Silicon Headlines for October 2024 - IBM, Snapdragon Summit, Arm vs. Qualcomm and more (with Jim McGregor)

Jim McGregor, Founder & Principal Analyst at Tirias Research joins Leonard Lee of neXt Curve on the October 2024 edition of Silicon Futures to talk about the hot topics in the semiconductor industry and technology.

It's busy season for analysts who are bouncing around the world attending the deluge of industry and company events. That means double, triple booking each week if not more. Then there are all those virtual session. 

October brought us Qualcomm's Snapdragon Summit, AMD's AI event, and a constant stream of headlines streaming of the industry including at Lenovo Tech World, the Westeros of the semiconductor industry where so many chip industry dramas are playing out at the moment. 

Leonard and Jim parse through the key announcements and semiconductor headlines that mattered in the month of October of 2024:

• The crazy season for industry analysts (1:06)
• Is IBM a phantom chip play with Sypre and Telum? (3:04)
• The chip industry is increasingly a systems game (7:09)
• AMD AI headlines with AI 300 Pro, MI 325 and more (7:47)
• Network and interconnect bound (10:01)
• GenAI scaling law bingo (12:57)
• Jim's next major architectural proposal and white paper (13:50)
• Hell freezes over with Intel and AMD - x86 advisory group (15:17)
• Is NVIDIA going open source with Blackwell? (19:56)
• Snapdragon Summit 2024 - Autos and Oryon 2 (21:12)
• Another Arm vs. Qualcomm flare up (22:03)

Connect with Jim and Tirias Research at www.triasresearch.com. Hit both Jim and Leonard 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.

Chapters

• 1:06 The crazy season for industry analysts
• 3:04 Is IBM a phantom chip play with Sypre and Telum?
• 7:09 The chip industry is increasingly a systems game
• 7:47 AMD AI headlines with AI 300 Pro, MI 325 and more
• 10:01 Network and interconnect bound
• 12:57 GenAI scaling law bingo
• 13:50 Jim's next major architectural proposal and white paper
• 15:17 Hell freezes over with Intel and AMD - x86 advisory group
• 19:56 Is NVIDIA going open source with Blackwell?
• 21:12 Snapdragon Summit 2024 - Autos and Oryon 2
• 22:03 Another Arm vs. Qualcomm flare up

Transcript

Leonard Lee: 0:06

Next curve. Welcome everyone to this next curve. Rethink podcast episode, where we break down the latest tech and industry events and things that are happening into the insights that matter. And I'm Leonard Lee, executive analyst at next curve. And in this, um, episode of Silicon futures, we're going to be talking about a lot of stuff that happened in the month of October, and I'm joined by The infamously fabulous Jim McGregor of curious research. Jim, welcome, man. How are you? Let me rethink this now. Wise man, wise man.

Jim McGregor: 0:57

Oh my gosh. I'm doing good, but I'm exhausted. There are too many events in October, too many events in November, and even Too many events in December.

Leonard Lee: 1:06

Yeah, this has been a very insane season. I just got back from 5G Americas, which we're not going to be talking about. But yeah, it's like constant. I barely have any time to write and do anything. I have time to do this podcast with you, my friend, and Hey, it's good to see you. You, we didn't have you on in the last episode Carl or missing Carl this week. But we have a lot to talk about. We're going to be talking about IBM, AMD Snapdragon summit. We're going to be talking about that interesting X86 partnership thing between Intel and AMD that was announced during OCP, but actually Pat and Lisa. We're in Bellevue at Lenovo tech world for that announcement. So we'll talk about that and the game of thrones that is happening in the AI computing universe. And Lenovo is the Westeros and we'll talk about that. And then finally some drama in the world of Silicon arm versus. Qualcomm. We thought this thing was squashed, but it keeps flaring up. So before we get started, remember to like share and comment on this episode and subscribe to the Rethink podcast here on YouTube and on Buzzsprout. Take us home on the road, on your jog, and listen to us on your favorite podcast platform. And also make sure to check out Tyria's research. Mr. McGregor and his team, they are legendary and if you are not tuned into what these guys talk about you are missing out big time and, and obviously you guys are luminary that is the only reason why you're on the show. I just want you to know that

Jim McGregor: 2:56

is that a good way of saying old,

Leonard Lee: 2:59

very flattering way of saying that you're old. So yeah,

Jim McGregor: 3:03

I've been around the industry way

Leonard Lee: 3:04

too long. I know where all the skeletons are. No but no, I think that there's tremendous value in that. And, as you and I have engaged over the years, you're. You're the man. It's great to, it's actually really great to be collaborating with you and your team. And let's get into this. Let's stop patting each other in the back and let's talk about your week. Looks like you were in New York or that you were New York, you were in Las Vegas. You're doing some stuff with IBM. Yeah,

Jim McGregor: 3:35

It's been a busy month. Especially on IBM obviously focused mostly on the cloud side. Yeah. And they announced at hot ships, their new talent processor Inspire Accelerator for their Z series mainframes. At the tech exchange this week, they started talking about, which is their developer conference in Las Vegas. Mm-Hmm. they started talking about power. Which is the next generation of the power series of processors for their systems. They're going to have a power 11 and they're even talking about power 12 and beyond. Wow. Yeah, power. It's still on the roadmap. So power 11 is going to be an evolution of power 10 some the one significant enhancement is that they're actually putting a interposer on there to improve power distribution. So not quite going to a chiplet architecture yet, but it's a 2. 5 D architecture in implementing that and obviously enhancements throughout the architecture that's going to be coming out in Q3 2025, which they mentioned and it's going to be paired with Spire. So they're going to be using their AI accelerator, both with the Z series and with the power series. That's interesting. The power series though, note that it did transition a few generations ago to the using PCI express. So with the power systems, you can use pretty much any accelerator you can use. You could use Nvidia, you could use AMD, you could use Intel, you could use any base AI accelerator in the system. They'll have a number of options for power. So it's still very relevant in the industry. And then with 12, they did mention that they're going to a chiplet architecture. But not just really for the processor, but to look at it from a system view. So I would expect you're going to start seeing the same thing we've heard from, Intel over this past year. And the fact that you really have to think about how you put the CPU the they're very excited about their roadmap and where they're going. And, and I, it sounds like they're going in the same direction as a lot of the industry. And it's going to be interesting to see.

Leonard Lee: 5:33

Yeah that's. That's really curious. So is IBM making a silent fabulous play here? And, do you think they're going to become one of those logos that we see entering the this whole semiconductor chip race? People don't think of I don't think of IBM anymore in regards to, um, Which is

Jim McGregor: 5:58

which is foolish because IBM still invests heavily in semiconductor technology especially through the university of New York up there where they've got, where they invest with a lot of different groups in future technology. Matter of fact, they're one of the leaders. They're even working with Intel on certain technology. Yes, exactly. No, they're not looking to become a semiconductor company again, but obviously they're still system services and software. And, part of that systems is having a really competitive offering. Matter of fact, it's funny. I still get this question from time to time. It's does anybody still use mainframes? Yes. And it actually fits very well into this whole hybrid data center or hybrid cloud architecture because data centers are still really good for those high secure regulated environments, especially if they've got that legacy code. But, even IBM is working on that. They've got some, they've got a whole bunch of AI tools coming out. That actually help you convert some of that old COBOL code and other code into Python or JavaScript or Java. So they're not making necessarily a play to be a bigger semiconductor vendor, but they're still making, they're systems and the mainframe's still an important part of that hybrid dataset. Yeah.

Leonard Lee: 7:09

But just like you said, Jim,, the chip game is becoming a systems game. Increase it is very much and, it sounds like they're building some and designing and taping out some custom stuff that, and they have their own, like you said, they have a roadmap. So, I mean, I'm just pointing that out because and asking you just because it's interesting, people usually don't talk about IBM when it comes to chip. So

Jim McGregor: 7:33

cool. No, and they've had a decades long partnership with Samsung. So despite the fact that. They sold off their fabs and fab manufacturing capacity to global foundries. They still leverage of Samsung very heavily going forward.

Leonard Lee: 7:47

Yeah. Okay, cool. And then you were also at AMD's event, right? Yes. They're

Jim McGregor: 7:52

they're advancing AI event. When you remember where I was it's morphed over the years. It's one of these events that keeps changing names. But they started this when they introduced the Zen architecture and it slowly integrated more of their products for the data center. And this year they made it more of an enterprise event. So not only did they talk about all their products that they're doing on their annual cadence for the data center, but also they started talking about their new processor for enterprise class PCs, which is the Ryzen 300 Ryzen AI 300 Pro. There's too many parts to these names these days. But But they didn't talk about their other pieces, which I would have liked to have seen Threadripper and Ryzen Pro, but not Ryzen Pro, Radeon Pro. But, they still do, more than anything, this event represents their continued cadence to deliver, just consistently deliver ever since they introduced Zen and they really underplayed it when they really, when they first introduced that, they have just continued marching. Now they're 34 percent of the server market in terms of revenue. So, I mean, it's still growing. They're very strong there. They've got their new Zen five architecture coming out with their 9, 005 series epic they've got their. New instinct, MI 300 25 coming out this year, then the three 50 next year and then the 400 in 2026, so they keep continue executing on that. Yeah. And this year, really, probably the most important stuff they talked about was Pensendo. Really, the networking technology. So, they've got a new Sienna 400, that's for the front end, basically system to system. And then they've got their Polara 400 for the back end. Really, connecting those accelerators. And it was really interesting, even though they didn't, we didn't really get, I don't think, a really good demo of this. But. It's AI enabled. This networking solution, this Polaris 400 is designed to map out the entire network and monitor and find out when there's blocks anywhere in the network and reroute everything. Really enhance the networking capability and improve the overall efficiency of those systems, especially when you're doing AI class workloads.

Leonard Lee: 10:01

Yeah, and it's interesting to see a lot of the attention continue to gravitate toward the importance of networking and interconnect. Right? Especially as these workloads are changing, going moving from emo ease to the. These now more advanced compound AI architectures the networking requirements are, seem to be shifting quite a bit as well, or evolving, I think, but evolving in a direction where you need to have much better timing, you need to have a much better band bandwidth, obviously. And be an intelligent, be intelligently be able to cluster all these resources across the system. And yeah, I, if anything, I think these are the innovations that are going to stick, right? Well, absolutely. You and I and Carl can argue whether or not these large language models, blah, blah, blah, are going to change the world. But the thing that I think is so frigging interesting. Did I just say friggin? It is so interesting is how all of these nascent networking and compute technologies are just finding a moment. Right? And this is the stuff that is not going to. It's not going to be backed out of investment and stuff like that, because, we're also seeing a lot of these innovations start to trickle down toward, edge infrastructure environments, or at least some of it starting to make its way down as people are looking toward how. The innovations in the, generative AI supercomputing environments can now be scaled down to, let's say, edge infrastructure or on prem or what have you. So that's another really interesting dynamic that I'm starting to see just simple because, I was at a Lenovo tech world and that's. A large part of the conversations that I have with some of the OEMs, you know what I'm saying? The customers of the Lenovo, that the NVIDIAs and the AMDs and the Intel's of the world. But yeah,

Jim McGregor: 12:08

it's, it really is. Jensen hung of Nvidia came out and said, listen, the new unit of computers, the data center, and Pat Gelsinger's reiterated that as well. And it really is, you have to think about making AI more cost effective. That means the entire data center has to look like one system. It means that you have to efficiently use all those resources because we're still facing limits in Moore's law. The power consumption is going up. We need more memory bandwidth. We need more networking bandwidth. That increases power. So you have to improve the overall efficiency of the entire system or systems. I should say, you know, especially we're in a hybrid cloud environment. So it is it is really about, how do we make all these things more efficient? And it's not just about shrinking transistors anymore. It really is about the system.

Leonard Lee: 12:57

Yeah, and, now you have everyone talking about scaling laws and everyone's inventing one. It's just becoming quite honestly nonsense. It's just becoming the next ridiculous thing. But I think you're onto something system scaling using different types of techniques. At a system level, and that enables the system to scale, but, 1 of the things that I always have debates on is this notion that Moore's law is dead. And then somehow there's these new, there are these new scaling logs that are independent of Moore's law. I think still everything is. Moore's law constraint until we find the next best thing. It's just at these system level, more holistic portfolio of technologies are enabling the system to scale. But, that's my take. I don't know what your reaction to that. I think we're going to see a lot of innovation in the

Jim McGregor: 13:50

systems. I think right now we're still, that's what we're seeing, right? Incredible. I'm almost surprised that we haven't taken a harder look at blades again. The VME compact PCI advanced TCA stuff that I used to do at Motorola. The, you know, we're maybe what's old

Leonard Lee: 14:09

new

Jim McGregor: 14:09

again. Is that what you're trying to do? I'm just saying that, you know, we went to these, we used to call them pizza boxes, but they're called trays now and they're really, slide in servers or when you to you for you type surface that you slide into a rack. We used to build and for specific reasons because you needed better density We built these blade servers, you know It changes the way you cool the way you power you can do and you're using a backplane matter of fact We even did 15 years ago. We did optical backplanes. They weren't cheap, but we did them I mean you can have optical interconnects. You could change the flow, airflow not to mention the way you have not the way you do your interconnects and cooling and everything else. It is very significant. I think there's going to be, I think within the next couple of years, and I've had some discussions with some of the, I've had discussions with some of the infrastructure vendors and some of the systems vendors and say, listen, we need to rethink this tray idea because it's not optimal anymore. So I think we're going to see a lot of innovation in system design. So folks, this is why it's good to be old. Hey, Hey, Hey, Hey, Hey.

Leonard Lee: 15:17

So, Hey, let's move on to let's move on to this really interesting. Thing I'm losing my words here. I have the worst words. This X 86. What do you call it? A counselor? I actually have a partnership, right? It is a council. Yeah. And, uh, Uh, Intel decided they're going to work together in terms of instruction set development going forward for x86 and that they're going to employ the use of an external council of OEMs and well, partners, I should say they're there. You know, you've got Broadcom, you got Microsoft, you got Dell, you got all these different entities that definitely have a interest in this. I wouldn't be surprised if some of the people on that council were the ones pushing for this. You know, having two different vendors with two different instruction sets for the same architecture can be a bit of a pain. Um, I'm sure that was part of it and also some competitive threats. Yeah, yeah the timing was a little bit odd and we're going to talk more about interesting timing, right? Later on in the conversation, but you're, yeah. And so you and I were on a call where we were pre briefed on this stuff, but yeah, it's interesting because it is a governance of 2, I think it was like nine, nine, they're not really stakeholders. Like you said, they're advisors or parties of interest. That's a wrong term semi stakeholders. And then it's really up to Intel and AMD to agree on certain initiatives and and standards or. What have you the folks on the, on that logo sheet, there's a question mark as to what sort of bearing they have on decisions that are made. Maybe they help set the agenda, but that was one of the things that I think was really curious for me. I know that in the, in discussions, you and I went back and forth with the the folks in the AMD, as well as Intel folks who were on the call. Yeah, that's 1 thing that I'm really curious about is how that's going to evolve. And then where are we going to when are we actually going to see impact for their customers? And when do we see this meeting of minds around, the X 86 standards and and so that's what I'm really curious to see because I thought the announcement was timed really weird. It almost was like, Hey, look, we're worried about something like what you said, the competitive pressure. Right? And so from a messaging perspective, I thought it came off a

Jim McGregor: 17:50

this is definitely early days for them. They wanted to get the message out as quick as they could. You have to remember that all the processors or SoCs that they're developing, they're developing two, three years out. So when will this impact actually the ecosystem and the future products? Probably not for a couple of generations. So, yeah, you might see some level setting of instruction sets in a year or two, but most likely two, three years out. But, and you're right, it's a governance of two, but it's an impact on the largest semiconductor ecosystem that we have to the x86. So it I definitely think it's going to help. It's going to be a huge benefit for software developers and everyone else not to mention OEMs. So it doesn't necessarily, it only applies to the instruction set. It doesn't apply to the implementation, the products, the pin outs, anything physical, it really is just the instruction set. But that is significant because we've seen the two kind of diverge, over time, throughout their history. And it's been a back and forth, and it is kind of mind numbing when you have different instructions for specific functions between the two. I'm sure Microsoft is definitely interested in seeing this kind of come to fruition. So they have a single instruction set to support going forward. Yeah,

Leonard Lee: 19:04

and yeah, good things happen when hell freezes over.

Jim McGregor: 19:08

Haha.

Leonard Lee: 19:09

Kind of like the way I look at it,

Jim McGregor: 19:11

right? That is true. You like that? Well, I asked him if I could use that as my headline.

Leonard Lee: 19:16

Speaking of hell freezing over, yeah I was at Lenovo Tech World. That's where this partnership was announced. I think the press release they had a separate simultaneous announcement OCP, but yeah, Pat Gelsinger and Lisa Sue were on stage. They actually shook hands and they had a Kumbaya moment with YY up on stage. And yeah, that was cool. I think it it was a big surprise at the event and yeah, I'm looking forward to seeing how the partnership progresses and I know that you and I are going to have a Hawks eye eyes. On this and anyway yeah,

Jim McGregor: 19:56

now I want to bring something up since you're on a post CP and the fact that we've been talking about systems and partnerships and everything else, there was a major announcement at OCP and the fact that NVIDIA is. Introducing or contributing the NVL 72 system architecture to OCP, which was a huge development by them. And we've already seen Meta take that and actually modify it for their own uses. The fact that, that was a drastically new system architecture. Like I said, I, the reason I think we need to re look at Blades, but they had to re architect a rack to support 6, 000 pounds, all these GPUs, the power, the liquid cooling, not to mention 5, 000 cables, combining everything into a single GPU architecture. It was, I think that was a very interesting move and a very good move for both NVIDIA and for the industry to help further that system level innovation.

Leonard Lee: 20:46

Okay, well, it sounds like you need to write a paper on that yeah, let me know when it comes out because I'm going to be really interested in reading it and we'll share it with everyone on that long list of to do list.

Jim McGregor: 20:57

I've been traveling for a month.

Leonard Lee: 21:00

People don't know. People don't know. They think we're just having fun. Yeah. It's a lot of work and speaking of funny timing arm and Qualcomm.

Jim McGregor: 21:12

Yes. Qualcomm was holding their annual Snapdragon summit down in Hawaii and announcing their latest series eight processor series, eight elite processor for mobile, as well as their solutions for automotive using they're also going to be drive elite and yeah, anywhere the other one. But no, Snapdragon eight, eight elite. It's eight elite and then they have the ones for the automotive too that Automotive. Yeah. Elite. Cockpit. And ride and drive, bro. Ride cockpit. Ride ride. Sorry. Cockpit and ride Too many terms in our industry.

Leonard Lee: 21:43

Oh, wait, wait, wait, wait. So last year ICI couldn't for the life of me. Get the Snapdragon X elite thing. Right. I kept calling it elite X and yeah, painful, all this stuff.

Jim McGregor: 22:03

And all of these products having that elite moniker. Are include this new Orion core, or I should say family of Orion course that they've developed based on, using the expertise of the Nubia team that they acquired. And that's been the whole issue of this lawsuit between arm and Nvidia. Obviously, Nubia had a architectural license. Qualcomm had an architecture license Qualcomm bought Nuvia, and the whole argument is whether or not that requires a new license. Yeah, ARM dropping, now, we should note that this has been going on for two years. We have a court case coming up that's set for December, but during the middle of the event ARM decided to drop another bomb saying that it's revoking, giving 60 day notice, revoking the architecture license of Qualcomm. So that impacts everything they're doing with those cores for mobile, for PC, for automotive and doing that in a timeframe when, where there's already a court case scheduled. To decide this is, it was kind of awkward. Uh,

Leonard Lee: 23:10

Yeah, it is curious and it's It'll be interesting what Qualcomm's reaction and play will be pre trial, right?

Jim McGregor: 23:20

The fact is that this is all going to come down to the wording of the contracts, which I don't think anybody outside those two companies has seen, other than maybe some redacted pieces. Yeah, you can surmise, but nobody knows for certain. Nobody knows, and how the judge actually interprets it. So that interpretation is going to be critical as well. But, you know, let's face it, the whole issue of Suing your largest customer is not necessarily setting a good precedent for ARM. They may win this, who knows, I don't know, but it definitely does not sit well, I think, within the ecosystem.

Leonard Lee: 23:55

Yeah, yeah, it's it's. Not a good thing. It doesn't help the whole thing has not helped. But you know, one thing that is pretty evident. Is that Gerard Williams and team have been able to deliver pretty solidly against some of the promises that they made last year at Snapdragon summit over there. Remember And at the time, there was a lot of controversy about the benchmarks and how it was not as open and, a lot of skepticism about some of the claims that Qualcomm was making. But as you and I found out in our conversations with Gerard at hot chips they made less more, you know, took a really balanced approach to their chip designs. Also their strategy and roadmap where I think what we're seeing with the gen two Orion. CPU seems kind of scary. I don't know what your peak

Jim McGregor: 24:51

is. Well, and they've got different, they've got different versions of it for performance and efficiency and stuff. So they pretty much have their own big little strategy with, with the Orion core. Right. They

Leonard Lee: 25:01

call it prime. Prime and performance. There's no efficiency. Okay. Prime and performance. You're right. Right. Yeah. Yeah. And yeah, to be clear, the gen 2 Orion is going to mobile. It's not going into the family, right? Which is there. Yeah. Right. Yeah. But we can anticipate. Probably next year, mid. At midpoint of next year for a new generation of Snapdragon X.

Jim McGregor: 25:28

This is their core architecture going forward. So we can expect to see it in pretty much across their entire product line going forward. Unless there obviously is, and it is interesting because typically when we see TIFFs over contracts, usually these things settle themselves out before court, but this one seems destined to be settled by a judge.

Leonard Lee: 25:50

Yeah, and that's the interesting thing, because I had thought that they might settle out of court, but, there's been a bit of radio silence between the 2 for a better part of a year, it was almost like, and that's why I think a lot of us had the impression that this thing might just either go away. Or it would get settled quietly. But yeah, it's it December will be, it'll be interesting to see how all this stuff progresses. But yeah, puts a little bit puts a bit of a damp towel or what towel on a. A lot of the announcements that came out of snap dragon summit in Maui this week. But

Jim McGregor: 26:30

yeah, I don't know. We'll, we'll have to wait and see. It's difficult to judge who's right, who's wrong. And once again, it's going to be up to a judge to determine that. So, it's, it's ironic because if arm actually wins, it would be devastating to them in PCs and other segments. So, so, Uh, if when you lose, I don't know that's kind of ironic here.

Leonard Lee: 26:55

Yeah. Yeah. Undoubtedly We'll be speaking about this more given that we're gonna be monitoring it. Yeah. The developments in our research and our coverage of these two companies. Anything else? Anything where else we're missing? This is like a whole bunch of material and you know what's crazy is like every month it seems like there's way too much stuff to talk about.

Jim McGregor: 27:17

I, I, I would agree. There, there's just so many things going on in our industry right now. It's almost scary. It is really scary. It's, there is I think I'm on every press list in the planet and I get on average 500 emails a day. I'm like, if you really need to reach me, don't go through email. uh, you, you really need a

Leonard Lee: 27:36

copilot, my friend Yeah. Hey, Jim, I think let's call it quits. Okay. While we're ahead. I think this is a good episode thus far a lot for the audience to chew on. And why don't you take a moment to talk about a curious research and how folks can get in contact with you, but also tap into curious research wisdom and Buddha knowledge.

Jim McGregor: 28:03

Yes, absolutely. Terrius Research, you can reach us through our website. That's terriusresearch. com. You can reach me and my colleagues directly. It's just our first name, jim at terriusresearch. com. Our contact info is up there and we're always available. We're also very visible through a number of media partners including Forbes e times e journal, ECT news, and some others. Definitely look for our articles, look for other podcasts, look for our partnership with our good friend, Leonard Lee here and everything else that we can do we are a custom high tech market research firm that deals with everybody from IP and technology vendors all the way through the cloud and service vendors. Yes, and he's not as old

Leonard Lee: 28:46

as you think he's actually a young guy.

Jim McGregor: 28:49

Well, you know, I'm the author and do things, you know, I, I hiked Mount Jade. Where were you? Okay. Well, on that note, you need to hide the Grand Canyon with me,

Leonard Lee: 29:02

Leonard. Yes, I will. I will. When I, when we find ourselves in Arizona at some point near the Grand Canyon, but yeah Hey, everyone, thanks for joining us and sticking with us thus far. Hopefully this has been a informative episode, please subscribe to our podcast and it will be featured on the next curve YouTube channel, which you're probably watching right now. And check out the audio version on Buzzsprout and find us on your favorite podcast platform. Also subscribe to the NextCurve research portal at www. next curve. com for the tech and industry insights that matter. And we will see you next time in about a month, right? And hopefully Carl joining us. So, all right, my friend, take care. You too. Thank you.