I think we're going to go ahead and get started. Good morning. I'm Rob Mason, the Senior Analyst at Baird that covers advanced industrial technology. Very happy to have Impinj with us today. Cary Baker, CFO, here with us. Just as a quick background, as you're probably aware, but Impinj is a leading developer enabler of RAIN RFID that's used to connect billions of items and provides item-level visibility, delivers real-time information to businesses about the items that they create, manage, transport, and sell. We're just going to kind of dive right into Q&A to get started. If you have any questions yourself, send those up to the iPad. We'll work those in, but also just feel free to raise your hand. Cary, thanks for being here.
Yeah, thank you for having us.
We're obviously, like most companies, not too far removed from third-quarter results, which maybe you just want to touch on the state of the business, how things are looking in 4Q, and how you see things wrapping up the year.
Yeah, sure. We delivered a strong Q3. Revenue was strong. We had record endpoint IC volumes. We had an exceptionally strong systems business, up 30% quarter-over-quarter. We delivered 19.8% in operating margin, which on a product revenue basis, excluding the periods in which we received license revenue, that was a record for Impinj. As we look into Q4, we're seeing incremental leverage at the gross margin line. We anticipate the M800 being our volume runner in the fourth quarter. As a result, we expect gross margin to be up over 100 basis points sequentially. Strong all the way around.
Yep. So that kind of endpoint IC strength as well, we kind of watch that on a seasonal basis also. Seems to be getting back to a normal seasonal pattern. Could you just maybe talk about what kind of the influences are fourth quarter, maybe into the first quarter as well, if you have any kind of visibility there around the seasonal trend.
Yeah. Our largest vertical is retail apparel. Upwards of 60% of the ICs we ship go into retail apparel. Our second largest vertical is logistics. Both of those categories have the same seasonal properties in that we ship in front of the holiday season. Q3 is our peak, and then Q4 steps down, typically down 5%-10% on a sequential basis in the fourth quarter for endpoint IC revenue. We actually guided on the more favorable side of normal seasonality for this quarter. As we look into Q1 from an endpoint IC perspective, and though we only guide one quarter at a time, the typical dynamics you see are endpoint IC volumes rising, but we also negotiate ASPs, our annual ASPs in the fourth quarter that go into effect on January 1. In a normal environment, those are down low to mid-single digits. You have those two offsetting factors that go into play when you look at the first quarter.
S afe to say we're in a normal environment at this point?
Right now, it's been performing relatively normal throughout the year.
Okay. There was reference as well, I think, some of the larger projects, just seeing some phasing there in the second half of the year, probably not atypical when you have some larger projects. Is that because of the overall macro or those kind of project-specific customer decisions, or just what's kind of influencing the flow on projects? We've kind of heard some of that elsewhere as well.
Yeah, it's really nothing more than customer timing. The second half has been a particularly strong period for systems for us, especially when compared to the first half. We knew going into Q3 that we were going to have a strong Q3 from a systems based on projects kicking off at current existing customers. That proved to be the case. It was actually stronger than we anticipated. As I mentioned earlier, we were up 30% quarter-over-quarter in the third quarter for systems revenue. Normally in the fourth quarter, we're also up in the systems business because of typical enterprise hardware buying behaviors. You're at the end of your fiscal year, you've got capital budgets to spend, and we benefit from that.
Ninety days ago, we thought we'd actually be up slightly in the fourth quarter, not as strong as usual, but because of the strong Q3, only up kind of on the lower side of normal seasonality, which is + 10%-20%. What we saw, though, was we saw those projects accelerate into Q3 and then stretch a little bit into 2026. The result is we're going to be slightly down on systems revenue in the fourth quarter. Those projects, the size and scope are unchanged. The customer's ambitions are unchanged. It's just sometimes the enterprises change the timing or they reprioritize one project over the other, and that impacted us slightly. From a systems perspective in the second half, it's still an exceptionally strong performance for us, and we're very pleased with the systems business.
Now, in the third quarter, have you talked to what areas you saw that kind of extra strength, whether it was in the apparel, logistics?
It was in logistics, yes.
What does that say about how do you read that as an impact on endpoint IC volumes in the future?
Yeah, I think anytime you have large systems deployment, it is a benefit to endpoint ICs down the road. These were actually the projects that we mentioned in Q3 that were kicking off were with two existing customers, our visionary European retailer as they're expanding their deployment beyond inventory visibility, beyond self-checkout and loss prevention, and then at our second-largest logistics customer as they too continue expanding their deployments.
Yeah. The food market opportunity is really elevated over the last 12+ months, I guess. There was an announcement in the quarter as well with Walmart selecting some categories they're going to apply RFID to, which followed Kroger. You just kind of touch on that's the two that we know of right now. You kind of touch on what that opportunity entails, how you see that being rolled out if you have that kind of visibility on a timeline standpoint, and how does it manifest for Impinj? Obviously, endpoint ICs would be where you would most think it would show up, but just initially, any kind of impact on the systems business.
Yeah. There have been two grocery deployments announced. Kroger announced a bakery deployment a little over a year ago at this point. This information is a couple quarters old, but the most recent we have is they were up to 700 stores deployed. They have roughly 3,000 stores, maybe a little bit less than 3,000 stores, so ramping nicely since they deployed. More recently, Walmart announced a deployment in their stores for bakery, proteins, and deli. The commonality amongst those two announcements is they're starting in the portion of the grocery store where they control the supply chain. They're packaging, labeling, producing, in many cases, the items behind their counter. It's an easy decision to make. They don't have to go negotiate with the consumer packaged good firms to put a tag on an item.
The goal in these cases or the objective for these use cases is very similar to retail apparel. In retail apparel, they're looking for freshness of fashion, as an example. In food, they're looking for freshness of the items. When deployed, they'll see benefits in reducing waste, better informed production, lower cost. If they follow the same path as apparel, they'll also see a sales lift. They see the sales lift because they know what's on the shelf and, more importantly, what's not on the shelf. That helps inform sales.
Yeah. What kind of numbers have you been able to put around the market size of this category? Think of it in different, when we say food, that's food broadly, but it's obviously different categories. There's probably different value drivers between those categories as well that impact that. Just what kind of color can you give us on that front?
Yeah, food is going to be a massive opportunity. If I compare it to the other three verticals that are more penetrated at this point, apparel is 80 billion units a year. General merchandise is 325 billion units per year. Logistics is 400 billion units per year. Food is bigger than all three of those combined. We're very early days right now with just two programs announced, but I would say that the activity in the pipeline is very strong. While Kroger and Walmart are ahead, there are others following behind them. Like apparel, which benefited from shared learnings, once we knew as an industry, we knew how to tag a pair of jeans, every apparel manufacturer knew how to apply a tag to the pair of jeans. Food's going to be the same way.
Once we know how to tag a steak, everybody will know how to tag a steak. Once we know how to tag a bag of bagels, everybody will know how to tag a bag of bagels.
How should we think about kind of the logical progression among the various categories as these roll out?
Yeah. So while.
How much time may be between those?
Yeah. That's an important question. First, of the three categories that are announced, it's not all SKUs within those categories. There's still work to do to solve some of the SKUs in each bakery, proteins, and deli. As we think about a deployment, while the use case is the same as apparel, there's still a lot of learnings that have to happen. That just takes time. You think about a store like Walmart, thousands of stores that need to be outfitted with encoding, endpoint IC encoding infrastructure on their existing weight scales. Not a huge lift, but work that needs to happen. A company like Walmart, which has experience in RFID in their apparel department and in their general merchandise departments, this is new for the food department. There are tens of thousands of employees that need to be trained on this.
We see that play out in terms of the pace of a deployment. Any large deployment follows somewhat of a similar path. Going from 0% to 20% deployed takes time. It is slow because you're training the employees, you're doing the infrastructure lift, you're gaining the momentum of having a foundation. Going from 20% to 80% goes really fast. You have the foundation, you know what you're doing. Now it's just rinse, wash, repeat. The final 20%, going from 80% to 100%, takes time again because you land at your edge cases, hard-to-tag items. As an example in food that we're dealing with right now is how do we tag a chicken, a chicken breast, poultry? Poultry has high water content. Water content interferes with RF propagation. We've solved it.
We, the industry, have solved it for steaks by putting a spacer underneath the tag that creates enough air gap for the RF to propagate. A steak has far less water content than does a piece of chicken. It is something that we will have to solve as an ecosystem. It is entirely solvable. It will just take more time. As I mentioned before, once it is solved, then everybody in the industry benefits from it.
Yeah. Maybe think about you're a technology leader in this space, innovator for sure. If you solve that, obviously it helps the industry, but is there any way that you can maintain a disproportionate amount of the economics from solving those types of problems with your system?
We think of the economics as share of the endpoint IC opportunity. We think the M800 is exceptionally well positioned, especially the M800 with Gen2X, to solve food, which is a hard-to-read category because of the product density, because of the properties of food, like the water content in chicken. We think the performance of the M800, the increased readability, will play very well in this opportunity. That is how we will drive economics to Impinj.
We've been waiting on the M800 to gain more volume. Like you said, it's creeping up on volume runner status. Maybe got there third quarter or will in the fourth quarter.
Will in the fourth quarter.
Fourth quarter. There is also Gen2X, which is another element of the technology stack. Could you expand on that and what that enables as well, whether it is in conjunction with the M800 or elsewhere?
Gen2X are a group of features that are extension of the existing radio protocol. It's based on learnings that we've had over the years working with enterprises. It's just incremental features that we've put into our ICs. It's not just the IC, it's also functionality in the reader to engage the IC. Here's an example. In a normal Gen2 environment, which is the global radio protocol that the industry subscribes to, which at its core level means that every reader can read every IC irrespective of manufacturer. It's how to transfer bits over the air. In a normal Gen2 environment, the tags constantly respond to the reader. If you think about a retail environment where there are thousands, maybe tens of thousands of items in a reading footprint, those items are constantly responding to the reader. Slows the reader down. It creates inefficiencies.
With Gen2X, we've put functionality in the reader to engage the IC, to tell the IC that once it's read to stop responding so the reader can focus on reading the rest of the tags in the environment. That increases read speed, that increases read range, and it drives a far more efficient solution to our end customers. That's a functionality that is proprietary to Impinj, and it's leveraging our competitive advantage, which is being the only player in the market that operates on both ends of the radio link. Now, we have Gen2X native in all of our M800s. It is also in all of our R700 readers and all of our E-family reader ICs. We've also licensed the Gen2X reading capabilities to some of our largest partners like Zebra, as an example. They now have Gen2X capabilities in their reading footprint.
It becomes more ubiquitous in the readers, but you retain the proprietary aspect of that by locking in the M800 as the endpoint IC.
That's exactly right. Our prize is the endpoint IC share. We think by developing the Gen2X features, we are not only helping end customers solve problems that were previously unsolvable, but we're also positioning Impinj to benefit from an endpoint IC share perspective.
For established programs, though, how common is it for the customers or the inlay providers to switch their chip for an established—I mean, will they do that, migrate to an M800, or do you need to wait for the next iteration of that program?
We'll see both. We'll see people migrate. We'll see multiple chips qualified in an opportunity. Where we see the sharpest movement is when we go to end customer and solve a problem that they haven't been able to solve before. As an example, we have a customer that has always wanted to do overhead reading, but the math never penciled. The read range was too small of the readers and therefore required too many readers. The infrastructure cost was too high. With Gen2X and the 40-ish percent improvement in read range, or increase in read range, the math now pencils. This was an account that was not a 100% Impinj share account. It was far lower than that. We're solving a problem for this customer that only gets solved with the M800 and Gen2X at play. It's an opportunity, again, to drive share to Impinj.
Maybe I'll pause right here to see if there's any questions in the audience. If not, we'll keep going. Maybe just to stick on the M800 a second, though, we've talked about, or you've talked about the gross margin implications for that chip. Is that what we're starting to see show up in your gross margin guidance to step up? How are you trying to balance that, manage that? Because it does give you some flexibility as well, maybe to go, I don't know, to perhaps go after market share. Or should we think it's all going to accrue to gross margin?
It's both. It's going to accrue to gross margin. It's going to accrue to market share. Because what the M800 has done, being a smaller die, is it spreads the largest cost, which is our wafer, around. It reduces the cost of the endpoint IC. And we share that with our partner ecosystem by giving the M800 at a lower price point than the M700. And we benefit the bottom line because of the lower cost. When the M800 is fully ramped, we expect 300 basis points of gross margin accretion. We're starting to see the first of that in the fourth quarter, as I mentioned, signaling that gross margin is going to be up over 100 basis points sequentially in the fourth quarter. We are not reaching the terminal mix of the M800 in the fourth quarter. I don't think that happens until 2026 sometime.
It won't blend for all of 2026. We'll see kind of a ramp as we move throughout the year.
Very good. Maybe just to zoom back out to kind of the major markets that you play in, retail, apparel, of course, as you noted, the most penetrated today. Just kind of update us there on what the customers are working on, where the major projects are that you're seeing, what's maybe next in line behind or in front of the projects as some of those start to wrap up or move to the next phases even.
Retail apparel, as I mentioned, largest vertical that we have, $80 billion units per year opportunity. We think by volume, we're roughly 40% penetrated. By brand, we're probably north of 90% penetrated. Most retailers have made the decision to move forward. Most retailers are not yet 100% deployed. They're following kind of a pragmatic approach to deployment. They're going geo by geo or brand by brand or product line by product line. They're just kind of increasing their penetration every year based on that progress. The goal is to get to 100% because then you can unlock more use cases. Up to 100%, they're using the only use case they're primarily driving is inventory visibility using a handheld reader. That's making sure you know what's on the shelf, what's not on the shelf, and moving it from the back of the store to the front of store.
When you get to 100%, you can extract even more ROI out of that tag. You can now do self-checkout. You can now do loss prevention. You can do smart dressing rooms. You can do better front store, back store management. Those are the use cases that are happening now that we expect to continue going as our retailers reach 100% tagging. In the logistics space, we see the same type of approach. Strive to get to 100% tagging and then add on, or sometimes in parallel to getting to 100%, add on additional use cases. With our second-large logistics provider, we saw them start with conveyors, putting our R700 reader on a conveyor, in some cases in place of the six-camera array solutions to track packages more efficiently moving down the conveyors.
The challenge that logistics has always faced is they could have throughput or accuracy, but never both, which is why you always see in the fourth quarter temporary labor staff up to support the increased volume. The RAIN reading capability solves that. They can have both, both throughput and accuracy. That's the initial use case. They started moving to preloaders to make sure what was getting loaded on the truck was more accurate so they can reduce the cost of misloading. They started moving to the trucks themselves. They'll move beyond that. They'll expand the RAIN reading infrastructure to recognize or capture all the potential value.
I think you noted, just in terms of a tagging perspective, you're nearing or at full penetration domestically at this customer. What does the international prospects look like?
Yeah. We reached 100% tagging on a domestic basis in Q3, not in the first half of the year. We achieved it in Q3, which means there will still be domestic growth on a year-over-year basis next year. The international opportunity has not started in earnest yet. That is about 20% the size of the domestic opportunity, but still an avenue for growth for us within that customer. The logistics pipeline is very active. Everybody, I believe, in the market realizes how far behind they are. They are trying to play catch-up right now. A logistics deployment, as I just walked through, on conveyors, then on pre-loaders, then on trucks, is much more complex than an apparel initial deployment where you are using just a handheld reader, which is to say it just takes some time.
General merchandise is another category that's started to emerge and obviously is a large market opportunity. Where do we stand there in terms of penetration? There's been Walmart has been pushing that. I think they're in phase two. There may be another phase beyond that. Just to be clear, I don't think that intermingles with the grocery opportunity. Just update us on where that stands and any others that may be on the heels of any other vendors, stores, retailers that may be on the heels of that or what you've seen.
Yeah. Walmart has announced two phases of general merchandise rollout, including categories like home goods, electronics, toys, stationery, some lawn and garden. Not 100% penetration in any of those categories, but announced a dozen or so different categories. Admittedly, we're behind on the project timing for that. There's still expansion left within those two phases that have been announced. It has been impacted by tariffs where apparel was able to move outside of China relatively easier when the tariffs hit. General merchandise was not as able to move outside of China. So we felt the impact from that. The whole ecosystem felt the impact from that. There is additional opportunity within Walmart. We expect a phase three, the category timing TBD, but we still expect that. And then there are other opportunities within the general merchandise space.
We call them piggyback deployments, folks that are leveraging the work that Walmart has done to start deploying a category. I don't have anything public to point you to on that just yet.
What have been some of the challenges, I guess, others might deal with? I don't know. They may like the scale of a Walmart, smaller retailers. Will they have any kind of distinct challenges as they try to follow Walmart's lead?
Anytime you're tagging an item or a category the first time, there's always challenges with doing it. There's learning that's involved in the phases that Walmart announced. We had to figure out how to tag stationery for the first time, how to tag a board game the first time. Entirely solvable, but just takes time. If someone follows Walmart's path and says, "I want to tag stationery," that highway has already been built by Walmart now. Their ease to deployment is significantly improved. If they want to do a new category and say, "I want to tag cosmetics," which hasn't been tagged yet, that will take some learning to do.
Maybe the inference there, I mean, Walmart, given their size and scale, they can exert a lot of leverage on everybody they deal with. Not every retailer has that same influence. I'm just curious, do you see any potential friction points as others try to adopt that Walmart didn't and had to deal with?
Potentially, yeah.
Okay. Maybe just real quickly here, Cary, I think we want to, AI has obviously been topical everywhere. I'm kind of moving down the P&L. I want to touch on how you're managing the business. I'll pose the question to you, just if there's been any kind of areas that you would highlight where Impinj is leveraging AI internally. You've been able to keep your OpEx actually pretty tightly managed. Is that having an influence there?
The OpEx has been managed, yes, but not as a result of internal use of AI. We're still exploring where we might be able to deploy AI internally. The AI resources that we have right now are focused on how do we put it into the product. Look for us to put machine learning at the edge of our reading environment in ways that help improve the ease of deployment in a RAIN solution. If you think back to Wi-Fi years ago, you practically needed a PhD to log on to a Wi-Fi network. God forbid if you had a Mac, you just had no chance at doing it. Today, it's easy. You just click the button and the Wi-Fi works. We are trying to make RAIN work like that.
We're trying to take the RF out of the RFID to make the deployments ubiquitous and easy to stand up.
Okay. You previewed some growth investments that will start maybe in the fourth quarter. Is that focused on this effort? Other parts of the technology stack that you're investing in that you want to call out and how we should be thinking about where you're headed with it?
Yeah. So what we called out was an investment in software. We think there's an opportunity to layer cloud or SaaS-like solutions on top of our platform to make deployments easier, to better manage devices. It's nascent today. We see the opportunity. We have strong ambitions. This is one of the areas that we're going to invest for next year is building out those cloud solutions. The other areas that we will invest next year really relate around how do we stand up and support more enterprise customers at once. Today, we can handle one to two lighthouse or large enterprise accounts at once, but the demand is much greater than that. Look for us to invest in the sales and the engineering type resources that we can deploy on site to customers to help bring solutions alive.
Maybe update us as well just on the partner network because that's critical to driving adoption outside of the lighthouse accounts. What can you do? What are you doing to help remove friction points around deployments? Continue to hear about how RFID has always been complex to deploy. We start to build up a base of learnings there that can be leveraged, I would think. What can Impinj do on that front as well?
Yeah. Part of it is making our products more easy to use. Making the readers more simple to use, the reader ICs more simple to implement, the endpoint ICs more simple to deploy. We're absolutely working on that. When we are working with a large enterprise customer, we're never able to be the ones that go deploy it for that customer. Oftentimes, they'll match us with one of their SIs to train the SI up on how to deploy. Our goal with that is to train them so that they can take the product to market in other opportunities. Those are just a couple of ways of how we address that.
Yep. Again, I can't see it. We're almost out of time. Any last questions from the audience?
I'm not wearing this one. So I know NXP is a competitor and you have a handful. What do customers choose you based on? Are you low cost? Are you differentiated from a product perspective? How do people pick?
The question is, why do customers choose us over our competitors? We are the innovators in the market. Our IC is the most performant, the smallest, the most cost-effective IC to implement. That's one of the reasons they would buy us. We are also the only ones with a platform. We can design and tune a solution to solve a use case that is not easily replicated by a mix-and-match solution. Where our IC and a competitor reader or our reader and a competitor IC, we try to solve and drive preference via innovation.
Very good. We need to break there.