The neXt Curve reThink Podcast

The M&M (Marvell and Mavenir) 2024 Events Recap (with Earl Lum)

• Leonard Lee, Earl Lum • Season 6 • Episode 52

Share episode

Send us a text

Earl Lum of EJL Wireless joins Leonard Lee of neXt Curve to recap and to share and discuss the highlights and the key takes from Marvell's Industry Analyst Day 2024 and Mavenir's Analyst Event that capped of an intense year of events. Leonard and Earl cover the telco and networking aspects of Marvell's event while Earl provides what he can from Mavenir's 2-day information-packed analyst forum.

Earl & Leonard discuss the following topic that mattered from M&M Tour 2024:

  • Marvell Industry Analyst Day 2024 Recap (2:09)
  • Marvell - Every networking & interconnect for GenAI supercomputing (5:02)
  • Marvell's telco story for 2024 and beyond (10:12)
  • Why Mavenir doesn't love Leonard Lee anymore? (11:30)
  • The many moving part of Mavenir (12:36)
  • Should Mavenir be designing and making radios? (15:15)
  • Can Mavenir to afford to boil the telco tech ocean? (18:44)
  • Mavenir's telco AI and AI-RAN story (21:33)
  • Thoughts on the state of Open RAN (25:16)

Connect with Earl Lum at www.ejlwireless.com. Hit both of us on LinkedIn and take part in our industry and tech insights. 

Please subscribe to our podcast which will be featured on the neXt Curve YouTube Channel. Check out the audio version on BuzzSprout - https://nextcurvepodcast.buzzsprout.c... - or find us on your favorite Podcast platform.  

Also, subscribe to the neXt Curve research portal at www.next-curve.com for the tech and industry insights that matter.

Earl Lum:

Next curve.

Leonard Lee:

Hey everyone, welcome to, uh, this next Curve Reaping podcast episode where we break down the latest tech and industry events and happenings into the insights that matter. And I'm Leonard Lee, executive Analyst at Next Curve. Um. In this episode, we'll be recapping Mavenir's Analyst Summit. Is it Analyst Summit, or? Analyst

Earl Lum:

Event. We're doing the M& M Tour, remember?

Leonard Lee:

M& M Tour, right, right. Or Bell

Earl Lum:

and Mavenir.

Leonard Lee:

Or, yes, yes. Uh, which took place in Dallas. And, you know, we're gonna, as Earl mentioned, we're gonna touch on The telco related stuff coming out of Marvell's Industry Analyst Day 2024, which took place on the 10th of December in Santa Clara. And since I've already mentioned Earl, I think everyone knows now that I am joined by Earl Lum of EGL Wireless Research, the infamous, notorious, Real,

Earl Lum:

not AI.

Leonard Lee:

Yeah, the actual,

Earl Lum:

actual

Leonard Lee:

Earl Lum. And so how are you doing?

Earl Lum:

I'm doing great. I. Made it through last week, a lot of flying around, a lot of information process for both companies and a lot of cool things that are going on for both companies. Yeah, we can dive right into it.

Leonard Lee:

Yes, but before we dive right into it, remember to like, share and comment on this episode and subscribe to the Rethink podcast here on YouTube and on Buzzsprout. Take us on the road and on your jog and listen to us on your favorite podcast platform, where everywhere just look for X curve rethink. So, yes, let's get started. Why don't we start off with Marvell industry analyst day, 2024. Let's just get that out of the way. Sure you know, obviously, there was a lot of talk about, if not only talk about gen AI supercomputing. For the data center, but let's focus on telco since that's what you and I talk about almost primarily exclusively. What'd you get out of the event this year in terms of telco? Cause they, they went really light with it in the morning and the afternoon sessions, right?

Earl Lum:

Yes, so telco was really during the one on ones and the breakout. Yep. On the afternoon, the whole morning was pretty much like you said, data center and whatever everything that they're doing in data center. And I guess the 1 word takeaway is custom.

Leonard Lee:

Yeah, exactly. Custom and optimized and they. Talked a lot about their interconnect and networking portfolio that spans from literally die to die interconnect or device to, or yeah, die to die interconnect on a server. And then at a rack level, at a data center level, and then. At a data center interconnect level. Right. So you've probably heard, right, I think

Earl Lum:

it went from tray to a rack to a row, to a data center, to multiple data centers.

Leonard Lee:

Yeah. Regions. And, uh, really power being the big, uh, constraint that everyone's trying to deal with. Because it's scarce, it happens to be scarce. And a lot of these. Mega data centers are geared for generative AI supercomputing are so power hungry and consume so much of it that a single data center may not be able to secure the power it needs to actually do what it does. And so now you have sort of these data centers distributed either. Within a municipality, or, regionally, and they're interconnecting all of them with fiber.

Earl Lum:

I think the thing that I learned was, it's not about a custom XPU for the AI engine or whatever is. Everything from the switch to the optics, to the processors, to the HBM is custom.

Leonard Lee:

Yeah, and this is a big game that Nokia is getting into Infinera, right. They happen to have the acquisition of Infinera and Marvell is a big player in this as well, quite obviously. Right. Right. but I think what's important to understand is there are all these layers of networking interconnect. Right? And so you can't talk about all this generative AI supercomputing or supercomputing data center. Connectivity and networking in simple terms, it's literally die to die all the way up to like

Earl Lum:

every connection, copper, optical, electrical, whatever is custom and has to be customized for that application, for that customer, for that data centers, topology, how they do the racks, how they do the rows, trace everything. So it's a very complicated. Scenario that has to be, end to end thought out as to how are you going to do this? And if you start changing things somewhere in the layers, you're going to screw something up, or you're going to need to change a whole bunch of other things that it touches in terms of its sufficiency and optimization.

Leonard Lee:

Yeah. Yeah. So anyways,

Earl Lum:

that was, yeah. So going back to the telecom side, I think. The technology at the end of the day and a shout out to Mark Kummerly and Wolfgang Sauter on their great presentation and a shout out to the JEDEC people fix your trays.

Leonard Lee:

Did he call it stupid trays?

Earl Lum:

Yeah, fix your stupid trays. And the key word from Wolfgang was chackage. It's not a chip, it's not a package, it's a chackage. But I think those are key technologies from a telecom perspective that are going to enable a client like a Samsung or a Nokia with their respective ASICs, or however you want to call them, products to get to that next level. Especially if we're looking at Stuff at the six, seven, eight gigahertz frequency band, where I have a much wider instantaneous bandwidth, or operating bandwidth. HBM might play an important role in that, depending on how I need to do that, right? Because if I'm off chip with all my DDR4s on a board, Is that limiting my bandwidth? And I brought some of that on chip in a very novel way or very customized way. How do I improve my digital front end? A six. How do I improve some of my beamforming capability is my layer one stuff, because all of that's off chip right now.

Leonard Lee:

Wow. Now you're like really thinking

Earl Lum:

about this

Leonard Lee:

box. Yeah. And that was one of the big announcements, the custom HBM, right, was announced. And so. The

Earl Lum:

beachfront property is not just in real real estate silicon real estate or die real estate is just as important.

Leonard Lee:

Yeah. So another shout out to Nigel Alvarez. Hey man. So yes, more HVM. That was one of the mantras that came out of that event. Yeah, it's re it was a really interesting presentation. And I

Earl Lum:

think the IP blocks, right? All of the soft IP that they referred to, everything that could be checked off looked like they had it.

Leonard Lee:

Yeah, which

Earl Lum:

makes it a very compelling foundation for you to start designs. Because you can pick and choose from anything and because they have it all

Leonard Lee:

yeah, and

Earl Lum:

I think that gives them potentially an advantage.

Leonard Lee:

Yeah, and it's like they have a sort of I would call it a shareable or a singular technology roadmap portfolio, and they've centralized their engineering R and D, right? But they're diversifying across a number of different let's call businesses or sectors. Telco being 1 of them. Automotive is another. Enterprise primarily networking. We'll see if they start getting into compute. Well, there's computing there as well, but, and then obviously the data center, all the super computing stuff. But yeah it was interesting to hear them emphasize advanced packaging as much as they did. Um,

Earl Lum:

absolutely. I mean, I think right now packaging is an incredible capability. And if you have it, you're doing fine. If you don't have it now, what do you do to get it? Because that could be the bottleneck for you.

Leonard Lee:

Yeah. And there was a lot of talk about Moore's law, whether it's ending and then what, what's the solution, how do you continue to scale? And that really being one of the levers, you know, there was a lot of talk about 2. 5 three D and 3. 5 D packaging. And so that being this new frontier, and I think you asked the question about thermals, right? Especially right. You're looking at packing all this stuff into bulk, multiple layers on die what are the eventual limits that you're going to face and deal with going forward. So that was an interesting discussion during the, I

Earl Lum:

think the whole interposer discussion on how many can you fit on a 300 millimeter wafer and how many wafers you need for interposers. Compared to just traditional die itself.

Leonard Lee:

Yeah. Yeah. So yeah, a lot of cool stuff. I'm sure you're going to have a report coming out. Any particular telco related stuff? I mean, they didn't mention Octeon even a single time, but when you're in the telco session, Yeah, I

Earl Lum:

think in that one on one session, I think there's certainly some continued opportunities that are still there. they just didn't focus on talking about those opportunities, but certainly the customers that they've had are still there. And still playing in the marketplace. And there's always new generations of chips whether it's a reef shark or whatever for Nokia or custom basics for Samsung, et cetera. So there's going to be another new cycle of technology that's going to be needed as we look at five and a half G or whatever we want to call six G and the six to eight gigahertz spectrum range that will be operating for those 5. 5G and 6G solutions, and they're going to need a lot more capability. And I think that they'll be well positioned. It looks like Marvell will be well positioned to help their customers get the right products that are going to be. The most efficient and highest performance products, performing products for those radios whenever they come to market.

Leonard Lee:

Yeah. Especially considering your HBM comment earlier. So, okay, cool. That's a really interesting. So let's move on. Let's talk about Maveneer's analyst couple of days. Is it summit or?

Earl Lum:

Uh, right. Yes.

Leonard Lee:

Yeah. You went I, analyst event. Yeah. Okay. Analyst. So you, it was

Earl Lum:

a long two days. There was a lot of stuff. It was literally drinking from a fire hose of all the information. A lot of it was shared under NDA. So I can't talk about that. However, there are some generic trends and things that I can discuss. Okay.

Leonard Lee:

Yeah, but you know what? There's one thing I really want to know, okay, is how come they didn't invite me?

Earl Lum:

Don't they love me anymore? You know what? You wore that Texas cowboy hat the best out of all of us several years ago.

Leonard Lee:

Oh, yeah. You look like J. R. Ewing from Come on, I thought John Baker loved me. I thought I was his number one guy. Yeah, I'm a little bit peeved. So, you know, you Mavnir AR team. Um, you know, my feelings are hurt.

Earl Lum:

Yes. Leonard's feelings are hurt. Yeah. However, with that said, I think there's a lot of moving parts with the company right now. There's the core business. Which is all of the software related packet core, IMS core, everything that they talked about before the RCS, multimedia, the voicemail, and then you have some new stuff like existing things that are now getting much more attention like fraud and security. On the software side, because every operator needs that given how many times they've been hacked.

Leonard Lee:

Yeah,

Earl Lum:

all over the world. And then you have the open brand business, which is where the majority of day 1 was focused on was really about. Where is that? Where was it? Where was it supposed to be? And where will it be going forward? And if I look at the technology across the board for every 1 of the product groups that they have for software and for open ran, I think they're solid. I think they have a great platform. The issue is, is that they are. Still trying to get the market moving and they are at the whim of whatever the market wants to do and the timing of whenever the market wants to do it. And as you and I know, telecom doesn't move fast.

Leonard Lee:

Yeah.

Earl Lum:

It takes years. Operator field trials take years. They don't take months. And it's always been about cash. Do you have enough money to last before you get to the end of a field trial with an operator to actually go into production? And do they give you more than a few hundred units? As your reward for the 3 years of continuous testing that they did with your gear in the labs and in the field and doesn't matter who you are. Everyone goes through the same thing. So they're in that kind of boat where they continue to be in POC hell and trials and such. And I think the issue becomes, should they be making radios? Probably not because it's not a high margin business, but I think they're being forced to make the radios to sell the software on the CU and DU. And if they could only sell the software on the CU DU, they'd be in much better shape because of the R& D investments and everything. But it's similar to the early days of Qualcomm, right? Where they had to make CDMA phones. They had to make the base stations. They had to make everything because no one else made it. And I think You know that, and that needs a lot of money to execute on that kind of a plan. And at the end of the day, how much money are you really going to get? It becomes the issue is even if you've got to one or 2 billion in revenues is that enough because of the constant R and D cycle that you have to spend to refresh every single year on all of those products and. Right now, they only have a couple of SKUs on the radios, but every new customer they get, they're going to add more SKUs. So there's going to be a SKU hell, essentially, and it's going to grow exponentially, and that's why I don't think people should make radios. Because unless you have five billion dollars lying around, it's a pain in the butt.

Leonard Lee:

Yeah. Two years ago, though, when we, uh, well, at least when I spoke with John, maybe it was three years ago, the problem was, is there weren't any radios, right? And that's why they made that foray into designing their own radios with, and building them ultimately, because their, the idea was that they would license those designs, right? That was the idea. And so, but basically, From the radio perspective, bringing in that sort of, let's say, less than optimal piece so that they can at least get that stack play going. But then, of course, as we well know, they always continued to evangelize very heavily open interfaces and, leading that March and maybe not focusing as much as, retrospectively, they should have in offering that stat solution and what the, let's take a smaller. Number of partners that could have helped them go to market and, uh, maybe capitalize on some of the opportunities in the last two years. So, um, right. I

Earl Lum:

mean, I, I think right now the issue is I have a great radio.

Leonard Lee:

Yeah. But that,

Earl Lum:

that if I'm at, if I'm at Marvin

Leonard Lee:

dumb idea, do you know what I'm saying? Right. I, I, it was, it was a. It was almost a necessary move.

Earl Lum:

It's a necessary evil that they had to go and then bark down this path. The question now is. If they have a great radio, whether it's open or not, from my perspective, I don't care because the radio is the radio. But it is part of this open initiative and I get what they're trying to do, but from my perspective. I think what they've been able to do on the radio designs, open or not, they should be commended. The question is, is it sustainable financially to continue to come up with 40 new SKUs every year?

Leonard Lee:

Yeah.

Earl Lum:

And dumping 20 percent of your R& D into that or whatever to sustain that. Or at some point, once the market has matured to the point where their stack is good enough on their own, should they divest the radio business to somebody else and go back to what they do really well, which is software.

Leonard Lee:

Right. And apparently services as well, right?

Earl Lum:

Exactly. I think they, their customers love them because they're great at customer service. They are very reactive. They're nimble, better than the traditional incumbent OEMs. And they're winning deals because of that. But the question is, there's so many moving parts to the business right now that they need to focus on. You need a lot of cash to be able to do that. And that's where I think they're starting to come up against the wall in terms of how to continue to gain, continue that momentum and have the amount of cash they need to fund that momentum.

Leonard Lee:

Right. And so what else did they talk about that you can talk about and share?

Earl Lum:

I think that there was a lot of different, um, again, on the fraud and security, that was really very interesting and dear to me because of all of the spam SMS messages that I get on my T Mobile phone. Brutal. That comes every day. Literally every day I have 10 to 20 messages that I can't get rid of. The spam calls. They're mostly going away now or they get filtered or says spam call coming in and maybe some of the spam messaging, but it still shows up and now I'm spending so much time. So I wanted to learn more about what they were doing and they have some very interesting uses of AI. That they've done and but it's really learning about what is the threat and how do they detect that threat, learn about the threat and then create something to block off that threat. And I think the fraud and security. Part for me was a big highlight of day 2 in terms of what they talked about, because there's some pretty cool things that they're doing with their customers. And 1 of them was. Uh, out of Australia that gave a talk about how they're working with that operator. And I think there's a lot of other ones that they're talking to. And there's 1 additional 1 that they should be talking to. If they're not already here in the US about getting some of these. Really innovative and very core to me types of services, which is everything is fraud today and everything is spam today and the IMS stuff and all the other parts, that's very interesting. And I use that, but what I really want is to spend less time worrying about what's actually getting through to my handset. So I can focus on what's really there. That's important. Just like in email, when we have all the spam filters and everything, I think we're lagging on the phone side and there needs to be a lot better technology and a lot more emphasis on that compared to all the other things that, is going on and, there's some new stuff in NTN that they're working on and everyone's trying to do something in NTN in that case. They're not building a satellite, thankfully, it's just their CU and their software stack. So I think there's a lot more opportunity for margin expansion and growth on their telco stack in the NTN world where they don't have to build the radio.

Leonard Lee:

Yeah, I assume they also talked about AI RAN a little bit, given that, they did collaborate with Well, they have been collaborating with NVIDIA for quite some time on the Arial platform or a software for RAN. So,

Earl Lum:

yeah I don't think they're pushing it in the way that everyone else is pushing it. I think it's more of a.

Leonard Lee:

Wow,

Earl Lum:

pragmatic perspective, like what AT& T was saying at 5G America is about ML and automation.

Leonard Lee:

Yeah,

Earl Lum:

and it's not gen AI of whatever of the, of my channel conditions on the downlink and uplink, and I need the GP, you at the edge at the base station. No, I don't think they're on that crazy bandwagon.

Leonard Lee:

Yeah, yet. Maybe yeah, but hey kudos to be joy well done. Sir.

Earl Lum:

So, you know, again I can't talk about the radios that they showed and the customers, but there's a lot of momentum going on over there. But all of that momentum needs money to fund it.

Leonard Lee:

Yeah, and so we're going to be pivoting toward in 2025 because, obviously, and based on what you've been sharing, they do have some core stuff that they continue to move forward with. I mean, I've heard there's announcements through the course of the year about and I even think I heard some stuff. Along the way as well, but, given that they have so many eggs in different baskets, where do you think things are going to land for them in 2025 in terms of direction and focus?

Earl Lum:

I think a lot, I think there's still a lot of focus that they will be. Doing on the open Rand, because that's where their investments are primarily going at this point. And if you remember the announcement from Rob Sony at AT& T, that Avenir is going to be one of the small cell radio players in addition to Fujitsu, which is good, but kind of a disappointment in that is not necessarily the major leagues. From the macro layer, it's kind of the minor leagues in the small cell layer, because the small cell layers, the layer I touch when I finished my major layer, and now I have to figure out how to fill things in. I want to add capacity. There's a fair, a fairly large opportunity in terms of the number of sites and radios, but each one of these is custom, just like on the Marvell side, every operator wants their custom radio for their custom bands and output power and instantaneous bandwidth. How it looks, it's got to have form factor because the small cells are a lot harder because they have to be stealth. So how do I fit this thing into a shroud on a light pole that I didn't really have to worry about on the tower. So I have a lot more physical limitations on the thermals, on the chassis and on everything else to make it. To where this is the flavor for AT& T. And if someone else wants it, how do I take my hopefully universal flavor and kind of just tweak it for them instead of just making a whole new radio for that one customer where they're only going to buy five or 10, 000 units and then I'm done.

Leonard Lee:

Yeah. But a win is a win. It's a good thing. Right. And

Earl Lum:

it was a win. And again, I think that they're in a good position at this point. The question is. Hopefully there won't be any more delays and pushouts, because obviously that just means you're sitting there burning cash, which is not a luxury that they have right now.

Leonard Lee:

Okay, cool. Yeah, thanks for that insight, Earl. Now what do you think about Open RAN? You're talking about the radios and their continued focus. So let's just broaden that aperture here to the state of, you know, Open RAN, which you and I talk about quite a bit. I think

Earl Lum:

it's going to be key to see in 2025 what happens to Vodafone Project Spring 6, because that's the big Open RAN prize that everyone's been talking about. And that 30 percent of whatever's on the European market. But as a reminder everyone, that 30 percent has shrunk quite a bit, because that 30 percent originally included Italy and Spain, which has been divested now. So, you've taken off a bunch of units off of that original number that Vodafone threw out there, dangling as a carrot. That carrot's a little bit smaller, it's not as big as it used to be, still a carrot, and it's still potentially a fair amount of sites. And the question is, is how does it get divvied up? If there's 5 winners of that 30%, that's not sustainable.

Leonard Lee:

Yeah, yeah,

Earl Lum:

because everyone's hoping that they get the whole 30 percent and you have a couple of front runners. I would say Samsung and Maveneer are probably front runners right now and Fujitsu as well. And then you have the incumbents within Ericsson and Nokia are still going to be there, at least on the European side. And then the African side, Huawei and ZTE will be there.

Leonard Lee:

Yeah,

Earl Lum:

because they're not banned in those markets. So it's going to be a very interesting situation where they announced that tender and who's who are the winners and how much of a percentage do they really get? And is everyone going to look at this going? You know what? I got so little. It didn't make sense for given. The amount of money I poured into R& D to try to win this tender, I'm not getting the payback.

Leonard Lee:

Come on, man. We're supposed to be a feel good show. Don't

Earl Lum:

Have we ever been a feel good show, Leonard? Come on. No. We're about the reality.

Leonard Lee:

Yeah, reality. I know. It's hopefully helped our audience here.

Earl Lum:

Because, 5 percent of that tender, it's a win, but that's a very expensive win.

Leonard Lee:

Yeah. Yeah. It's a, Hey, don't be mad at us. Call us, we'll help you out. Right. Don't be mad. Don't be mad. Don't be mad. Reality bites sometimes. That's just. The reality, but no this is good stuff. Going into MWC 2025, I'm really going to be curious as to whether or not they're going to continue to do the first future first stuff, or are they going to start focusing on the press?

Earl Lum:

Yes. The present problems

Leonard Lee:

rather than, setting sites and, uh, A future that won't come unless you start to work with the operators and, getting them, uh, up the maturity and learning.

Earl Lum:

And what will be their stance on Open RAN as we talk to them? Yeah, that's, that's what

Leonard Lee:

we're going to be. We're you and I definitely are going to be looking at that stuff when we. Land in Barcelona, but cool. Hey, Earl, thanks for the update. That's cool.

Earl Lum:

for having me.

Leonard Lee:

I had a great time and I totally It was great seeing you at Marvell. Yeah,, that shouldn't have been a surprise, but I don't know why we were surprised that we saw each other there. That was, I guess that

Earl Lum:

was because I wasn't expecting that you were going to be at Mavineer.

Leonard Lee:

Yes, don't. Give me a paper cut and pour lemon juice on it. It hurts already that they did

Earl Lum:

and then put the band aid over it.

Leonard Lee:

Maybe next year, maybe next year. Well,

Earl Lum:

I think it was a good it was a good M& M tour. This will be December is the M& M tour for last year. It was. Mavinir first and then Marvell this year was Marvell first and next and then Mavinir. It'll be was it'll be whatever it is every single year, but it will happen in December and we will have an Eminem tour update.

Leonard Lee:

Yes, yes. And there you have it. And hey, everyone, thank you for joining us. We really appreciate your viewership and listenership if you are listening to us on an audio podcast. And so. Look, if you want to get in contact with Earl, just hit him up at www. ejlwireless. com. And that's the website and research portal for his firm, which is EJL Wireless Research. And you can also hit him up On LinkedIn. I think you're starting to get a little bit more active on LinkedIn. Right. I know you were kind of lazy before, but I see

Earl Lum:

I was lazy before I'm much more active but in a very sensible way in that I will only write something if I feel compelled and it's important that I say, whatever I say.

Leonard Lee:

Yes, yes, yes. And you can reach out to him again on LinkedIn. You can reach out to me as well. And take part in the discussions that we instigate. I think that's the best.

Earl Lum:

Yeah, I think that's a good word. Instigate

Leonard Lee:

instigate. Yes. And, uh, oh, please subscribe to our podcast, which will be featured on the next curve YouTube channel. Check out the audio version, as I mentioned before on buzzsprout. Okay. Or you can find us on your favorite podcast platform. Also subscribe to the next curve research portal at www. next curve. com for the tech and industry insights that matter, and we will see you next time. Thanks a lot, Earl. Have a great

Earl Lum:

week. Happy holidays to you.

Leonard Lee:

You too.

People on this episode

Head Shot

Leonard Lee

Host

Podcasts we love

Check out these other fine podcasts recommended by us, not an algorithm.

IoT Coffee Talk Artwork

IoT Coffee Talk

Leonard Rob Stephanie David Marc Rick
The IoT Show Artwork

The IoT Show

Olivier Bloch
The Internet of Things IoT Heroes show with Tom Raftery Artwork

The Internet of Things IoT Heroes show with Tom Raftery

Tom Raftery, Global IoT Evangelist, SAP