We'll go ahead and get started. Welcome everybody to the third day of Needham's 26th Annual Growth Conference. My name is Quinn Bolton. I'm the semiconductor analyst here at Needham & Company. Thank you for joining us. It's my pleasure to host this fireside chat with MaxLinear. MaxLinear, as many of you know, is a leading supplier of RF analog and mixed-signal semiconductors for the broadband and connectivity, infrastructure, and industrial markets. On stage with me is Dr. Kishore Seendripu, CEO, and Steve Litchfield, CFO and Chief Corporate Strategy Officer, and we also have Leslie Green in the audience, who handles investor relations. So Kishore, Steve, Leslie, thank you for joining us. I wanted to start off, guys, with maybe some macro questions or just industry questions.
Obviously, the entire semiconductor industry is facing inventory digestion as you have, particularly in the broadband and connectivity business. But, you know, what are your latest thoughts on where we are in that inventory digestion process? And, you know, how long does it take to clear inventory in, you know, your various end markets?
So I think, you know, not much has changed since our last earnings call. We, you know, we are burning through inventory, right? And we are under shipping demand quite substantially. And if you see what took our revenues down is really almost entirely concentrated in the broadband connectivity business, and they're one and the same as far as revenues are concerned. But the remaining portions of our business is infrastructure, industrial multi-markets don't have too much inventory, this thing. So broadband and connectivity, we've been tracking it, and we're vastly under shipping demand, even though the demand seems fine. So we expect that maybe second half of this year is when we will start seeing some recovery to happen. And, you know, it's like fits and starts.
We're seeing some cases, some parts are in short supply, and you see some bookings, and the others you don't see, so I think it'll really be the second half of this year. Regarding the infrastructure and industrial multi-market, we don't see, we didn't have that overshipping problem other than some exposure to the macro markets that are softening. So we expect softening in industrial and infrastructure markets in the first half of this year as well. So-
But that's more macro?
More macro than really any large-scale inventory issue that we are seeing in the broadband and connectivity space.
If you're right that broadband and connectivity, where you have the bulk of the inventory problem is, you know, clear, sort of by the end of the first half, and hopefully gets into a better situation in the second half, at what point would you start to expect to see bookings begin to pick up? Is that, you know, a quarter ahead of, you know, kind of the recovery? And, like, how far in advance would you start to see it in the order books or at least order trends becoming more consistent?
I mean, typical lead times are 16-18 weeks, so they-
16 weeks before?
Yeah.
Okay.
I was gonna say, it won't be any faster than that. But I guess I would just... I mean, look, I think bookings are improving. I mean, we're starting to see signs of improvement, even on the broadband side. I mean, even where there's inventory in the channel, there's never a perfect mix of inventory-
Yeah
... right? So there, you're starting to see signs, and so things are improving. It's not as much as we want, no doubt about that, but definitely signs of improvement.
Got it. Okay. You mentioned sort of normal lead time, 16 or so weeks. So are we back to, to that level at this point?
Certainly.
Yeah.
Certainly.
Okay. You know, I know there are a lot of folks who would love to know what's going on with the SIMO arbitration. I know it's private, so rather than me ask you a lot of questions you can't answer, maybe I'll just open it up to you. So can you give us any update-
Sure
... to where we are in the SIMO arbitration?
Yep. So yeah, familiar with the question. Really not a lot has changed. It's not a lot of an update. So, Silicon Motion filed for arbitration in October. We responded, and so now we'll go to Singapore, per our existing contract, and go through arbitration. It's expected to take 12-18 months. It's all done confidentially, so we don't expect to have a lot of updates for you in between now and then. Another common question, I'll beat you to the punch, is, you know, like, what is the expected outcome? Clearly, we've filed, terminated the deal, don't expect to pay a breakup fee whatsoever.
I think their position is they want a breakup fee and, legal expenses, damages, et cetera, which, you know, as I'm sure many of you know, typically, our position and most people's understanding is that kind of gets incorporated into the breakup fee. So the breakup fee is $160 million, so most folks have kind of framed it around $0-$160 million. Certainly, there could be other outcomes, but that, that's kind of where most folks frame it.
Okay. Is it a binding process?
It is.
It is. Okay, so no, no appeal.
Correct.
When the ruling is given, it's final and binding.
Right.
Got it. Okay. Just moving to the broadband connectivity, you recently introduced your Puma 8 solution. Tell us a little bit about that. I think it has both Ultra DOCSIS 3.1 and DOCSIS 4.0. Maybe what is Ultra DOCSIS 3.1? You know, and how do you see adoption of Puma 8 over the next couple of years?
So what we've announced is what is called the Puma 8 chip, which is basically a full-blown DOCSIS 4.0 chip. And what DOCSIS 4.0 does is that it gets you, you know, 10 Gb downstream just to keep it, you know, approximately 10 Gb downstream speeds. And upstream, you can go somewhere between 5 Gb and 10 Gb upstream data speeds. But the problem with DOCSIS 4.0 deployment, it needs a massive network upgrade, which is very, very expensive. To put it simply, that you need to, you need to maintain a certain signal quality to every home that receives DOCSIS 4.0. That means you need to put a lot of hardware in between to make sure the signal doesn't get destroyed as it goes through old lossy cable coaxial lines, right? That's the thing.
But the trick is, how do you provide the speeds to these service providers, to enable to provide that data rate that competes against the fiber guys, you know, with existing network? So what we call Ultra DOCSIS, some people call it enhanced DOCSIS 3.1. So what it does is that it says, "Okay, guys, today you're getting less than about a gigabit of upstream speed. What we're gonna do is we're gonna shift the partition of how much bandwidth we reserve for upstream and downstream.
So we're gonna give you somewhere like 3 Gbps-5 Gbps , you know, upstream speed by occupying more of the cable bandwidth for upstream and shifting the downstream a little bit higher so that you still get somewhere between 5 Gbps-10 Gbps on the downstream." So what that does is it gives a bridge for the cable operators to nicely with very little expenditure in the network upgrade for the same box, to scalably use it to transition to 3.1 Ultra DOCSIS and eventually to DOCSIS 4.0, full-blown, to get to 10 Gbps speed.
Yeah.
Right?
Yeah.
So it's a cost-effective way of transitioning so that you can pace your CapEx if you're an operator, if you're a service provider, at the rate at which customers will want service to be upgraded, right? That's, that's the beauty of it.
Got it. So you upgrade the CPE device today with a Puma 8-
Yeah.
you can enhance the service without doing-
Yeah
a lot of network investment. When they're ready to do the network investment, they've got a DOCSIS-
Yes
4.0 capable CPE already.
Exactly. They've got the DOCSIS 4.0-capable CPE, and they turn on the DOCSIS 4.0 capability in the network based on the user upgrade point, right? So to speak.
From the cable MSOs, you know, as RFPs are coming out, are they now starting to specify DOCSIS 4.0 or this enhanced or Ultra DOCSIS 3.1 feature? Or do you think that those RFPs are still, you know, later this year?
So I would say that the... when RFPs come out, they want to specify both, right? It has to do this, it has to do that sort of thing. And you'll hear a lot about unified DOCSIS and so on and so forth. That's just the buzzwords for Ultra DOCSIS and then DOCSIS 4.0. So the silicon has to be capable of supporting both, basically.
Okay.
Yeah.
Okay. You know, you mentioned competition from fiber, and I know you have a fiber play.
Yeah.
What's the size of the overall cable modem gateway market today? Is it growing? Is it actually starting to shrink? I think in the past, we've probably talked about 30-40 million units a year. I'm just trying to level set folks. Is that still the right unit TAM? I know we're going through an inventory correction, so it's probably not the right TAM to this year or in 2023. But is that kind of still the opportunity for cable, or do you think it has shrunk?
Oh, yeah, but right now, it's definitely shrunk, right?
Right.
That's why it depends on who you ask the question. But I think it has shrunk, it has pulled back. I don't think it grows, I think it's stable. And then I want to, the units-wise, but there are two kinds of units in cable. One is the gateway units, and there's the pure modem units. The silicon opportunity in modem is half the BOM, the bill of material of a gateway, let's say, just to keep it-
Yeah
... because if you look at a modem, it's just got the DOCSIS chip, right? And then it's got the front end for the DOCSIS, right? Whereas a gateway has got the modem, plus Wi-Fi, plus Ethernet as a connect, you know, the full, and then the network processor that can handle the Wi-Fi routing. So you've got basically a router and a modem combined in one. The other one is a pure modem. So the content is increasing in the gateways, whereas modem, you know, the... So with the upgrade to the new standard, the ASP is increased by 30%, let's say.
Okay.
Right? So I don't see... So I say it's a flat TAM, if you discount for the inventory correction. If you offset for the inventory correction, it's probably a flat TAM.
Okay.
Right, and I know there is a lot of talk about subscriber losses to telco and all of that stuff, right? It's very, very... in this period, it's hard to sort of, you know, gauge it, but I think it's a flattish one as well as dollar TAM.
So, units probably down a little bit, content's going up-
Yeah, so it's kind of flattish, yeah.
Is share position of Broadcom, do you think... I, I think, I know there's a lot of noise here in the last year, but do you think it's still roughly a 50/50 split, or, or do you think there's been some share shifts within cable?
It takes a while to settle down because this whole deployment model of DOCSIS 4.0 versus DOCSIS 3.1, it all has to be solved. And, I don't expect substantial shifts. You know, I would say it's always been, the world has been 60/40 or 50 ±. So could it be 40? Yeah. Could it be 60? Don't know. You know, it, it really varies, you know. So I would say nothing substantial has changed other than the dust has to settle who's deploying what on DOCSIS 4.0 versus modems versus gateways, basically.
Okay. Lastly, Puma 8 shipments begin when? Later this year, 2025?
I really think that it's really a next year phenomenon for the industry as a whole.
Yeah.
I think the upgradable, scalable Puma 8 equivalent functionality starts shipping production-ready probably first half next year.
Okay. Switching to the fiber side of the broadband business, you know, this was obviously smaller, starting from a smaller base. You'd previously talked about sort of $10 or less than $10 million in 2021, I think $30-$40 million or so in 2022. You had expectations for growth in 2023, but obviously, we know the entire broadband business was tough with the inventory correction. Did you actually achieve growth in 2023 in fiber? And can you give us any sense of the outlook for 2024 in terms of growth in fiber?
So let's just focus on the past. It was barely anything three years—I mean, before 2022, and then the numbers you said are right, and it was like about $30 million in 2022, and we—2023, we expected double. We didn't quite double, but we actually-
Okay.
grew quite substantially.
Okay.
Right? Let's say it's $50 million range-ish, right? So we grew less. We thought maybe it'll be $80 million when we told it'll double, but it turns out to be somewhere in the $50 million-$60 million range sort of thing.
Okay.
So we kind of did well, but, but there's one Tier One operator. So in, in fiber, our position is most of our shipments have been so far on the SFPs. If you look at the fiber PON market, there are two deployment models. One is you terminate the fiber at the entry point to the home, and then you connect to an Ethernet gateway router inside the home. That's one way to think about it. The other model is get the fiber straight in and terminate everything in the big gateway box itself. So far, our shipments in PON have been not gateway on the SFPs. That means we've done quite well, actually, very well. But if you look at the gateway, we've just ramped a, a Tier One operator in North America. We got some good traction going on, another Tier One operator in North America.
In Europe, the bids are going out right now. Frankly, they've been pretty slow, the RFQs, so they'll all be finalized by the end of this year. And then maybe you got more gateway ramps next year and some other smaller Tier Two gateways also happening next year. So I think it continues to ramp, slower pace, and it'll pick up more pace next year. So PON opportunity is almost 2x+ on the cable side, so that's going to be the exciting growth for us since we have such a small position in fiber.
Got it. In competitive landscape, is it mostly Broadcom? Do you see any of the Asian or Taiwanese vendors active in North America and Europe?
So I would say from a full BOM, the entire content platform-wise, it's Broadcom and us, right? But there is these Taiwanese vendors, a combination of, you know, Airoha, MediaTek or Cortina sort of thing. You know, there's some combination there where they're usually the third player, and obviously, it puts pricing pressure on the ecosystem. That's what they achieve. But generally, the way the market plays out is that there's a bias towards a full solution because ultimately it's more cost-effective once the pricing pressure is imposed on players. So we expect for us, it's a growth business. We're not going to be losing share because we're building from one thing to the other, and... But that's the dynamic there.
Okay. On the connectivity side, maybe just speak a little bit about your design win momentum in Wi-Fi 7. I know you've said there's a content uplift as well, and maybe just reiterate what that content uplift is.
So, okay, you know, today, all our Wi-Fi revenues are Wi-Fi 6. Wi-Fi 7, we just announced that we have been selected, along with three other players, as a certification bed. Once again, we had that position in Wi-Fi 6E. We got that again in Wi-Fi 7, which is a big deal, which means that we're automatically qualified. Everybody else who wants to launch a Wi-Fi product, who's not these four players, they have to certify against each other, against us. So that's good news, but mostly right now, the RFPs are being issued are for Wi-Fi 7. So the ramp really happens middle of next year sometime, and maybe we will ship some Wi-Fi 7 at the end of the year, but nothing meaningful because, you know, the quotes are going on now. It takes a long time.
Another thing is that iPhone 16 may have Wi-Fi 7. The current new model of iPhone doesn't have Wi-Fi 7. So those are the dynamics that drive it. The content uplift is usually 30%, right? If you look at Wi-Fi 6, maybe it's $8-$15, you know, $8-$12 to $15. The other one will be, you know, maybe $10-$20 sort of range.
Okay.
The shift, basically.
And, you know, before the inventory correction, part of the Wi-Fi growth was you'd done very well, sort of in the service provider market. You were starting to kind of move a little bit more, I'll call it retail-ish-
Yeah.
because I think it's in Asia, the service provider, you know, some of that channel is through the retail market.
Yeah.
But, you know, is there an opportunity for Wi-Fi 7 in sort of the retail or that, that same sort of Asia channel, in addition to the more traditional North American, European service provider?
Absolutely. We had very strong traction there, and we're getting a bunch of design wins, and we had silicon supply shortage, so we overwhelmingly biased towards supplying the, the bigger players, North America customers. So we starved those guys, and therefore, we kind of stopped our own momentum by a decision who we allocate the silicon to. So we're back at it again, and definitely the opportunity is there on the service provider class, Wi-Fi, which is kind of a different model, retail. In, in China, for example, Huawei and ZTE dominate the market, but on the Wi-Fi, Ethernet, and stuff, we can also ship basically, right? So...
I'm sure you've gotten this question. I'll ask it because I think it leads to a more interesting question. But, it's been a few months now since the Vantiva-CommScope deal was announced, and, Vantiva's been historically more of a Broadcom shop. You've been more of an Arris, CommScope shop. You know, some people worry that, geez, maybe they consolidate around one silicon provider. Your thoughts there? The more interesting question is, you know, do you think you see a move away from that OEM, you know, gateway model entirely, and it becomes more of an EMS or contract manufacturing business?
Okay, so that's a, that's a good question. First, firstly, let me answer the Vantiva-CommScope one. Really, the decisions about whose silicon is used is not made by the OEMs. The operator is the one who chooses it, and for them, supplier diversification is very, very important. So what we expect is that more ODMs will become OEM-like, and they come on board first, right? And then that's one. So I'm not worried about that. As I think the silicon choices are made by the, by the service providers. And when I say that, that means we have more knowledge than, than that, even to say that we feel confident that the service providers pick the silicon. So we're not worried about that. So I just think the, who we supply to could change based on decisions to be made, number one.
Number two, whether they go EMS, I think they have tried that before, and I think that some have succeeded. We have seen that the satellite world, for example, with DirecTV, and you saw with BSkyB in, you know, Sky in the UK, who did that very successfully, but then nobody else has been that successful. I think the operators are moving towards a direction of owning the designs and everything and giving it to a contract manufacturer to do it as a backup, rather than doing it themselves. But they're gonna own more and more of the middleware software and above. They're trying to standardize it so that it's just a commoditized situation.
Okay. Okay. Lastly, on the sort of broadband business, you know, what are you hearing from the cable or fiber broadband service provider customers in terms of, you know, when they think the RDOF or the infrastructure dollars really start to flow?
That's a bigger question. Really, from our point of view, it's, you know, whatever gets deployed, they'll have to use our silicon if it is a PON side or is it a cable, cable play, right? So we should benefit but on the PON side, you benefit from the tier two players-
Like the-
... Calix's and the Arris's and those sorts of things, you know, and we should receive some of that money in some form. But it's really for their CapEx, not, you know, and then it'll translate to silicon. So really, with this whole inventory stuff, it's very hard to say, you know, for how it's gonna play out. I think we just give it another six months and, you know, things will clarify a little bit.
Okay.
Okay.
TBD at the August conference then.
Yeah. But you promise me next time you're gonna ask me more about infrastructure as well.
Oh, no, I've got six or seven here on infrastructure.
All right. Okay.
In fact, that was where I was going.
Okay.
Obviously, a lot of momentum in data center spending, optical infrastructure, especially as AI, you know, continues to be adopted. You know, we talked last week, you seemed most excited last week in our meeting about your Keystone DSP and the optical DSP progress you're making. And so maybe give us the latest thoughts on the optical DSP traction you're seeing.
So I wouldn't specifically bucket it around the optical DSP. I think we're very excited about infrastructure. We've been investing for the last five years to build the infrastructure business, practically from zero, right? Now, the infrastructure business is about $200 million per year, and we got multiple growth vectors in each. Out of which, in those incremental growth opportunities, data center is the biggest opportunity, right? We expect that the revenues based on 800 Gb deployments and our anticipation and proof points that are developing about design wins, that we should get to about a $100-$300 million range of revenues in a 3- to 5-year window, based on a market share of about 20% or so. Right? So if you look at the transceiver market, you know, the numbers are all over the map.
Let's assume there are 20 million transceivers this year annually shipping, and then we expect that to double in three years to 40 million transceivers, right? And out of which, 70% will be 800G. And the expectation is that 800G, really 100G per lambda, will be one of the longest, longest life duration deployments in the data center for high-speed interconnects. And even though there'll be announcement about 200 gigabit per lambda, which is the 1.6T PAM4, probably in OFC, they won't see the light of day in real shipments till 2026, 2027, right? So it didn't give the last lot. So we got the right, right horse and right jockey, so to speak, right now. So out of that, we got the 5 nm Keystone DSP silicon. That's our 800G horse, and it's got very good traction.
It's got extremely low power and performance proof points that are resulting in design wins. Actually, compared to Q4 call, you know, we feel increasingly bullish and confident that the shipments are happening now. We actually started shipping some volumes now, and there are more ones to come online, so feeling pretty positive about it.
As you're working with the hyperscalers, can you tell, do you know, is the application for general purpose—or I'll say, just the general switch hierarchy, is it an AI cluster too... Will they tell you sort of what the intended use of that optical module is, where in the network it's going?
You can approximately guess, but because all transceivers are not equal, they're long reach, short reach, and then what kind of lasers they use. From that conversation, you can kind of figure out where they're going. And there are many flavors of it, right? From the server to the rack, and that's a different. You can pretty much tell by what they're qualifying the, what kind of, module level they're qualifying, you can tell where it's going. But really, when you talk of AI right now, it's really the big demand is really NVIDIA, and, and that is you had—you have to have been there when that, that bubble, kind of ChatGPT thing happened. We are not there, but they'll also be diversifying the supply base, and hopefully, we have a shot at, being a participant in that.
So at the data center side, they don't give you enough information on that, but you can kind of glean through that. I think it's we sell the same DSP to every category, so it doesn't matter to us whether it is, we have a couple of SKUs, depending, of course, optimization, power optimization, so it really doesn't matter to us.
But you can tell if it's a VCSEL-based, probably short-
Exactly.
That might be, you know-
Yes
... one application. Okay. So do you think that you're being qualified within sort of AI applications in at least some of the design wins?
We're definitely qualifying designs in VCSELs right now.
Okay.
Right? So that's... So you want to understand how this sale is processed. We, the data center guys you have to work with, you have to work with module guys. And the data center guys tell the module guys whose silicon you want to have modules ready, but they play different. They ask multiple module guys to do multiple silicon stuff. They pick who your horses are. They probably have four or five module vendors, and they'll end up choosing three. Two get the majority volume, and the other one gets the minor, you know, you know, smaller share. Some you win, some you don't. You know, on an aggregate, we hope to get 20% market share, right? That's the thinking. So, so when people talk of AI and everything, from my point of view, on the PAM4 DSP side, it's really about what kind of module.
Are we getting enough design-ins in different categories? Absolutely. In fact, we have design-ins and design wins in long-reach type data transceivers. We have in short-reach, we have VCSEL category, we have MM, we have active optical cables, and also active electrical cables, which right now is very, very tiny, but we expect to grow to about 10%-15% of the market in three years from now, right? So we got eggs in every one of them, but right now, the shipments are the ones that are getting past the design phase or the regular transceivers and active optical cables.
Okay, perfect. Another area of strength that, I think certainly caught us by surprise over the last quarter was your Panther storage accelerator.
Yep.
You know, that was something I think that originally came with the Exar acquisition. Didn't really seem to generate a lot of revenue in the first couple of years, but, you know, you sized it last call, where I think it's gonna be, you know, kind of, it was less than $10 million , maybe a double this year-
Mm-hmm
... and $30-ish million in 2025. So starting to see some pretty good traction. Can you just talk about the application, the use case, and what's driving that growth?
Okay. So, we acquired this company called Exar, and they had this really a hardware accelerator for storage compression. If you look at today, when we had low-speed memory, like the hard disks, you know, memory, you know, the magnetic storage and stuff, very low slow access speeds. So they use a lot of processing power in the, in the CPU to do the compression. So, but as the speeds increased and you get faster and faster memories like NAND flash and things like that, the, and ferro memories and stuff, so the access speeds dramatically increase with the memory. So now you are underutilizing the memory speed and how you do the computation and everything. So what they started using, using the high-end Intel processor, for example, do 50 cores and stuff to do it in software.
But still, that's very, very expensive because now you're using a 50-core processor for a peripheral function, number one. Number two, it's very, very power intensive, hugely power intensive. So the natural logic is just like the AI processors, is to offload into a hardware acceleration engine. And, and the thing is that that's what we, we basically took that kernel of the idea and a very, very large enterprise storage appliance maker, one of the top, top-tier ones, and we designed a new chip for the future, which is called Panther III, which is a 200 Gb, you know, various levels of compression ratios, and security is a big part of it. And it basically runs at 200 Gbps , and you don't have to buy this fancy processor. You save a lot of money, and we'll be shipping the chips.
So this year, we told you correctly, it'll be about $20 million-ish, and we said a three-year window, we expect the revenues to be between $50 million and $100 million. Okay, that, but that's for enterprise storage appliance, server-type applications. But the real goal in the end is to take this data processor, the DPU, if you will, for storage, and move it into data centers. That's where the big, even bigger opportunity is. So we expect this business to be $50 million to. So let's take the algebra again, right? So infrastructure today is $200 million. three to five years from now, we expect, PAM4 to be over $100 million-$300 million. We expect this to be $50-$100 million, and then we also have enterprise Ethernet wins, Ethernet PHY wins we're having.
That'll be the same range, $50 million-$100 million range.
Oh, for the switch?
In the five-year window. Yeah.
Okay.
For switch.
Switch 5.
This thing. And so you can see infrastructure has a legitimate chance with incremental $300 million-$500 million opportunities in three to five years to be a $500 million business, which would be an exciting thing because generally, it's very sticky, stays for a long time, great gross margins. And that's the reason we're giving a longer-term model, about 62%-65% in gross margin expansion. Even though intermittently, we may have some turbulence, but we feel pretty good that we could grow the business.
Perfect. Just two more questions. I know the millimeter microwave backhaul has probably been the biggest part of the infrastructure business to date.
Mm-hmm.
You saw strength in 2023.
Yeah.
I think expecting a little bit of a pause in 2024. What's behind that pause? And then, you know, what's the outlook as you maybe get into the second half of 2024 for the millimeter and microwave backhaul?
We expect the first half to be softening, not because of any inventory issues, just the macro softening we are seeing in the telco situation right now. But we expect a recovery in the second half because there's not much inventory in the system. There's a lead time issue with these products, so we start seeing, expecting some recovery to start. And you saw announcements from AT&T and others of the big network spend coming to... The North America is not big on microwave, millimeter wave backhaul, but they do use them, right? And it's really outside North America. So I think the new spend that comes with it, we should benefit from it, right? And the big thing that's driving for us is not just the units, because the units are always flattish, right? But the amount of content in those units is increasing.
First, historically, they've been selling only microwave backhaul or millimeter backhaul. Now they're putting hybrid systems, millimeter and microwave backhaul-
Oh, in the same system.
In the same thing. And then they're doing also MIMO of that, like, you know, putting more of them so that you can make the pipe fatter. And then the RF content is increasing too, so we are benefiting from that. Not just the modem, but the RF content as well. And, you know, so that's the reason we feel that second half we should start seeing some recovery in backhaul revenues.
Last question on infrastructure. Any updates on the base station transceiver business? You know, any discussion with your traction relative to market share leader Analog Devices?
So look, the transceiver business absolutely has not been as big as we anticipated. You know, our real big horse was Huawei. I mean, now that they're on the... I can tell that freely. We would have run the table with Huawei, but when China went away, we kind of lost some traction. We have tier one design win and tier two design wins in the non-China market, but that itself is a small fraction of the whole arm. So right now the focus is to really increase the content on the access side, this whole O-RAN and everything, really increase the content on the bill of materials we can sell to these players. And I think that opportunity is a lot more in the O-RAN type of deployment space.
Okay.
Because if you look at the tier ones, they're trying to do custom silicon and really thin out even what our competitor can do. They're really, really trying to minimize the value proposition of component vendors like us, basically, right? They're really trying to keep it all in-house. I think that's the dynamic, so you're going to see a change in the strategy of us and our competitors, where the focus is in a brand-new area, where this silicon content escalation beyond the transceiver into the digital parts side as well on the remote units.
Okay.
I think we are very well positioned competitively for that because that's familiar territory for us.
Okay.
Okay.
Perfect. I know on the industrial and multi-market business, you said that that's not really a business that has been largely affected by inventory. But, you know, what's the outlook for that business in 2024? And, you know, are there any particular growth drivers in that industrial and multi-market business that you would want to call out?
If we just step back, right? We have seen infrastructure and industrial market really hold on nicely. They've grown actually in 2023, and we said we'll see a softening here. Even there, we don't have as much inventory situation there, but there is some China exposure, so, and, you know, all these industrial markets, you will have some macro exposure right now. So that's the only thing we worry about. But we-- but the growth has been coming through launching you know, small... You know, the way the markets, it's like bag of parts, right? You're launching new, new parts every so often at a cadence, and they take some time to bear some fruit. The growth has been coming that way. But we generally expect normal GDP-type growth sort of playing in that, in that market.
I think as we launch more products, as we take our IP in other markets and put it in that market, I think growth continues along that fashion, basically.
Okay.
So this whole broadband connectivity is the one that needs to recover for us because we don't see any headwinds. We actually have tailwinds in both big time infrastructure, but decent ones in industrial multi-markets. It's really a wait-and-see game for the inventory to flush out, you know?
Okay.
So in summary, right, infrastructure is a huge growth factor. We've got some very, very strong incremental $300 million opportunities infrastructure in the next three-year window. The industrial multi-market, decentish growth will continue, right, after the softening phase in the first half. And PON market will be a good growth driver, and Wi-Fi will drive BOM expansion. And, you know, all we really... From where we are right now, even if cable doesn't play as big as it used to be historically, that's growth.
Yeah.
Good. We got plenty of growth vectors to get to that $1 billion back within the next few years, right? So that's where we are.
Okay. And maybe just one or two for Steve. Steve, on the OpEx, there's a recent 8-K kind of detailing some of the October restructuring charges that you're going to incur. But was there any update to sort of the 24 OpEx that you had talked about last quarter? Or is this just more that 8-K just kind of detailing what you had previously incorporated into guidance?
Yeah, it's really the latter. I mean, we did announce this on our earnings call. We had taken the action. We had done restructuring, reduction across the entire company. There was some actions that were taken in Europe specifically, which is a higher cost region, but it's also a region where you're having to negotiate with work councils. It takes a little while to work through that. So we highlighted some cash and non-cash charges associated with the RIF that we took in October. With regard to the outlook for 2024 OpEx, no, no change. I mean, this doesn't reflect a change at all. You know, with this reduction, I mean, we do expect to see OpEx come down in 2024, you know, on the order of $290 million. They didn't give us an exact number, but-
Okay.
That's the-
But that was sort of incorporated into your thinking-
That's correct.
Last year.
It was.
Okay.
It was.
Okay. It sounds like, you know, if I was gonna ask you to rank order the growth segments in 2024, sounds like infrastructure is probably at the high end, and it's just a question of when inventory burns off and connectivity and broadband and industrial multi-market probably a slower grower.... Is that the right way to think about the relative allocation?
Absolutely. I mean, I think infrastructure by far is the bigger grower. Now, you know, we get through enough inventory. I mean, naturally, broadband and connectivity both could react quicker.
Yeah, I mean, they're obviously down a lot.
That's right.
If they come back, they can come back-
That's exactly right.
Yeah, yeah. I mean, there could be some-
We've also got some, you know, kind of year-over-year comps that are a bit challenging. Q1 of last year was quite strong.
Right.
If I think about what last year looked like, just kind of the shape of it, it was kind of declining throughout the year, and clearly this year I think it'll look just the mirror image of that.
Yeah, you come up against some much easier comps-
That's right
... beginning in Q2.
That's right.
Yeah. Lastly, now that the SIMO transaction's been terminated, what are your priorities for cash generation? Are you gonna try to accumulate cash? Are you thinking about other M&A opportunities-
Yeah
... and just what are you doing?
So, I mean, I'd say right now we're just kind of heads down, you know, looking to generate cash. We have this restructuring, there's some cash costs to that. Revenues are down, so we're just very, very focused on profitability. We did the cuts that I had mentioned. I think we're frankly kind of encouraged by some of the knobs that we've been able to turn. We have, you know, you can really make some dramatic changes in the business in these downturns, and I think we're really excited about that as we come out the other side, so that's encouraging. We still have a little bit of debt. You know, the debt cost is a little high, but yeah, ultimately we'll get back to kind of looking at some smaller tuck-in acquisitions, I think, in the not-too-distant future.
I mean, nothing imminent, of course, but that's been a big part of the strategy, and it will continue to be.
Yeah. Hasn't changed in the long-term strategy.
Correct. Correct.
Got it. Okay. You know, before we conclude, are there any closing thoughts you'd like to share with the group, Kishore or Steve?
I think, you know, you can only control what you can, right? But what we can control is superior execution, and I think the integration of the Intel team acquisition has finally closed in a non-remote scenario. I feel very, very good that all the products we worked on, invested heavily on, are really coming to fruition, have come to fruition. They're sampling, and I think it's a matter of now really bedding down the customers and getting there, and I feel pretty confident that the team is really executing very, very well, and we should do quite well. That's my-
Perfect. All right.
Great.
Kishore, Steve, thank you very much for joining us at the Needham Conference.
Great.
We really appreciate it.
Thank you, guys.
Thank you, everyone.