Broadcom Inc. (AVGO) Earnings Call Transcript & Summary

Hey, guys. This is Chris Danely, your friendly neighborhood semi analyst from Citi. I guess it wouldn't be a Citi Tech Conference without some huge IT problem. My Internet connection is down. But the most important part is we have Hock Tan here, the CEO of Broadcom. Hock, can you hear me okay? I'm just coming in over my iPhone.

You're using iPhone. Yes, I'm fine, Chris. Too bad, I can't see you.

Great. Well, let's go ahead.

But that's okay.

Yes. Well, I'm pretty small anyway. So let's go ahead and get started. I think you guys were the most recent semiconductor company to report. So if you could just give a brief sort of summary of how the quarter was and the outlook, and then we'll move on from there.

Sure. I mean, we reported our fiscal Q3 ending -- ended, I would say, end of July, and we're now halfway through Q4. And what we did was -- I mean, we reported pretty much what we expect to see, which is year-on-year top line growth -- revenue growth that is of 16%, and that comprises semiconductor growing at 19% year-on-year, very strong semiconductor for semiconductors. And on our infrastructure software side, we're growing 6% year-on-year and that was pretty much what we expect. One is, of course, also what we expected and -- but nonetheless, very pleased to see is that EBITDA margin -- well, let's start with operating profit year-on-year grew on revenue of 16%, 24% operating profit growth. And our EBITDA margin reached -- north of 60% revenues for the quarter. For Q4, we see more of the same, in fact, very, very strong demand in terms of bookings and a fairly disciplined way we have of shipping our products. But with Q4, we see the -- particularly with the seasonal ramp up of our North American OEMs, we're guiding up towards from $6.8 billion of revenue in Q3 sequentially, up to $7.35 billion of revenues. So a very strong sequential step up and again, year-on-year in the mid-teens growth. And basically, what we see is fiscal -- that will be the last quarter of fiscal '21, and we pretty much came in where we began the year in terms of what we expect to see.

Great. All sunny skies over at Broadcom. I have asked this of your peers and contemporaries like Rich Templeton, and I'd just be interested in your opinion. Do you think that this is the strongest upturn in semis ever? Is this the strongest, I guess, business conditions for Broadcom that you've ever had?

It is -- I don't know I can think back that far to figure it out, but it's probably -- if not the strongest, it's definitely one of the strongest point in time. And I guess what this translates to is are we in a super cycle upcycle? And I guess, I have to say, yes, we certainly are.

Yes. And then I'm already getting some questions coming in from the audience. How about shortages? How are shortages impacting Broadcom?

Well, I don't consider those shortages as much as -- frankly, one way to look at it on the semiconductor side, just on semiconductor side, demand in the form of bookings showing up, if I compare year-on-year, we're up about year-on-year, and if I take a 13-week running average of bookings or orders coming in, we grew -- we are growing roughly 40% to 50% year-on-year on demand bookings. And as you've seen, we are shipping out -- we are pushing our shipments at a rate of around 20%. So if we look literally at that kind of demand of bookings versus revenue, sure, there's a difference. But having said that, we look at a time horizon, a time factor to the demand and that 50% year-on-year increase in demand in the form bookings is actually over a more extended time frame than I think people -- than customers coming in are truly asking. In other words, what I'm saying is not all 50% booking demand are real. These are people. They need it, but they don't need it in the current quarter, they probably needed the quarter after or the quarter or even 2 quarters further away. So if you time base it, my view is true demand is around 20% year-on-year.

And Hock, 2 trends that are driving your business are the 5G upgrade cycle and the enterprise recovery. I guess the most recent one is the enterprise recovery so we can start there. You mentioned that in the most recent conference call and also the one previous to that. Would you say that this recovery in enterprise demand is about in line with your expectations, a little better than your expectations or a little slower? And then maybe talk about the trends behind the enterprise recovery and how you expect it to play out?

Well, I guess the strength of bookings demand -- back to booking demand is, obviously, as I indicated earlier, very, very strong. But then one has to put into context perspective over the last, say, couple of years. In 2019, we start back then, we were -- almost I would say the bottom of a semiconductor cycle, down cycle. 2019 was a recession period for semiconductors, and the large part of that was enterprise demand was very weak at that point. Just starting to recover in early calendar '20 when COVID-19 hit, which obviously, with the lockdown environment, enterprise demand quietened down. And I would say enterprise did not spend much in 2020 and only started to step up in calendar '21. And I would even say really stepped up in -- over the last 3, 4 months in terms of their demand. Prior to that, for the last 12 months, at least through COVID-19, demand and shipments were driven by hypercloud and followed shortly thereafter by the service providers and telcos when they pushed for broadband and starting to upgrade 5G backhaul networks. It's only 3 months or so ago that enterprise started recovering in their spending and starting placing large orders with us. And that's continuing on today. And we see revenue shipments, I would say, out of those demand will drive back half of '21 until '22.

And would you say that this is being driven, Hock, by the, I guess, the end of work-from-home-and folks going back to the office? Or is there something else going on out there as well?

I think it's a very [ hot ] trend here in the sense that, as I pointed out, there's been pent-up spend demand out of a lack of spending in '19 and '20 over 2 years of lack of enterprise spending. I think there's a catch-up going on. And exacerbating -- exaggerating it, I would say, is also that spending has been particularly weak in 2020 because of work-from-home environment. And as the economy recovers, businesses opened up, We're seeing that in '21 with a vengeance, I guess.

And if you had to estimate what percentage of your revenue is "driven by the enterprise," what would that number be?

Well, enterprise represents some 35%, 40% of my revenues, of semiconductor revenues, I would say, take aside infrastructure software. Infrastructure software, it's all enterprise, as you know, but they are on long-term 3-year contracts in a fairly subscription manner. So that's managed -- that's also driven separately. But on semiconductors, about 35%, 40% is pretty much enterprise.

Yes, that was going to be my next question is, how is the sort of recovery in enterprise impacting your 2 businesses differently, software versus semis?

Well, on semis, we will see, as we are announcing continuing strength in entire semi demand, initially with hypercloud, then a service provider and for probably '22, most of '22, we'll see the big driver of growth as enterprise spending recovery, very sharp increase. And that enterprise will kind of drive the double-digit growth we have been seeing in '21, just for enterprise. Now in infrastructure software, however, this tends to be 3-year contracts with the largest group of enterprise customers out there. And we recognize revenue on a ratable basis, and it comes off a renewal and there will be more capacity usage, obviously, as enterprise opens some and recovers. But again, it's all spread over 3 years. And so the growth rate here is much more modulated to probably a mid -- around 5%, 6% year-on-year growth rate on infrastructure software, just partly the way the subscription is written and the way the revenue is spread over 3 years.

Yes. Well, I guess you don't have to worry about shortages in the software business. On the semiconductor business, we've seen shortages out there. As the enterprise recovery continues to ramp, how well do you feel that Broadcom is prepared from a product and inventory perspective? Could there be any extension in lead times or any shortages there? Or do you feel like you're in pretty good shape to sell this increased demand?

We have kept our lead times on products pretty stable now for 5 months, and we didn't feel we have the need to extend it. So I think things are relatively much more stable and much more predictable, but lead times are still rather extended. Having said that, we have kept it stable, but very extended.

Yes, you're not alone in that quandary. You mentioned hyperscaler demand earlier in the conversation. There are some concerns that as enterprise recovers, we might see some weakening or softening in demand from the hyperscalers or the cloud service providers. Can you speak to that? How is the tone of that business with you?

We have not seen any softening of demand from hyperscalers even as enterprise demand picks up over the last 3 months. So I'm hoping that it will stay at these elevated levels. It may not grow much more from where it is, but we have not seen and I do not expect to see a reduction in demand.

Yes. So let's switch gears and talk about the other big trend, which is 5G. I guess, first of all, maybe let's dig into the wireless business overall. This business was more or less was put up for sale and now you're keeping it. Maybe talk about your sort of long-term thoughts on the wireless business and then we'll jump into the 5G trends because, obviously, that stretches into some of your other businesses as well.

Sure. Well, this -- I mean, make no mistake, the 3 product lines, 3 product divisions we have in our wireless business, which is really more consumer driven and is very limited to just a couple of large customers in very specific product franchises, in this case, 3 that I articulated. Each of those product lines, again, we sell -- as we sell all our other product lines, only on technology. I mean we got the best products, we have the best technology, and we execute the best. And that's what drives this business. So in many ways, in terms of product technology characteristics, it means it fulfills the requirements we have as a core franchise. The key is sustainability of demand. And as we articulated before -- or announced before, for us, it really validated the long-term sustainable nature of these particular franchises. When we actually step up and sign a long-term arrangement, long-term supply and technology enablement agreements with our largest customers, it kind of validates the fact that these are sustainable franchise. And they have been for the last 10 years, almost 10 years, and we see that continuing into the future industry areas that we have. Now 2021 is rather unusual because COVID-19 created a launch of the last generation 2020 phone model. And I expect, we all here today, 2021 is on schedule. So the year 2021 has unusual aspect of having 2 peak seasonalities in 1 year, which probably leads to a higher -- an additional tailwind in terms of year-on-year comparison on revenues growth. So the question is what '22 will bring. If '22 reverts back to a normal pattern before COVID-19, then that's a headwind. If the number of phones continue pretty much in '22 as it has run in '21, then we have a continued, I would say, sustaining at a high level of growth.

Great. Well, even though it's "only 3 customers," Hock, those are 3 very good customers to have. How do you see -- some of your customers are talking about potentially going internal for some of these products? I mean do you think that's going to happen eventually? Do you think that it's going to be way out in the future or not going to happen? And how does Broadcom sort of deal with that or combat that?

No. We have seen that not just this lately out in the news. This is something that is inherent in the business model we work on. You're talking about not just some of these big phone OEMs, you can even talk about some of the large hypercloud guys because -- and what is more overriding than anything else is each of these guys have the scale to consider creating silicon -- having silicon created for them that meets very specific needs, whether in the case of cloud, it's workload needs. In the case of phone OEMs, it's particular features that they would use to differentiate their products. And in the case of cloud, that makes their workload, run more effectively, whatever it is. They're looking for hardware engines that can address those specific needs. And frankly, if you can get it outside from the marketplace, why do it yourself? Because it tends to be a painful exercise in trying to do things yourself. It's not just painful, it's expensive because you stand it one generation, keep it going, as these things go, multiple generations. You keep investing. It's not -- it's literally you get on treadmill. And so my sense of it always is if they can get what they need externally, they don't need to do it internally, and we've seen that happen over the years many times. And frankly, for many of these large customers, when we sell them specific franchise products on our site, most of all these products are somewhat customized. They are in -- some more so than others. But they are. These guys are big enough to ask for specific tailored features and we're happy to provide it to them because they have the scale. From our point of view, we get the ROI to do it. So we'll do it. And if we keep doing it and meet what they need, my frank view is they don't need to go outside. It's not worth it to buy the cow when you can have the milk, so to speak.

Interesting analogy as always from Hock Tan. One last question on wireless, as far as the 5G upgrade cycle, Hock, where do you see us in this cycle? Are we halfway through it? Is it a 1/4 of the way, 3/4 of the way? I just appreciate your thoughts on kind of where we are in this upgrade cycle.

Are you talking about phones? Or are you talking about the infrastructure, Chris?

Well, yes, since we're on wireless -- sorry, since we're on wireless, phones. And then I thought we go to infrastructure after this question.

On the phone side? I don't know. I think we've just seen pretty much the first 2 -- first one or 2 generations of fully 5G phone. So I think we're still relatively early in that whole process because I would imagine 5G is going to run for 5, 6, 7 years. If you look back at 4G. It ran for 8 years. But drilling in, as you say, drilling every year or every couple of years, we get more and more add-on features, more and more bells and whistles, attach on, but they all basically are 4G phone -- 4G technology. And on 5G, we expect to see the same more bands, more capabilities, more features that added on, and they'll all be derivatives of the underlying basic platform of 5G phones. And so I think we're still early in that whole 5G cycle.

Sure. And then if you could just take the second part of the question, talk about infrastructure and then I'll transition us to the key ASSP business.

Well, yes, on infrastructure, which is also whether it's the fronthaul, the radios, the RAN, so to speak and as well as the backhaul, which comes after that to -- which are the networks that holds this high traffic from the front out to the Internet to be used to the core, I think that's even early in that cycle. Now we're seeing that happen obviously simultaneously as more and more base station or RAN gets put up there, they are 5G capable. And we have seen that start to happen. And we are also simultaneously seeing, as we did the last 3 months, 4 months, a lot of large operators expand, upgrade their backhaul networks, which is really Edge, core, metro networks to handle expected high amount of traffic. And when that -- what I mean by is we see that -- we see a lot of demand for our merchant routers, especially at Edge, metro and core. That pulls the backhaul of wireless infrastructure.

Okay. Another question I've gotten from the customers listening on the line is your, I guess, scaling your ASIC business, the hyperscale opportunity with all these alternatives to X86 coming on out there, all these ASICs, whether they're ARM-derived or anything else, how is that business trending for you guys? And then what are your sort of long-term expectations there?

Well, we have been doing this ASIC business as a core product division for 15 years and probably longer. And it started with largely networking, and we've seen it evolve as merchant silicon in networking increasingly displaced ASIC that sits with OEMs. But having said that, filling in that blank, filling in that space have been more and more ASIC silicon, special purpose silicon for hypercloud, particularly. And we have -- we have indicated things in machine learning that we've gotten quite a bit of traction. We have also started doing -- we have also been doing ASICs in what people call compute offloading where instead of using general-purpose CPU in service, in the compute side of a data center in hypercloud to offload some of these specialized workloads, whether it's for virtualization, orchestration, security, encryption on and out towards the edge of the server to handle that. And we have our fair share of that in our ASIC business. And it's a business that is sustaining very well as more and more of this hypercloud to build -- decide to do their own machine learning chips. More of them are going on and new application coming, for instance, in video transcoding, where you see the likes of YouTube, TikTok and even Facebook site, why they want to have better video, enterprise-grade video for their subscribers and start to run video workloads on specialized silicon as opposed to trying to run it on software on general purpose CPU. I call that all-compute offloading. And we are very much part and parcel of that initiative over the last, I would call it, 5, 6 years now among the hypercloud guys, pushing it. Now as far as are we looking at trying to do ARM core general purpose CPU in those kind of environment, up to now, we have not. We stayed away from that.

Got it. And then as part of all these growth opportunities, we hear all about foundry pricing going up and issues with supply. How do you feel about your ability to procure wafers? And is pricing doing anything to you reasonably passing that along to customers?

Yes. I think we're in an environment coupled with the fact that all our products are very mission-critical products with various sole-source characteristics, we'll pass it on to our customers. And we have been doing that, frankly, for the past 12 months as the supply chain in semiconductors, depending on various areas, various parts of supply chain, have step-up costs, have reduced what would have been the cost reduction road map. All these have increased our cost and we have been very, very able to pass it on to all our customers, and we'll continue to do the same.

Great. Shifting gears a little bit because we just have a few minutes left. I'd be remiss if we didn't touch on the software business. I had a couple of people basically e-mail you the same flavor of question to ask you, so I'll ask you first. What exactly is the synergy between the software and the semiconductor businesses?

Well, I tell you where we are benefiting right now in these synergies. When we -- to sell infrastructure software, this software are all sold to, used by -- adopted by the largest 1,000 enterprises, traditional enterprises in the world out there. This -- there's a footprint of our 5, 6 products in infrastructure software that we have sold to this same 1,000 guys largely. To cover these guys very, very well, we have a very large enterprise sales organization directly covering, supporting and selling -- cross-selling, I would say, of 6 different products to those 1,000 customers. And who happens to have the largest IT budgets around. This also happens to be a big part of the enterprise demand out there for IT infrastructure. Their service on hardware site, system site by the OEMs, the likes of Cisco, Dell EMC, HPE, IBM. They also happen to be. And in our semiconductor space, when we talk about enterprise, we really sell our chips to the OEMs who in turn system integrate it, put in software, box surrounding and sell it to the same thousand and more end users who are enterprises. So when we get all this huge demand from the OEMs, one of the first things we do is triangulate it against what our enterprise sales guys are telling us when they cover these 1,000 guys. One example is suppose a big bank wants to build a big data center in New Jersey of $100 million, $200 million with a whole suite products. Before they even do an RFI, 2 or 3 of those OEMs would already have been in the running, have knowledge of this, what is needed and what -- in this environment of long lead time, be coming to us from a semiconductor requirements point of view and placing orders for the same project. And this way, we get to know.

Right. I guess we have time for one more question, Hock. Just on M&A, it seems like we've been waiting for some sort of acquisition on the software front. Are we just waiting for Godot? Or do you see something out there? Or could you possibly do something in semis? Maybe talk a little bit about that.

Well, you're right. Never say never in semis because there are still some opportunity in semi, not many that are very -- that could be of interest. But you're right, more opportunities in software. But even then in this environment, I guess a better way describe it is we're getting very selective. Part of it is when you're growing as a whole company, 16% a year, the compelling need to annually increase our -- not just top line, but our EBITDA double digits is perhaps less compelling, which allow us to be much more selective in what we pick. And this particularly comes about when you look at how our earnings have grown over the last couple of years. Right now, our free cash flow is $13 billion to $14 billion a year. And if you put that against our market capitalization, you're talking about cash -- free cash flow yield on our market capitalization, our stock price, so to speak, of close to 7% today. Now you know our criteria for selection, our financial criteria for selecting the acquisition, other than our other criteria or market leadership and all that, is we've asked for a 10% cash-on-cash return on what we pay out. And we're beginning to look at this 3% premium as really not necessarily sufficiently risk adjusted. I mean we could buy our own shares without second thought and get an implied 7% return on cash versus buying someone else at 10%. So probably, we have adjusted our selective hurdle rate to probably closer to 12%, 13% now, which basically makes the selection of a target more challenging. But we'll keep looking. And I will put it on record, we're looking. Just that we're more selective, and we haven't found anything yet.

Great. I think we're out of time. Thanks again, Hock, for joining. Sorry, everybody that we had the technical difficulties. But hopefully, the rest of the conference goes well. Thanks again, Hock. Have a good one.

Good. Thanks. Thanks for being here.