operator. Welcome to Serve Robotics Third Quarter 2026 Earnings Call. With me today are Serve's Co-Founder and CEO, Ali Kashani; and our CFO, Brian Read. During today's call, we may present both GAAP and non-GAAP financial measures. If needed, a reconciliation of GAAP to non-GAAP measures can be found in our earnings release filed earlier today. Certain statements in this call are forward-looking statements. You should not place undue reliance on forward-looking statements. Actual results may differ materially from these forward-looking statements and we do not undertake any obligation to update any forward-looking statements we make today, except as required by law. For more information about factors that may cause actual results to differ materially from forward-looking statements, please refer to the press release we issued today as well as the risks and uncertainties described in our most recent annual report on Form 10-K as supplemented by our most recent quarterly report on Form 10-Q and in other reports and filings made with the SEC. We published our quarterly financial press release and our updated corporate presentation to our Investor Relations website earlier today, and we ask you to review those documents if you haven't already. And with that, let me hand it over to Ali.

Ali Kashani: Thank you, Steve, and good afternoon, everyone. We delivered third quarter revenue of $2.8 million ahead of our expectations and up over 300% year-over-year. We are also raising our full year 2026 revenue guidance to $12 million, which is from the previous guidance of $9 million to $10 million. That is an almost 30% increase. We are really encouraged by how the team is executing. We've been focused on broadening our demand for robots and also improving our economics for the business. And this quarter, the results show that our efforts are paying off. Our goal is to use Physical AI to make local delivery more affordable for merchants and consumers and we are working towards that by making our network easier to access and also expanding into new markets and geographies while also lowering our cost of operating the robots and generating more revenue from each robot. So let's start with our newer platform partners, where we are seeing a lot of delivery growth from DoorDash and Grubhub. Daily deliveries with them has more than tripled from the end of Q2 to the end of Q3, and the share of all deliveries has more than doubled in the same quarter. Also, the number of merchants from the new platforms grew over 2.5x over the same period, and it continues to scale. These partnerships give us multiple paths to delivery growth. We're really focused on turning that broader merchant base into more work for our robots. The growth that we are seeing in the delivery side is actually underpinned by the continuous improvements we are making to our technology. Currently, our robots are 13% faster than they were in Q2. That is a 13% speed up in just a few months. The robots now travel 7% more miles per delivery, meaning that they can cover more business to perform a timely dropoff. Our reliability is another key to unlocking this growth. Our delivery completion rate was 99.9% in Q3 as it was in Q2. That means that of every 1,000 deliveries we pick up, 999 are delivered to the consumer successfully and no tip needed. Advances in our Level 4 autonomy are helping our robots move faster, and they're also reducing their staffing costs needed to support them. So stay tuned because we have some big announcements about our autonomy in the coming weeks. Now alongside these technical advancements, we've also been reducing our day-to-day operating cost for the fleet. We delivered results on a slate of cost reduction programs in the quarter that benefited our cost per delivery. And in aggregate, they saved us about $1 million in the quarter. Let me give you a couple of examples. First one is the improvements we made to our procurement and maintenance programs to drive significant efficiencies when it comes to robot repairs and replacement parts. The second is how we streamline the way we deploy robots every day, we made them faster and less expensive. And this cuts the external cost of deployment by 75%. Taken together, these efforts improved fleet margin by over 100 percentage points from Q2 to Q3. Now in order to build on our growth momentum, we are making our delivery network easier to access. Our relationship with Wonder and Grubhub has added another channel for demand in markets such as Washington, D.C. metro area. We are also working directly with merchants through Beacon, which is what we announced last quarter. Beacon allows us to integrate seamlessly with more merchants, and we are seeing great progress with the initiative. We recently launched our first beacon-enabled locations in San Antonio, Texas. The easier we make it for others to use our network, the more potential there is for them to bring their own ideas and their own customers. We know our fleet can support users and use cases beyond what we've developed ourselves. So we'll have more to share as some of these opportunities open up. Another important foundation for growth is our geographic reach to enable rapid growth, we have to have really robust playbooks for deploying to new markets. We have to have mature AI and autonomy models that can easily adapt to new cities and countries, and we have to have regulatory and policy support to launch new geos. So we have developed these robust playbooks now for expansion and policy and have matured our AI models after scaling to more than 20 cities in the U.S. So we are now taking these learnings into new international markets. and we are able to get operational in new cities with less friction. Last week, we announced our launch in Gold Coast Australia, a market that we are really excited about. Gold Coast is our first international market expansion and is in an exclusive partnership with DoorDash. Customers can order through the same DoorDash app that they already use, just like they do in the U.S. Clifford University hosts our local operations teams. This is really exciting. It's the first university to host our operations. Beyond Australia, we are also considering other international destinations. You may remember Vancouver, Canada as one of the examples where earlier this year, the City Council approved our pilot proposal. We're now working with the province for what could be our first Canadian city. As we continue to expand, we are learning how to reduce the cost and time it takes to get to new markets and get them up and running. And that actually brings me to an update about micro depots. Micro depots give us a small footprint local base for charging and staging and maintaining robots so that they can be very close to where they work. These are our most efficient deployment format. In our early deployments using this strategy, we are expecting to launch cities with 1/5 the cost and in a fraction of the time required compared to our legacy deployment formats. We are pretty excited about the results so far and continue to invest heavily into this. Lower costs are one part of the evolution of our fleet economics. Another way to make local automated logistics more affordable is by increasing robot monetization beyond delivery fees. You've heard me discuss this before, but this actually helps us diversify our revenues. And one really good example for this is branding. Branding is an another source of revenue from robots that are already working out there in the community. Our fleet now includes 2,000 robots in 40 cities and 27 hospitals. That presence gives both national and local brands a unique way to directly reach people. Our branding revenue grew nearly 80% from Q2 to Q3, and major campaigns are returning now. We have a strong pipeline of upcoming opportunities as well. For example, the underdog sports campaign ran for 6 weeks across Dallas, Atlanta, Miami and Los Angeles. 455 robots generated 144 million impressions, and they traveled nearly 100,000 miles. That's an amazing amount of progress in our sidewalk business in Q3. We are reaching more merchants, reducing the cost of operating, expanding our operational footprint and also generating revenue beyond just deliveries. All of this is putting us in a better position to make local delivery more affordable for merchants and for consumers. But beyond sidewalk, another way we use Physical AI to serve people is with our Moxi robots in hospitals. They move supplies and materials to help nurses focus more of their time on patients. Our health care sector continued to deliver solid revenue in Q3, in line with our expectations. And importantly, year-to-date, we have signed multiyear contract extensions with 12 existing hospital clients. Three of the hospital clients have actually expanded robots in their fleet. Overall, we are now serving 27 hospitals across the U.S., following the addition of 4 new hospitals in 2026 through the end of Q3. Hospitals and sidewalks are very different and unpredictable environment, but both of them forced us to solve these practical problems of robots operating among people in diverse and challenging environments. They help us train better models from a vast and diverse proprietary data sets to teach robots, how to coexist with people. Besides our years of work on onboard AI and autonomous fleet software have also generated data and operating experience that other companies are asking us to use inbound inquiries about our foundation model data are increasing, and we are beginning to serve our first customer. Our robust experience real-world edge cases and gather insights every day across a range of environments and scenarios that we believe makes us quite unique in the marketplace. It's early, but the cities, as a tipping point for building another path for the work our robot already do to generate value which is to help accelerate the Physical AI revolution by teaching foundation models about the real world. This interest we are seeing extends beyond just data. It also extends to our technologies that power are robots. We are building one of the industry's strongest Physical AI stacks, and we want our technology to reach more people beyond sidewalks and hospitals. I've always said delivery is just a starting point for us, but we are really building robots that can navigate human spaces. But we can't go and make every robust form factor ourselves. New robot forms are being developed across many industries, and we want to help them come to commercial scale faster. That's why we are not just stopping with 7 Moxy robots. Any company building a new kind of robots to be able to leverage our proven technology and instead focus their efforts on the unique problems that their robots is meant to sell. As Jeff Bezos says, they should focus on what makes their beer taste better. We see an opportunity to become a supplier to those companies, trying on what we've built for ourselves. So earlier today, we had a very exciting announcement about our Physical AI services. We think the Amazon AWS moment for robotics is here. Just like Amazon's technology, which was first developed to power its own retail website has become the foundation for millions of companies to build their web products. The technology that we have developed to power our sidewalk and hospital robots can help accelerate new robot manufacturers to bring their products to market faster and cheaper. At the start, our initial offering brings together three services: A high-performance data link that connects fleets to offboard intelligence in real time, a catalog of real-world foundation model data and a data curation infrastructure for the fleet scale AI learning. Our technology is already powering some of the largest robotic fleets in public spaces in the world. Now we get to bring the benefits of it to more people. So, our Serve Link, for example, is one of these three initial services. It helps connect robots in the field to the computing infrastructure beyond their onboard intelligence. It helps keep the operational data flowing reliably as conditions change so that robotic teams can really focus on advancing the autonomy, among our launch customers is Bedrock Robotics, which builds autonomous systems for construction. Bedrock use Serve Link for network connectivity across LTE, 5G and satellites. Another service we are announcing is the data curation product. It has helped us train our robots to navigate some of the wilder sources of edge cases imaginable in the streets of L.A. and now across the country while also providing us a safe and near perfect track record of completing deliveries once we receive the goods. It developed at times, acquired and have advanced these technologies along with many others to run our own sidewalks and hospital fleets. Physical AI services gives us a way to build a new business from that same investment we've made across a broader Physical AI market that is accelerating. So I covered a lot of ground today. To sum up, we delivered Q3 revenue ahead of expectations. We are raising our full year outlook from $9 million to $10 million to $12 million and we are advancing our technology, making our robots faster and more reliable than ever. We are reaching more merchants and expanding domestically and internationally while at the same time, also lowering the cost of operating our robots and improving gross margins. We will keep strengthening the core business while continuing to develop our diverse base of emerging business opportunities to fully take advantage of this historic moment that we are living through. With that, let me hand it to Brian.

Brian Read: Thank you, Ali. I'll walk through the quarter's results, the improvement in our cost base and assumptions supporting our updated FY '26 outlook. Total revenues during Q3 was $2.8 million, up from roughly $700,000 last year, an increase of over 300% year-over-year. Compared to prior quarter, Fleet Services revenue grew about $0.3 million even as Uber volume declined as we told you it would in August. Meanwhile, software services revenue declined following a completed customer contract during Q2. We are still in the early stages of development for our software services and anticipate quarter-to-quarter volatility. Recurring revenue growth continues to outpace nonrecurring revenue. And in Q3, recurring revenue accounted for 69% of total revenues supported by Serve Link connectivity growth of 23%. Branding revenue was significantly higher in the quarter, growing nearly 80% sequentially. A portion of that was a 6-week campaign that does not repeat. Gross loss for the quarter was $7.6 million and gross margin was negative 266%. This was our third consecutive quarter of margin improvement. Our fleet margin improved by over 100 percentage points sequentially as we implemented the fleet cost reduction programs that Ali mentioned. The actions drove the results. Autonomy improvements, which lowered remote operations cost per delivery, nearly $1 million of procurement, repair and maintenance savings and a deployment redesign that cut daily external deployment cost by 75%. Last quarter, I said we believe the path to gross margin positive is inevitable. Our focus on driving revenue per robot and operating hours, better operational productivity and a growing mix of recurring software and platform revenue should enable our gross margin improvement trend to continue. GAAP operating expenses were $53.6 million, an improvement of 7% from Q2. excluding stock-based compensation of $12.5 million and amortization of intangible assets and acquisition-related expenses of $1.9 million. Non-GAAP operating expenses were $39.2 million down $1.2 million from Q2, our first sequential decline this year. In August, I named a few areas of cost focus. Here's where those stand. First, discretionary spend and consultants. We terminated several consulting engagements and cut nonessential spend. That program removed nearly $8 million from the second half spending versus our prior plan. Second, deployment infrastructure. That shows up in the cost of revenues, not operating expense and we covered that a moment ago. Daily external deployment costs was down 75% and total cost of revenues were down 14%. Third, head count remained flat throughout Q3. We prioritized engineering hires for autonomy and Moxi 2.0. Other functions were flat or down in the quarter, which we anticipate to continue through Q4. R&D was $22.6 million or $16.7 million non-GAAP, up from $14.9 million. It is the 1 line item that increased and the area I said in August, we would continue investing in to support core autonomy to our next-generation platform for Physical AI services and Moxi 2.0 development and engineering compute resources. GAAP G&A expense was $20.2 million in Q3, which is down 19% from Q2. On a non-GAAP basis, G&A expense was $12.4 million. Operations expense was $7.3 million or $6.8 million non-GAAP, both down 8% quarter-over-quarter. Sales and marketing expense was $3.5 million or about $3.3 million non-GAAP, down 18% and 16%, respectively. Our investment philosophy remains anchored in ensuring that every dollar goes towards revenue quality, margin improvement and platform differentiation. Net loss for the quarter was $59.2 million or negative $0.58 per share. Non-GAAP net loss was $44.6 million or negative $0.51 per share. This came entirely from gross loss improvement and $1.2 million in operating expense discipline. Capital expenditures were approximately $5.3 million in the quarter before the benefit from an additional $1 million in tariff refunds received during the period. We ended the quarter with more than $198 million in cash and marketable securities. That is a real advantage and enables us to make the decisions we made this quarter from a position of strength. We are raising full year 2026 revenue guidance to $12 million, up from the previous guidance of $9 million to $10 million. This implies a fourth quarter revenue of approximately $3 million. In August, we said we expected to grow revenue by 3.5x and we are now on track for 4.5x our prior year revenue. We expect full year non-GAAP operating expenses of approximately $150 million which is $20 million below our expectations at the start of the year. We expect full year capital expenditures in line with our prior $15 million to $17 million. Stock-based compensation should remain approximately $50 million. we just discussed our Physical AI services offerings, each built on connectivity, tooling and data that we already use to power our fleet. That revenue sits in two existing lines today. Data contracts within fleet revenue and Serve Link within software services. Our Physical AI services require no new robot build so no new incremental capital expenditures. and it consumes no fleet supply hours. So it does not have to compete with delivery or health care for capacity. Let me close against August. Q3 performance was ahead of expectations from August with the commercial progress moving towards a stronger financial model, resulting in our increased 2026 outlook. I said margin improvement would come from cost discipline. Fleet margins improved over 100 points on a 14% reduction in cost of revenues. I said we would manage spending against the runway and keep investing in autonomy and software. Non-GAAP operating expenses declined for the first time this year with R&D investments remaining focused. And we ended the quarter with $198 million in liquidity. Serve is building a robotics platform, not a single-use delivery fleet. With that, operator, we can open the line for questions.

Operator: [Operator Instructions]. Our first question comes from the line of Colin Rusch with Oppenheimer.

Colin Rusch: Ali, can you talk about the breadth and depth of initial customer interest on the Physical AI services. Clearly, there's a need for some of these elements here, but just want to get a sense of how mature the sales process is.

Ali Kashani: Yes, absolutely, Colin. I can take that. We've been working on this for some time. We just are finally discussing it publicly. We have customers ranging from -- we mentioned Bedrock Robotics which is in construction, but also between the customers and pipeline do you have anything from mining to even defense, there is a very wide range, which is pretty much what we expect because, as I mentioned, People are building new form factors of robots, there's kind of the moment for Physical AI right now. And we saw enough of evidence and traction to decide to make this a new business line that we're pretty excited about.

Colin Rusch: Excellent. And then from a cash management perspective, as you think about optimizing the spend, how should we think about this as you get into 2027 and in other areas where you may be able to optimize and save a little bit of cash.

Ali Kashani: Yes. I guess I'll start and Brian can add to it. We've been very, I think, responsible with cash, as you can see, we've been making savings along the way every quarter. A part of this is as we are improving our technology you're getting better at operation. When robust gets faster when we build more demand, all of that actually helps with both the overall financials and the unit economics. So this is something we're working on an ongoing basis. And then, of course, when it comes to things like hiring, we've been very disciplined, and we're going to continue to look at that, kind of, spending the same rate especially the current transitions. Brian, if you have anything to add?

Brian Read: Yes. I mean I think, Colin, in our prepared remarks, we talked about quarter-over-quarter trends, how we're remaining disciplined all the way from the margin at the top to our OpEx throughout the P&L. So that mindset is not going to change right now. And in fact, we're going to keep paying more attention to it as we get into 2027.

Operator: Next question comes from the line of Taylor Manley with Guggenheim.

William Taylor Manley: Congrats on the strong results. First, on Australia, obviously, that's in connection with the University. How should we think about the opportunity as it relates to college campuses? And is it kind of -- is the TAM too small in the U.S. for that to be worthwhile?

Ali Kashani: Yes, good question, Taylor. We are you've always been focused on urban density. So college hasn't been something we've just looked at exclusively as a market. And I don't think that's changing right now. This just was the right partnership in the right environment, but it's not really changing our strategy about the college campuses.

William Taylor Manley: Helpful. And then second, maybe one for Brian on diligence. It seems like some strong traction there. Any thoughts on kind of like top line growth trajectory once you lap kind of the initial acquisition contribution.

Brian Read: Taylor, thanks for listening in. So we will reiterate what we said 2 quarters ago and last quarter with respect to $7 million on Diligent, really strong performance throughout the year. We talked about the new customer acquisitions even just this quarter and the hospital is now up to 27. So we're looking at ways to enable that growth into 2027, which is supported by the CapEx numbers and the investments that we've talked about the last couple of quarters.

Operator: The next question comes from the line of Alex Latimore with Northland Capital Markets.

Mike Latimore: This is an exciting news here with the Physical AI services launch. I had one about daily active robots. Is there a way to think about how many you might have exited in the year you added over 100 this quarter, so maybe 1,000 plus?

Ali Kashani: Alex, yes, I wasn't anticipating this. So I don't have a number to share with you right now. I just want to just say is we talked about being cost conscious and disciplined. So we are reacting to what we see when we have demand, we add robots, when we have opportunities like Australia and there are some other cities and countries we're looking at, we would expand. But we are not going to force this doing irrational things, and we want to be very cost cautious. So that's why the number is just growing organically as it makes sense.

Mike Latimore: Understood. And then on the Physical AI services, how do you price those services? Are they usage-based or on a recurring fee per system? And then what are the gross margins?

Ali Kashani: Yes. So things like Serve Link, it's been recurring revenue per usually device perceives that's the model. And really good gross margin. This is software, what you would expect with software. It's helped our overall gross margin as well. And then, of course, this is just an initial set of products, and we expect that we would have more pieces of our technology that we can offer. And there could be different pricing models as this matures, but a lot of it right now is the kind of software service model.

Mike Latimore: Cool. One final question. I saw that DoorDash recently acquired Instacart. Does this open up the door for Serve to do grocery deliveries over time?

Ali Kashani: We've done groceries in the past with -- directly with merchants in Los Angeles, for example. And yes, it's definitely a market we've been paying attention to. And as you know, we have relationship with DoorDash. So I believe this is going to take them about a year or 2 to close, if I recall the process. So it might be a while, but we'll be very excited to speak to them and see if the opportunities there.

Operator: Next question comes from the line of Jeffrey Cohen with Ladenburg.

Jeffrey Cohen: Could you talk a bit about Beacon and some early experience as far as the rollouts there as far as San Antonio? And maybe talk about some of the providers and what you're finding as far as optimization and how we may see that expand?

Ali Kashani: Yes. Yes, happy to take that, Jeff. So we have our first merchants now using Beacon. This is just for folks who may not remember, we talked about it more advance in the last earnings. It's a small tablet that enables merchants to get notifications and a robot is doing their deliveries. It's off the pretty big bottlenecks when we are trying to integrate with more restaurants, especially when you -- when they using POS systems, and we want to try to have a direct relationship with them. So it's still early days for that, but it's pretty important missing piece of the puzzle that we've now brought to the table. So I expect we would see more adoption over time.

Jeffrey Cohen: Okay. And then secondly, as a follow-up, could you talk about Moxi a little bit and the 2.0 development and maybe it gives us a flavor of commercial and sales channels and existing in new hospitals as far as POs and your sales organization.

Ali Kashani: Yes. So the 2026 was really an investment year for Moxi. So we did the acquisition at the beginning of the year, we invested and help build a new Moxi 2.0 hardware. With that now rolling out, we are hoping to kind of switch into both getting the current fleet upgraded, but also getting them into new hospitals. And there are definitely conversations with partners and distributors about what 2027 and 2028 is going to look like. So I'm hoping to have more to say about that in the coming months.

Operator: Next question comes from the line of Andres Sheppard with Cantor Fitzgerald.

Unknown Analyst: This is Peter on for Andres. Congrats on the quarter. So first, we were hoping you could spend a little bit more on the Uber relationship. Do you think there's maybe an opportunity to resume this partnership in the future?

Ali Kashani: Peter. I don't have much new here. The relationship continues right now. We still are doing our contracts, which goes, I believe, through part of Q1 next year. And I don't have anything that has changed. We'll kind of continue to evaluate that as time goes by. But I guess just worth mentioning where we are seeing a lot of growth and momentum that we Grubhub are daily delivery growth. Something about 60% to 70% just in the last month. And I mentioned DoorDash as well, something like tripled in daily volume. So it's a really good growth with our newer partners and then we'll see how things go to be, but I think it's been fairly transparent about what you're thinking as of last quarter.

Unknown Analyst: Got it. Very helpful. And then maybe you had -- when we're heading into 2027, can you talk a little bit more about some of the operational goals, whether you be targeting more increased utilization of your existing fleet or looking to scale up the number of your bots deployed?

Ali Kashani: Yes. So I think we are always pushing three things at the same time here. So first one is just improving the technology. So I mentioned that we have increased the speed of robots just since Q2 by something like 13%, which is really phenomenal progress there. Also, there's going to be some big announcement we have coming out about autonomy. So we're going to keep pushing the technology, and that has downstream impacts on the operation as well. On the scale, as we are building demand, especially with new partners, I mentioned how we have more than doubled the number of merchants with Grubhub and merchant in the last quarter. So we are going to see opportunities perhaps for more robots or new neighborhoods, new cities. And then the third pillar here is always cost down. So just getting better at operation and driving the cost of operations down. And you can see again in this quarter, we made good progress there too.

Operator: And that's all the questions we have. Ladies and gentlemen, this concludes today's call. Thank you all for joining, and you may now disconnect.