Extreme Networks, Inc. ($EXTR)

Okay. Good morning, everyone. My name is Tomer Zilberman. I cover 2 areas here at the bank. I lead coverage of vertical and back-office application software, which not so relevant for this discussion, but I also support coverage on networking and by extension, I cover Extreme Networks. Today, I'm joined by Kevin Rhodes, CFO of Extreme. Kevin, thank you for joining us.

Yes. Thanks for having us.

Of course. Kevin, maybe just a high-level question just to start here for investors that are newer to the story, can you talk a little bit about Extreme, where do you sit in the campus network infrastructure? What's your opportunity in campus switching versus WiFi equipment?

Sure. So we deliver, obviously, WiFi solutions, both hardware and software solutions that when kind of combined together given a very elegant networking experience for our customers, we believe that software is the real key differentiator. And what we've been really focusing on in the last 2 years is, a: we have this fabric technology, which is different than the competitor set. Typically, competitors have an IP fabric, which is more like a 7-layer cake. And ours is -- the way that our architecture is built is more like a spider web and it doesn't have IP addresses. And so at the end of the day, what -- you can't move laterally within our network, which gives you a lot more secure architecture for how you configure your network. The hardware itself, access point, switches. Those are pretty standard in the industry. We use Broadcom. Broadcom is an amazing partner of ours, and that's who we use. And we feel very, very confident in working with Broadcom in that regard. And we've kind of solidified. We talked about recently, our supply chain for like next fiscal '27. We are kind of really focusing on is gaining share, gaining share against Cisco, against HP, against Juniper. Naturally, there are some forces that are happening in the markets right now that are getting us some opportunity to kind of win more share than the normal competitor. I do describe this as an ankle biter, but we're working our way up to the knee at the end of the day because it's a large, large market that we can go after at the end of the day, we feel like we're really winning head-to-head on technology, but we're also doing it with software.

Right. Absolutely, I want to dive into the technology and software aspect. And you are seeing, if you look at the market share data that you are gaining share, right? I think you were kind of 3% market share last year, 3.5%. I think that momentum is continuing. But before we get into that, I want to talk about the Broadcom piece, right? I think the topic that [indiscernible] nowadays, which is memory supply shortages and your management of that. So as I said, you guys have been executing very well. Stock wasn't really working up until recently until kind of last quarter-ish when you came out and you said we have line of sight to visibility of memory supply until 2027. Can you first walk us through how you got there? What are the steps that you have in place to ensuring that you really have all the supply?

Yes. It's really 2 things that happened last quarter, right? One, we said about the supply chain, and two, we actually had pretty good margins because people are worried about margin pressure, which we've actually kind of solidified through a couple of price increases. And I'm sure we'll cover that. But on the supply chain side, we -- first of all, we have an excellent team on the supply chain side, and they've been with us for 10, 12 years. And they've -- COVID was [indiscernible] of 10x the problem that we faced here. The reality is working with our partners, in particular, Broadcom and others, we were able to solidify all the components within the next fiscal year. And that's really Q4 through fiscal year 2026 -- 2027. And we feel very confident, and that's what we said in our third quarter coming out is that our supply chain is covered. And it's not just covered from a contractual commitment perspective, but we've actually bought a lot of the components ourselves as well, and they're on consignment with our ODMs, but those ODMs is really kind of sitting either on our balance sheet or sitting on the ODMs balance sheet. But at the end of the day, these are components that we've now secured. And we feel very, very good about our ability to achieve and deliver on our promise that we can fulfill 2027 forecast.

So as you're just kind of referring to, right, you have a mix of inventory on hand, but also supply that's committed that maybe you don't have on the balance sheet just yet. We just hosted Arista this morning and in their last quarter, they talked about decommitments. So what's the risk that your partnership with Broadcom or I think you've also talked about having direct relationship with some of the memory vendors that they come to you and they say, "We know you committed, but we can't just supply it?"

Yes. I mean on the Broadcom side, number one, we sit at the lower level than like Arista, right? Arista is at the kind of 5-nanometer and a different type of chip that's coming through Broadcom. We're at the lower level, the [indiscernible] Trident level. And we feel good about our ability to -- I mean, we've committed -- we've heard from them that they've got enough supply to cover us for that fiscal '27. So I don't think we're going to see [indiscernible] key commitments there. And like I said, we feel very confident in our ability to continue to get that supply from them. We wouldn't say that on records unless we actually felt it, and we heard it directly from Broadcom leadership all the way up at the top. So I feel -- and Ed has had that relationship for a long time.

It's less of a, call it, a strategic partnership level, meaning the size and scale of the companies, it's more -- the actual chip you need is not constrained versus where it is for Arista?

Correct. It's a different chip.

Right. Understood. Maybe going back to the growth aspect, right? You had your Analyst Day, I believe it was towards the end of last year?

Yes. End of November.

And you set out a 10% long-term growth framework. Last few quarters, you've been growing nicely above that. I know between ...

11% last quarter. 12% for the year.

Yes. How do you think about that, right? Where is the growth improvement versus your long-term expectation coming from? And how sustainable is that?

Yes, a few things. I'd say from a demand environment perspective, we continue to see very strong demand. And it's not necessarily a pull forward. We see -- there's a few things. One, there's a refresh happening right now. Intellectually, if you think about it, people weren't really in the office at all during COVID, '20, '21, '22, even '23. People started to come back in the office in '24 and '25. Here we sit in '26 and full on, everybody is back in the office. Well, now people are back in the office, but the networks weren't refreshed over the last 4 or 5 or 6 years. And now you've got all this Claude usage and other LLM usage that's happening and it's running over the network. And so people are realizing very, very quickly that the network that they have today are not going to sustain them for tomorrow. That, coupled with Cisco's announcement last June, that they're going to have a massive refresh and that they're end of [indiscernible] lifing a lot of hardware has got a lot of customers thinking about, "Hey, do I need to refresh my network?" And then above and beyond that as well, you've got a lot of customers that are just thinking about lead times, like it's now 52 weeks with the memory shortages. And so they've got to get ahead of what their refresh is going to look like, well more in advanced now than they ever had do before, right? And so that's causing everybody -- and then you get the price increases happening across the board as well, with memory shortages and other components. And so that's what's causing a lot of customers to think about, "I got to get ahead of this. I got to make sure that I'm being planful about what my network refreshes look like." Oftentimes, we work with our customers and our customers tend to -- I think about like a higher education -- they do higher education university, they'll do their athletics first, and they'll do their dorm second, the they do administrative buildings, then the classrooms. There's this cycle that just happens kind of consistently on a 4- or 5-year period where they're always refreshing. Well, now you've got WiFi 7, 50% of all of our wireless bookings last quarter were Wi-Fi 7, 37% of our shipments were WiFi 7. That was actually faster than we expected it to happen. And so we see this kind of convergence move into WiFi 7 as well.

So if you think about the growth opportunity, is it more about displacing or moving up the knee, like you said, versus some of the bigger incumbents? Or is it more about this kind of secular WiFi 6 to WiFi 7 protocol? Or is it kind of sprinkled in between for the both of them?

I think it's those 2 things. But I think the third is really we're embracing AI differently than our competitor set. And the traditional way that people run networks is they have a network that's installed and then wait for problems to happen. And people put a [indiscernible] trouble ticket in and they say, "Hey, I'm having problems with my video." And then they go investigate and they look at a lot of the log files and they find the access points get a firmware upgrade that's due. And then they go fix it, and then they go and tell the person that is all set and then they close out the tickets. And it's very reactive and it's very like trouble ticket-centric. And we've now moved the entire industry towards a proactive approach to managing our network and this is an approach that is like let the technology help you solve these issues before they become issues and let the technology help you become safer so that you can look at your network overnight while everybody is sleeping and then they can come and tell you in the morning, "Hey, this is what happened in your network overnight." Those are the sort of kind of paradigm shifts that we're seeing with our technology now that customers are really embracing. And you want that because the new technology has that, but the old doesn't.

Right. I think that conversation will come just a little bit about the new Agent ONE platform that you guys debuted pretty recently. But maybe going back into the market dynamics for just a second. If I quantify the market right now, it's give or take, Cisco is about 50% of the market. HPE, Juniper is about another 10% to 15%. As I kind of mentioned earlier, you're around the 3.5% mark. So you've talked about 2 growth dynamics or 2 competitive displacement dynamics. One is Cisco through this end of refresh, but also HPE and through the kind of integration pains with Juniper. Now if I just look at the scale of the market share of both of them, is there a difference in how much you're winning against Cisco versus HPE? Or is it still kind of broad-based or winning against both?

I think it's fairly evened. I think the dynamic of HP and Juniper will continue to play out over time. they've got to converge the road maps. They've got to figure -- they still have 2 independent sales organizations that they're starting to integrate together, but they're still going out with 2 different products. And people are not like they're smart and they're asking which product is going to survive. And if you can't give them a straight answer, I think that, that creates fear, uncertainty and doubt around what I'm buying for the next 5 years? And is that product going to still be there? So I think that's going to kind of shape itself over time, they're doing really well, selling servers and whatnot, but we feel like we're starting to gain some share on the enterprise side because of that fear, uncertainty and doubt. And they've got $1 billion, I think their number to take out. And so a lot of employees are being let go as a result of that. So that's just naturally going to create challenges. Your account manager leaves, you no longer have a tie to the company like you used to, that tends to help us win those customers. On the Cisco side, some interesting dynamics here. Obviously, they've announced this refresh that they're forcing customers through that refresh with the end of life of certain hardware. They also just change their partner program to point system. That point system is enabling their partners to make money, but only if they sell all things Cisco. I think there's a lot of partners out there that are disenfranchised when they sell Palo Alto and they sell side-by-side with Cisco, but they're not going to make the points that they used to and the rebates that they use to if they don't sell everything. And so we're hearing from partners as well that are interested in being authorized resellers of Extreme because they want to be able to not have all their eggs in one basket because they're not sure what the rebate structure is going to look like in the future and obviously, it's dollars and cents is what matters to resellers.

Right. Also if I look at the pricing environment, I think Cisco has now made 3 -- 4 price increases?

4, 4 is what we've heard.

Yes. And you are at 2, right? And I think each one was about mid-single digits. Correct me if I'm wrong, but...

Yes, a couple of things there. One, I think this is actually -- so a, 4x versus 2x, we tend to sit 10% to 15% below Cisco. That might even be more wider now with the 4 versus the 2. I think the one thing that we also are doing going to market right now is that we're giving more certainty around the pricing that we're giving. So our quotes right now, if a reseller deal [indiscernible] with us, we will give them almost a 6-month price guarantee through November 1. We just announced that about 3, 4 weeks ago. And so our resellers love that because now it's price certainty, especially as we kind of cross over years with a lot of governmental agencies who are starting their fiscal '27 themselves in July. They're looking for quotes now for the next year. And all of a sudden, if you can't rely on a quote that's only 2 weeks old with Cisco and that the price could increase, they could cancel that order before it ships and give you a new price increase, that creates a lot of fear, uncertainty and doubt around what are you going to buy. And so we're trying to make ourselves, again, we try and make ourselves the easier partner to work with -- a technology partner. So we're giving them a little more price certainty and resellers, margin certainty on their end is one.

But how do you think about that in terms of you said you're widening the gap between your list price versus Cisco's list price, but the opportunity set that because Cisco is such a big piece of the market that you could also in tandem raise your prices.

Yes. I mean we can, and that's -- there's obviously a delicate balance. We want to be very sensitive to price and customer elasticity there. And we want our customers to be understanding that we are trying to pass on real cost to them, [indiscernible] we're not price gouging. At the end of the day, we want these customers to be happy with the technology that they're buying and realize that there's real value there behind it. And again, with the technology, the software, in particular, we're trying to make them more productive than they were before, managing their network. And even one research company just came out and said that we are about 32% cheaper than Cisco over a total cost of ownership over a 5-year period. So it's just not the upfront, and it's not just the licensing simplicity that we have and the cost of the licensing, but it's also like the things like the POE, just the energy usage of that equipment over time, we save you costs. So that's what we're trying to differentiate.

And how do you think about it from a profitability perspective. Two quarters ago, you did see some margin pressure from the memory constraints, but you came out last quarter and said you can protect the 62% gross margin level, 15% operating margin level [indiscernible] right. Do you think that that pricing stays even over the next 1 to 2 years as memory eventually starts easing? And does that help both growth and profitability?

It's a good question. Time will tell whether the cost of memory starts to come down. I don't think that we are going to find ourselves in a situation where memory constraints are lifted and that there's just not enough new memory suppliers coming online in the next 12 months. We just -- no one sees it. And so while we've secured that and we've got even some customers that are delivering DDR4 memory to us, and we've got other ways of being able to get that DDR4 memory into 2028. WiFi 8 is going to go in DDR5, and that's about 18 to 24 months out. So that's where I think we're all going to see as WiFi 8 being at the DDR5 side. I think at the end of the day, for now, we feel good about where we are from a supply chain perspective in sourcing the memory. We've bought the memory for 2027. That is for sure and we feel good about being able to deliver that.

Right.

And then margins like -- the margins, like you said, 62% at the low end, we said at the low end. We think as we continue to drive more platform on adoption, revenue mix will continue to increase and [indiscernible] that we get margin improvement that way as well.

Maybe to turn to the AI angle a little bit. I think AI presents an opportunity for you from 2 sides. One, Cisco talked about, as we move more towards inference, there's going to be a kind of higher sustained traffic pattern on the network that facilitates some need for this camp infrastructure upgrade. I think you guys have a little bit of a different tune when you think of that opportunity?

Well, there's a few things. I mean: a, I do think network traffic is going to increase over time. I would agree with Cisco on that. I think two, but it's not going to all be in the hyperscaler environments. I think there's going to be a hybrid. I think that customers are going to come back to building out their own data centers. When you think about customers, they have different types of data. If you're a company in clinical trials, that clinical trial data is really, really important to you, you may not upload that into a hyperscaler environment. If you've got HIPAA data, that may not go outside your 4 walls, either. And so the sensitivity and then, of course, in Europe, you've got data sovereignty and that's a big issue there. I mean at the end of the day, we're seeing enterprise customers starting to think about more to have a hybrid -- [indiscernible] of using the hyperscale environments, but also that critical data I keep and I put in my own data center. And again, we can kind of go from the data center, which we due today, about 15%, 20% of our total revenue, it comes through data centers all the way through [indiscernible] to the core campus all the way to the edge. And so kind of extension all the way through with our fabric technology, with our AI, you can get full visibility of your network all the way from end to end. That's what we think is going to really help us and be able to drive more volume.

Right. But if I think about the data center opportunity, historically, at least, you've been a little bit more tied to the service provider use case, right? I think you called out partnerships with Verizon and Ericsson, as an example, right? You had your user conference a couple of weeks ago, which I was at and you debuted or you discussed an enterprise data center switch. I think it was 800 gig.

400- and 800-gig.

So can you talk about that opportunity? Why enter the market now? Does that -- are you limited to just the enterprise network? Or do you think that you can grow into, I don't know, the hyperscaler or maybe the Tier 2 clouds, the neoclouds of the world?

I would say that our focus is on enterprise. I think it's always been. I think that's the first and foremost because our reseller network has got relationships with these customers, and we're a trusted adviser to those customers. So it's much, much more natural for me to have -- from my resellers and my salespeople or my sales engineers to have a conversation with an existing customer that knows my technology, and they're going to go from the core all the way out to their data center. That's very, very natural. Could a 400-gig or 800 gig switch go into a different data center? Maybe. But that's just not our DNA right now as a company. And so I wouldn't want to overarch just say, yes, we could easily. The reality is we just don't have those relationships at this point. Can we create some of those relationships? Sure. It'd probably be through a reseller relationship, [indiscernible] necessarily direct.

Got it. You started talking about it earlier, but I want to go into the Agent ONE platform that you announced. If you could first go over what it does, I know you have 2 different forms with 2 different pricing models. So if you can just discuss the differences between the two.

Okay. Okay. And maybe we'll back up a little bit to Platform ONE like what it was when we first went out, right? So we introduced Platform ONE [indiscernible] and the Agentic model was really, what I would say, a query model. So our first generation of Platform ONE that we announced last July was you asked a question just like if you would ChatGPT or Claude, you ask it a question and it will give you an answer and you can query and you could constantly query and you're trying to get those answers back and you're trying to basically pull information out of the system that way. The new generation of Agent ONE, which we're calling basically more agent actions, there was a coworker mode and then there's an operator mode. The coworker mode, think about it as a virtual desktop IT agent that's working for you. And so you're able to work with this agent and basically have it do tasks on your behalf. Now you're not just doing queries, now you're actually doing actions, agent actions that are helping you protect your network. So that is a light year ahead of where everybody else is from a networking perspective right now. And even Cisco has kind of announced what they are doing from a AI perspective like a day or 2 ago. But the reality is that's akin to our first generation, and we're already leapfrogging what they have with our second generation, which is coming live. We just showed it on stage live demo. We have not seen any live demos whatsoever, but [indiscernible] Cisco has. What we showed it live on stage as a demo at our user conference, and we're going GA with that. It's in limited availability with certain customers today and we'll go live with that in July. That's exciting. So then there's an operator mode, which we talked about as well, which is our agents working with other agents, [indiscernible] like an agent exchange and so creating like a ServiceNow agent, working with our agents and being able to manage tickets simultaneously with our networking agents to be able to solve issues.

Right. And the pricing mechanism growth?

Yes. So right now, Platform ONE tends to be about about 15% higher than if you bought our old, what we call Extreme Cloud IQ, ironically IQ is what Cisco is calling [indiscernible] there is now and then we also had support contracts separately. We've bundled those together and then we added our AI on top of that, and that's what Platform ONE was. Then you've got Agent ONE, which is going to be slightly higher than that because what we're going to do is we're going to monetize the agent actions. So within your Platform ONE subscription, you're going to have, what I'd call, a reasonable amount of agent actions that you can use that's embedded within and then you can buy more blocks of the agent actions should you choose to go and do more right? And that's the monetization there.

Is this an area that needs customer education, meaning, is this more about future-proofing the portfolio now for this world of Agentic network operations? Or do customers kind of understand the problem, understand the benefit from adopting agents for their networks?

Yes. Well, it will take some time, naturally, like anything. But I think this is not a use case where a company is going to spend their time trying to use Claude or Claude code and try and solve this issue on their own. I'm going to market to solve the issue for them as a product feature and not necessarily force them to go do it themselves. And the benefits they're going to get out of it is they get a better view of their network. They get a better proactive view of what's happening within their network and naturally there are natural security ramifications of that if you've got over-the-top security today on top of your network, well, maybe you don't need as much of that security in the future. And so there is -- at least for the smaller businesses, they can actually manage very well with what we're doing from an agentic AI perspective versus buying some software on top. It's not going to get [indiscernible] rid of checkpoint, but it's going to be just better visibility of what's happening in your network.

Yes. And are you competing in the space more against the vertical players or the horizontal players. Meaning are you going to see more companies like Cisco and HP come out with their own agent platforms? Or will you see AI native companies in the space?

No, I think that we'll see the, I think, slower than us because we seem to be more innovative, more nimble. I think that Cisco and HP, Juniper will come out with their own Agentic AI. It's just going to be 18, 24 months behind us. We were already doing this if we -- when we launched last July, we had backed up from that 18 months prior to that to build that agentic layer that we had.

Yes. So we only have a few minutes left here. If there's any questions in the room?

Technology is one aspect. What about go to market. AI change your go-to-market at all for the same [indiscernible]?

We're using AI actually to help our go-to-market. So a few examples, RFPs. We have put all of the historical RFP responses into a container into a RAG model. And then we are -- naturally any new RFP that comes through, we are basically putting Claude code against that to be able to kind of have 90% of their RFP response ready and going and then we're filling in the rest. That's one thing that we're doing. We are using it also for empowering our resellers differences between Cisco and Extreme, enabling them from an AI perspective to be able to fill in those queries, ask those queries and be able to be very responsive to our customers. Obviously, cell sheets and things like that, being able to generate those as well. So I think we're embracing it from a sales ops perspective and being able to -- not only our own sellers but also our partner resellers to be able to make them more efficient and more productive for sure.

Have you changed -- so you're running very fast on the technology side. Talk about your go-to-market in general, talk about AI, what have you done in the last few years to broaden to the market [indiscernible] what are the efforts -- what are the part of the efforts that you have done on the old margin.

One of the things we did, Tal, is that we looked at our reseller universe and realize that we have a great, very sticky, very loyal reseller base. But from a growth perspective, many of those resellers, owner owned, they're very happy with us, but growth was not necessarily top of mind. They have [indiscernible] $30 million a year or $50 million a year, and they're very happy with the size of where they were operating. What we realize is that we need to move up market and we needed to move upmarket with larger resellers, more reach, more ability to kind of generate more demand for us. And so that's -- we've been on a crusade, if you will, for the last 18, 24 months, we actually hired the head of partnerships from Juniper. We actually hired the Head of European sales from Juniper. These were 2 wonderful adds to our management team. They have obviously very good relationships with their resellers and also with sellers within Juniper as well who are maybe out of job. And so we're getting just more and more, I'll call it, halo effect from a go-to-market perspective based on some of those things that we've done. And so things like [indiscernible] NTT East, things like companies like [indiscernible] Softcat or computer center or WWE or CDW, we're building much stronger relationships with these larger resellers that have a lot more volumes.

Last question is about the portfolio. So we've seen -- if I take a 5-year on a quarterly I've seen switching companies, expanding the portfolio into adjacent areas. So away into SD-WAN, anything to WiFi before the view is to provide a much wider portfolio your vision, your own vision of complete you where you want to be? Or do you think you need to grow horizontally into adjacent markets in order to provide a wider portfolio over.

Yes. It's a good question. I would say from a product perspective, we've got a niche and that niche is just enterprise and enterprise networking. We're focusing on the WiFi 7 portfolio expansion. We've now created a ruggedized switch. So that's kind of an interesting 1 that people were asking for for a while. So now we have that. Now we've got 400-gig to 800-gig coming online. I do feel like we are expanding the portfolio more than we've done in the past. The reality is, could we do -- you could always do more, but it's really doing it all, doing it well. is an important element for us. And we believe that we have very sticky customers. And the reason why we have such loyal sticky customers because we deliver for them, and we do it with high quality. And we don't want to basically take away from that by expanding too much adjacently and then basically not doing the things that we need to do right.

I think Tal might have looked over my question list because he addressed some of the things that I was going to get to you. But unfortunately, we're out of time. So kevin, we'll bring you back next year to round it out. Thank you so much.

Thank you very much, Tomer.