Okay, we'll go ahead and get started. Welcome, everybody, to the 27th Annual Needham & Company Growth Conference. My name is Quinn Bolton, and I'm the semiconductor analyst for Needham . It's my pleasure to host this fireside chat with Synaptics, founded in 1986. Synaptics is leading the charge in AI at the edge, bringing AI closer to end users and transforming how people engage with intelligent, connected devices, whether at home, at work, or on the move. The company powers the future with its cutting-edge Astra AI native embedded compute, Veros wireless connectivity, and multimodal sensing solutions. Joining me on stage from the company are Michael Hurlston, President and CEO, and Munjal Shah, VP and Head of Investor Relations. Michael, Munjal, thank you for joining us at the Needham Conference.
Quinn, good to see you. You're looking very dapper today, I must say. Very dapper.
There's a lot of caffeine here. A lot of caffeine. Okay. That's the secret. That's the secret. So maybe, Michael, just starting off with some background. Synaptics, as I mentioned, founded in 1986. Initial business focused on touchpad and touchscreen applications for PC and consumer electronics. Can you walk through the company's transformation that began in 2017 towards a more IoT-focused business?
Yeah, I mean, when I joined the company, I joined in 2019, so a bit of it was underway. But when I joined the company, we were very heavily concentrated on mobile phone. And in the mobile phone sector, really one customer that was Apple. And we actually knew that the Apple revenue at the time I joined the company was going away. So Apple, our primary product there, was attached to their LCD phones, their older phones, and they were in the process of transitioning to OLED. So we had a pretty big hole that we had to dig out of. And I guess necessity is the mother of invention sometimes. And we said, okay, let's figure out how to dig out of that hole. And we essentially filled $1 billion for a $1.5 billion or $1.2 billion company.
$1 billion is a pretty big chunk of revenue. And we figured out how to backfill that with more IoT-centric products. We did a bunch of acquisitions to bring us wireless connectivity, some processing capability to build what we think is a future franchise in edge IoT processing and connectivity. And in the process, we're able to kind of dig ourselves out of a pretty appreciable revenue hole that was developing in the face of this Apple piece.
Perfect. Last week, obviously, Consumer Electronics Show. Ahead of that show, you guys had, I think, a pretty big announcement with Google. You announced that you will integrate Google's MLIR-compliant machine learning core onto the Astra platform to accelerate the development of AI devices. Let's unpack this a little bit. Can you first tell us about the machine learning core that you will be integrating from Google? The price performance, how scalable is it?
Yeah, I mean, I think Google went through a pretty exhaustive process. I don't know all the details, but our estimate is that they kind of did a bake-off between six or seven microcontroller suppliers and found our solution to be the best in terms of inferencing, which is the measure of when we talk a lot about this machine learning, you're talking about the number of inferences. And we have the best power per inference in the industry. It was head and shoulders above anybody else. I mean, we're a relative unknown in microprocessors and microcontrollers. And for us to come out ahead, we had to be a lot ahead. And we clearly were.
To your question, Quinn, right, what ends up happening is we are now going to incorporate, on top of this truly differentiated product, where our own neural network will continue to carry the bulk of the load, we will add the Google neural network. Now, what does that do for us? Number one, it's a great validation point. It validates our technology.
As I said, we were in a pretty significant bake-off and ended up head and shoulders above many of our competitors. But number two, what I do like about the Google system, what they're trying to do for the IoT is largely what they tried to do for mobile phone, which is democratize it a bit, right? The Android operating system is what we all know. They are having a different operating system that will go with the IoT.
They are going to open source a toolchain and compiler. The compiler is really essential as we think about these models that run on these edge devices. Having an open source compiler that can take those models and run on a microprocessor is essential. All of that is being open sourced. We will be the platform of choice on which all of this can run. We think, one, it opens up an opportunity at Google. Google has lots of home devices that can benefit from this. But probably more importantly, it's the whole ecosystem. It's all the rest of the people that Google does such a good job of enabling that we think we can now take a lead position on.
Any customer using this open source compiler, that compiler is likely going to be compiling that code, putting that model onto the Astra platform, because today you're really the only hardware partner that's been announced for this platform.
Yeah, you have it right, Quinn. So you think about garage door openers. You think about home automation equipment, right? These folks don't have the software capability, the ability to go compile any kind of software onto a microcontroller. Google is enabling that by open sourcing it. And we're sort of the beneficiary on the other side.
Perfect. Since you'll be integrating the core from Google, can you say whether there's any licensing or fees that you'll pay as part of this? Or is it truly sort of a partnership where they provide the core, you'll be the hardware partner, and you monetize it simply through the device shipments?
Yeah, it's more of the latter. In fact, there may be some relationship where actually Google's paying us some money to help get this to market faster, right, to really accelerate the time to market. They want to see this thing hit market very, very quickly. And certainly from a resource standpoint and others, they're going to contribute to accelerating that project and product. But money certainly won't flow the other way. We're paying them no licensing fees or anything of that nature.
Perfect. And then any thoughts on sort of when you may be sampling this product to customers? And I assume it's probably just with anything in microcontroller, probably a year or two plus to kind of time to revenue. But any thoughts on samples and revenue?
Yeah, you have it right. You're not the best semiconductor analyst without reason. So that's exactly the right timeline. I think we'd expect to see production sometime in our fiscal 2027. We've talked about Astra really going into production, as you know better than most, sort of end of our fiscal 2026, which we're offset by two quarters. So it's mid-year 2026. And then we'd expect that this follow-on with integrating the Google core, the MLIR core, is probably end of 2026 into early calendar 2027.
Great. I want to step back now, and we're trying to ask a lot of our management teams here at the conference to sort of general or industry questions. The first is, with the Trump administration coming in next week, to the extent that they do raise tariffs, do you see any direct impact on the business? I understand that there may be indirect impacts if the products your chips are incorporated into have higher prices, but any direct impact on the business from a higher tariff?
Yeah, I think the simple answer is no. I mean, we wouldn't expect our chips to be tariffed. You alluded to the right point, and that is, I would expect end products into which we are designed, ASPs, to go up, and that probably has some impact on demand, but in terms of direct impact, we don't see any.
Okay, perfect. The other question is just, I know the entire industry has been battling excess inventory for probably a year or two now. Are there any parts of your business where you still do see some lingering effects of excess inventory?
Not really. In simple terms, I think we were first in, and you and I had conversations probably two years ago now, where we were starting to see a lot of inventory buildup as a smaller billion-dollar-ish semiconductor supplier. I think we were more subject to customers pushing us to ship when we understood the demand on the other side was weaker. So we built up a fairly appreciable inventory position that corrected off faster than most. Right now, I would say we do not see inventory anywhere in the channel. And we're seeing pockets now of short lead time orders where we're actually chasing business for the first time in probably six or eight quarters.
That sounds encouraging. I wanted to now pivot to the Core IoT funnel. Obviously, Core IoT, as you talked about earlier, kind of the growth engine now of the business. You've set sort of a longer-term growth expectation for 25%-30%. That funnel last year grew about 30% from about $2.2 billion-$3 billion. Can you just sort of talk about what makes up this funnel? And how does it split between sort of wireless and processor-based solutions?
Yeah, so our Core IoT, I mean, really where we want to go as a company is being the leader in edge IoT devices, right? And we think we have all the ingredients to do that. The sensor piece, any IoT device has a sensor on it, as you well know. And we have touch sensors. We have audio sensors. We have video sensors. We have a lot of different sensors in our portfolio. We don't talk about that as much. Core IoT, for us, is made up of the processor and the connect piece of an edge IoT system. The processor, we just talked about with the Google announcement, we uniquely can take machine learning and run that on the edge rather than the data center. And then our wireless connectivity has been kind of the star part of our portfolio.
So to your question, if you think about how our funnel is growing, it's largely driven off of wireless today. So our growth drivers for this year are largely around our wireless portfolio. It's done phenomenally well after seeing a big inventory correction. We've kind of come out of that inventory correction and seen decent growth as we come back to normal shipping levels, layering in on that some new design wins that we've talked about with Bose, with GoPro, with some really big names that we're happy to add. The second thing that happens this year, as you know, is our Wi-Fi 7 for IoT starts shipping sort of mid-calendar year. And then we layer in this broad markets opportunity, which we're super excited about, which opens up a whole new segment of the market, more of a data center-centric piece of the market.
The funnel growth, to your question, is largely driven off of wireless. But in the early, I think we gave a number on our last call of the Astra funnel was $300 million. Based on what I saw at CES, that number should grow very, very significantly. Again, earlier in the funnel, it's a lot of opportunity rather than design in and design win. But that customer conversation, that customer engagement has really started on the processor piece. And we would expect that to pull through appreciably over the next four to six quarters.
Perfect. In the Core IoT business, I wanted to start just the processors you ship today, I think more sort of microprocessors than microcontrollers. Can you talk about what end markets and applications are those processors addressing today before sort of Astra ramps?
Yeah, again, a good question. We are, as I said, largely a newcomer to processors. And when I say that, I mean more general purpose, MCUs, microcontrollers, and MPU, microprocessors. What we do have are very narrow targeted processors today, one of which an MPU ships into set-top boxes. It's a video conversion IC that runs the heart of a video set-top box. And that business actually for us has done very well. We've been able to take share. It's obviously a market that arguably is declining. It's not the most exciting market in the world. But we've been able to grow share significantly against a large U.S. supplier and also against a Chinese supplier using some of the geopolitics to our advantage. The second class of processor is narrowly targeted actually to, again, not the sexiest market in the world, but to desktop phones.
So there we compete again with the same large U.S. supplier, a trillion-dollar market cap company that had previously had the majority market share. We've now taken almost all of that market. So it's been a market share gain story for us. And that's been a good one. Our MCU, our microcontroller, is actually, again, very narrowly targeted at a use case where we are going into this human presence detection for PCs.
So it's where we've been able to get a lot of our advantage, for the Google relationship, is coming out of this development that really started as this human presence detector that goes largely into the Dell laptops, where if somebody is sitting in front of the PC, it's fully powered. They step away, look away, whatever it might be. The PC actually goes into a power-saving mode. So that relationship with Dell really was the precursor to a broader AI-based microcontroller that's turned out to be a real winner in the portfolio.
Perfect. You talked about some of the strength in Wi-Fi in answer to the funnel question. But your momentum, I think, there has been pretty impressive. I think on the last core call, you talked about doubling your Wi-Fi design wins. You're just starting to ship Wi-Fi 7 this year. What are you most excited about in the Wi-Fi business?
Yeah, there's three legs to our story at the moment. We've seen a lot of the growth that you just cited and correctly has been on the back of our traditional high-performance video centric business. These are applications like security cameras, like the set-top boxes we just talked about, in some cases drones, anything that's moving video over a wireless link, and we've seen that business sort of return to normal state. Our design wins are layering on top of that, and that business is all of the growth. Anything that you see is coming from that high-performance segment. We layer in now Wi-Fi 7. You asked about that, and the Wi-Fi 7 helps us in that core business raise ASPs by $1-$2, right? It's a nice ASP kicker.
And we'd expect 10%-20% of our existing SKUs to migrate over to Wi-Fi 7 this year in a fairly short order. But more significantly, it opens automotive to us. So we've never been able to participate in the automotive market. There's one supplier, a large supplier down in San Diego that has a huge bundle that they pull together. And they have done very, very well in the automotive market. We won't be able to attack that. But for the other 50%, based on NVIDIA, TechWell, any of these other homegrown processors, they need a Wi-Fi and GPS solution. We think we can provide that. So the automotive market is open to us for the first time by virtue of Wi-Fi 7. The last piece is what we call broad market. And you're familiar with that. But it's a data-centric connectivity solution.
Moving out of this video and now connecting little streams of data that need to go over a Wi-Fi link. Think of home automation, industrial, toys. There's a ton of things, even wearables, the watches and things like that have Wi-Fi in it that is data-centric. We've not really had a part that could compete there on price, on performance, on power. We've introduced a new device that we think can attack that market. In so doing, open up another three to four billion of TAM that we think we really think we have a running start into given the performance and power and cost that we can deliver to that market.
You've mentioned that the ASP uplift for Wi-Fi 7, a dollar or two higher ASP. Can you talk about the margin benefit for that? And then as you look at Wi-Fi 7, how long do you expect that sort of premium ASP or margin to continue? Does that tend to fade pretty quickly? Do you think you can hold on to those margins for a period of time?
Yeah, and it's obviously very good from a margin perspective. If we lay out our margin profile, generally speaking, as you know, our Core IoT, our target as a company is 57% today. We're operating 53%, 54% gross margin. The Core IoT, in general, is kind of right in that zip code. The wireless is generally a little bit below. And Wi-Fi 7 is a big uplift. I mean, Wi-Fi 7, we think, is high 50%, 60% gross margin. What's unusual in normal cases, the scenario you're hinting at is correct, where we would expect a pretty quick reversion to the mean. For some reason, it seems like we're far enough ahead of competition here that we have at least a year of greenfield, maybe more, before we'd see some of the traditional IoT competitors coming into this landscape. As such, in, I would say, a rather unusual set of circumstances, we probably have a longer runway than normal with this one.
Perfect. I wanted to move to sort of the broad markets opportunity. I think it was in December you announced your first Bluetooth broad markets Bluetooth Thread device that supports Bluetooth 5.4, Zigbee, Thread. Can you tell us about that solution and how has customer feedback been on that device?
I mean, a couple of different things. We announced it as Bluetooth standalone, right? So kind of the third segment, we talked about high performance. We talked a bit about this broad market. Bluetooth is the third leg of the stool where we've really had no presence, and in December, we did announce a Bluetooth standalone chip as our first foray into Bluetooth standalone. That chip actually has been very well received. It's a higher performance, as you said, Bluetooth 5.4. It actually operates dual mode, meaning Bluetooth Classic and Bluetooth Low Energy. It's found places in remote controls. It's found places in some sensors, so we like it. We really need to do more there, I think, to capture that opportunity. We've got to get some more chips. We see our competitors out there with five and six different BLE devices.
So we've got to invest more, I think, to really make a dent in that. Just for clarity, when we talk about our broad market initiative, that combines Bluetooth and wireless LAN. So this BLE, Bluetooth wireless LAN, the broad market, the data-centric thing that I was just talking about, we call that broad markets as well. And that's a combination of all the standards you mentioned plus the Wi-Fi as well.
Got it.
No.
Did you have more? OK.
No, that's great.
How big is this broad markets TAM and maybe talk about the go-to-market for that part of the business?
Yeah, it's three to four billion of which we're a zero player. It's a much more fragmented market. You're talking about industrial. You're talking about, as I said, toys is an interesting market. Wearables is an interesting market. Home automation is an interesting market. It is a very, very. It's an equal size TAM to the high performance. And we know all the customers there, the Samsungs, the TV makers. I mean, that's a huge, huge market. This is a much more fragmented market. So even given our advantages in performance, in power, in price, right, when we think we have advantages in all of those, our challenge is the go-to-market because this is not Samsung or Google or Amazon. We're dealing with a whole different class of customer.
What we've built out over the last six months, Quinn, is a fairly significant sales effort, collateral effort, software effort to simplify the product such that it can sell into these broader markets. There's not as many knobs. If we're dealing with an engineering team on the other side, they can typically deal with lots of knobs. And they like that. The more tweakability there is, the better it is. When you're going into this broad market, we want no tweakability. We want to be able to ship a turnkey solution to the extent we can. So we've made this broad markets wireless chip as compartmentalized and as turnkey as possible, number one.
Number two, we've spent money on our go-to-market motion in terms of website, in terms of collateral support, in terms of documentation, and then developed a whole set of relationships with distributors and large channel guys such that our salespeople don't have to go and visit directly 100 unit, 200 unit, 500 unit, 10,000 unit accounts, right? We want to focus our people on the one, two, three more million unit accounts.
Perfect. I wanted to switch now to the enterprise and auto business. Maybe just as we head into 2025, I don't think we've seen a widespread increase yet in IT spending, but maybe spend a minute just talking about your sort of expectations or your outlook for IT budgets, and then maybe sort of wrap in there kind of your thoughts on PC unit growth in 2025 just at the market level.
Yeah, again, a really good question. One, I mean, it's been interesting for us. We were just talking about in our last group session. Munjal has done, I think, a terrific job because if you look at the IT spending, actually IT spending is up, right? And so everyone's like, well, why are your numbers up? Well, the secret, of course, which you know and I'm sure most of the audience knows, is all that's going to AI, right? It's all going to data center. It's all going, whether it's outsourced and in the cloud or you're building some on-prem data center. Our IT guy, who is a terrific human being, is spending all of his money on on-prem data center. So it's hard to tease apart in these IT spending numbers exactly what's going on.
What Munjal has done is been able to isolate sort of what's going to gadgets and what's going to data center. It turns out the gadget, no surprise, is way down, right? The spending on what we care about, PCs, enterprise headsets, enterprise telephony is down fairly significantly. We would hope we don't yet see. We would hope that eventually that has to come back because these gadgets have been kept for longer than normal. We expect a PC refresh cycle typically is three and a half to four years. We're now on year five. These are well beyond sort of a normal usability pattern. Our docking station business, which you know and the audience probably is less familiar with, again, there's been very little in terms of refresh.
So we would expect this year to see some uplift with kind of the rest of the gadget ecosystem. PC, to your question, has kind of normalized a bit. I think PC last year did about what we saw pre-pandemic. I mean, there was this big spike. There was a correction down. Then it kind of came back to this 250, 260, 270 level that we would expect. We don't expect, even though there's all these sort of tailwinds in the PC business, AI PC, ARM-based PC, Windows 11. At least right now, Quinn, we're not forecasting, Munjal, correct me if I'm wrong, any big growth, unit growth in the PC sector.
No, you're right, Michael. I mean, what we're seeing is all the drivers are in place. Everybody's talking about those drivers. But given our lead times, we're not necessarily seeing that. Now, if it does happen, it's a unit play for us. So from a unit perspective, we stand to benefit from it. But we're not calling it at this point. But having said that, we do have some content that is new, like the user presence detection that Michael talked about earlier in the conversation that is incremental for us.
Yeah, I was going to go to the user presence detection in a minute. But I think even in the core enterprise business, on top of whatever the PC unit market may do or IT budgets may do, I think you've talked about the ramp of new products. And maybe that includes user presence detection, but also some share gains. And so I don't know if there are specific areas for share gains or new products outside of user presence that you're expecting to help perhaps provide some growth or just better than market performance in 2025.
Yeah, we would expect to do better than market. Two things. Munjal touched on it. User presence. User presence is really only a Dell phenomenon at the moment, really taking advantage of all the features and functionality. We'll just start shipping this year, and I'd expect the attach rate to be somewhere between 10% and 15% of the Dell laptops. We expect to extend that to both HP and Lenovo. It's a pretty compelling feature that ends up saving battery life to the tune of 25% or 30%, so the PC manufacturers really like it. We would expect that to flow through, but this year, it's a nice incremental gain for us, even at the 10% and 15% attach. We'd expect Dell attach rates to double over the course of time.
And then the other thing for us is we had lost all of the, for some technical reasons, we'd lost all of the Dell fingerprint business. We won't get back to normal market share this year. Generally speaking, we have 50% to 60% share in the fingerprint market. But we will have a big step up at Dell because we go from zero to, let's say, 30%, 40% attach this year, which will be another good incremental adder. So even in the face of flat PC units, those two growth drivers add quite a bit to our PC revenue line.
Have you guys, on the user presence detection , given a sort of dollar content per laptop, is that a single digit, low single digit kind of range, or should we be thinking about kind of?
Yeah, it's $3-$4, right? So it's a decent ASP. Our job is to grow the attach rate, right? As I said, I think Dell is really the company that's taken full advantage of it, but still 10%-15% of their machines. I'd expect that to more than double over the next year, and then we've got to get into HP and Lenovo.
Got it. And competitive landscape and user presence detection, I know some of the small FPGA vendors are targeting this application. What competitive advantages do you have? And it sounds like maybe at least at Dell, you might have displaced some FPGAs with that win.
Yeah, you're right, Quinn. I mean, it is, that small FPGA maker has made a lot of noise. I mean, I think they're doing a lot of things right. So all credit to them. But I think in this particular area, we've basically run them out. So I would expect us to have absolutely the majority share at Dell. And then we're flipping over some of the wins that they have in the other accounts. It comes down to the simple thing. Once you have a purpose-built semiconductor, it's always going to be better from a space, cost, and power standpoint than an FPGA. The FPGA did a great job sort of trailblazing in the market. And they won a lot of nice business. But in the end, a purpose-built SoC is going to do better.
Great. Looking at the outlook for the video interface products, I think a lot of these go into docking station type applications. Anything to call out on that part of the business? I think you had previously seen some excess inventory there. Were we mostly through the inventory digestion for that product line?
Yeah, you're right. So the inventory, I think, was most pronounced there because, frankly, we have very, very high market share in the video interface business, which is primarily, as you correctly called out, going to docking stations. We would expect now a more normalized condition now as inventory is cleared. So from what's been true is even through the pandemic and everything else, modulo inventory, one docking station ships for about every two and a half laptops. And we have not seen the sell in and sell out commensurate with that just because of the inventory condition. We would expect to see that return this year. So our docking station business this fiscal year is probably undershipping that kind of rule of thumb by 40% or so, so a fairly significant amount.
I think as we get through the back half of this fiscal year and into next fiscal year, we should see that normalize and get back to levels that we saw in 2019. And this is a big part of our margin story, as you know well, right? Our mix has shifted away from higher margin parts of the portfolio, enterprise headsets, docking stations, monitors. Even we have solutions that go into printers. These are super high margin products for us that have been particularly challenged. And look, as IT spending gets better, that gets better. As PC units, I think that the constant with the two and a half to one will end up bearing out through this calendar year.
Great. Lastly, in that segment of the business, auto you've been pretty cautious on. And I think some of that's a reflection of just what's going on in market demand, but also sort of a decline in the legacy DDIC solutions. What could turn that business around?
Yeah, we have been cautious. I mean, it's been one of these things where we keep signaling rain is coming, and it hasn't necessarily come. Our numbers have actually held up really well in the automotive business. I still think rain is coming. I mean, it's just too many signals out in the hall here that companies are having difficulty with their automotive business. And eventually, I think it catches up with us. But I mean, we have some good things that are going on. One, more and more production cars are shipping off with our TDDI solution.
And that's been a real tailwind for us that would continue. And then the second is SmartBridge. As we look out, we are still optimistic that we've not talked about any design wins. I think we'll be hopefully in a position to do that here in the near future. But the SmartBridge bundle with the TDDI is really a good one, Quinn. And we think we could offer a solution that's super competitive at really good margin. Won't impact this year, won't impact next year. But in the outlook in the further out time frame, it should be a really good solution.
OK. Just moving maybe to a couple of financial questions. In November, you issued a $450 million convert with 75 basis points coupon, used proceeds and cash from the balance sheet to pay down the term loan debt. I think you also repurchased some shares in that transaction. So maybe tell us about the transaction. And as you look at the balance sheet today, is it just the $400 million of 4% senior notes plus that convert that remain on the balance sheet in terms of the debt?
Mr. Shah, you're the expert on this one. This is tough math.
This is tough math, yeah. No, you're right, Quinn. What we did in November was we took out a $450 million convert. The primary purpose or primary use of that was to take out the $580-ish million term loan. So we reduced our interest expense significantly. And as you mentioned, we bought back shares. So overall, the transaction, very accretive transaction. We also lowered our gross debt. So our gross debt came down. So it's a positive in terms of from a credit perspective as well. And we extend or we changed the curve on a maturity. So a very good transaction for us, and we achieved a lot of things with it. What's left is now $850 million of total debt, $400 of loan, and then $450 of convert. So that's on the balance sheet.
Perfect. With the term loan now repaid, what are the company's highest priorities for cash flow generation going forward?
Yeah, so the priorities haven't changed from a capital allocation perspective. Continue to invest in the business. So Core IoT growth drivers, we talked about that a lot. That is an area of investment for the company. So we will continue to do that in a measured, aggressive way. Where we need to, we will do it. Beyond organic investments, organic growth is a priority. So that has not changed. We continue to look at a lot of things. And so depending on what comes through, we will execute on that with what makes sense in a financially disciplined way and a strategically accretive way. So those are the priorities. And absent those, we don't see the need to build a ton of cash. I mean, that's where we were before we did this transaction. So absent those, we will buy back shares.
OK. I'm going to see here, pause here and see if there are any questions from the audience. We've got a couple of minutes left. Everyone? No? All right. I've got one final question. Just what are you seeing as we kind of come through the downturn on the pricing front? Are folks getting more aggressive? Is pricing not too bad? Any thoughts on the pricing environment?
Yeah, I mean, it's a challenge. We obviously had a big step up, as everybody else did during the pandemic. That step up, for the most part, has held up. I think if you look at our margin profile, it's down 8%, let's say, from 62% to 53%-54%. So 8%. Probably 5% of that is mix, where our high margin products have seen the most, have seen the worst demand conditions. Two to three of it is the price. So we have not seen our input prices go down all that much. We've done some things to work with our suppliers to get some of that worked off. But our price has gone down more.
And it's in our most challenging business. The businesses that have recovered the best, the Wi-Fi, our mobile business, our PC business, that's where the pricing challenges are. In our businesses that are obviously stronger gross margin, video interface, audio, things like that, less pricing pressure, thus much better gross margin. But for whatever reason, they have not come all the way back. And that's what put pressure on our overall gross margin line.
Got it.