Silicon Futures for March & April 2025: Tariffs, Geopolitics, and Supply Chain Turbulence

Jim McGregor of Tirias Research joined Leonard Lee of neXt Curve to recap March and April 2025 in the world of semiconductors and accelerated and non-accelerated computing on the neXt Curve reThink Podcast series, Silicon Futures. Jim and Leonard share their thoughts on the state of this semiconductor industry in light of a volatile pair of months that saw the starting of a global tariff regime instituted by the White House that has roiled the markets and industries and trade relations with the U.S. face unprecedented uncertainty. What are the implications on the semiconductor industry and the global electronics supply chain as the tumult continues?

This episode covers:

➡️ Global tariffs and its impact on the global electronics industry (3:19)

➡️ Jim's impression and takes from Embedded World 2025 (34:45)

➡️ Edge AI and the diffusion of AI accelerates (39:22)

Hit both Leonard and Jim up on LinkedIn and take part in their industry and tech insights. 

Check out Jim and his research at Tirias Research at www.tiriasresearch.com.

Please subscribe to our podcast which will be featured on the neXt Curve YouTube Channel. Check out the audio version on BuzzSprout 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

• 3:19 Global tariffs and its impact on the global electronics industry
• 34:45 Jim's impression and takes from Embedded World 2025

Transcript

Leonard Lee: 0:10

Welcome everyone to this next Curve Rethink podcast episode where we break down the latest tech industry events and happenings into the insights that matter. I'm Leonard Lee, executive Analyst at ncur, and in this Silicon Futures episode where we focus on all the things, related to semiconductors in the industry itself. we're gonna be talking about the happenings in the month of March. April. So we're doing a little bit of a makeup here, and I am joined by my good friend and fellow industry analyst, Jim McGregor of the Famed Cheerious Research Group. Do you have a group? You don't have a group at the end of your No. Yeah. It's just curious research. And so, Jim, how are you? Good. How are you doing today? It's been a while. We've been super busy, right? Yes. It's been a very hectic year. Dare I say it's still the beginning of the year. Is, is that No, it's almost the middle of the year. Where's the time going? I know. It's really crazy. So, but before we start, please remember to like, share, react, and comment on this episode. Also subscribe here on YouTube and but Sprout to listen to us on your favorite podcast platform and, Also would like to remind everyone, opinions and statements made by my guests, in particular Jim, in this instance and in this episode, are entirely his own and don't reflect mine or those of next curve when we're doing this to provide an open forum for discussion and debate and everything else on all things. Related to consumer tech, commercial tech, and the semiconductor industry. So, Jim, this installment, we're gonna try to do a little bit of catch up, right? We, didn't do our March bit. unfortunately, Carl Fre, who's, another partner in this, endeavor that we have with this Silicon Futures podcast isn't able to join us. we're gonna hit on March and April, right? Yes.

Jim McGregor: 2:14

that's a lot to hit on.

2:15

I know.

Jim McGregor: 2:16

March I was gone the entire month of March, the entire month.

Leonard Lee: 2:19

I'm proud to say that I managed to spend some time at home, but you're right. we have a term for that. It's called lazy. Yeah. Wow. Hey, Jensen won. Come on. that's basically what he was saying, right? Yes. Make my people lazy. Lazy. That's a secret to success. Exactly. speaking of March, we had, and you know, Jensen Wong, we had GTC as well as MWC and Embedded World. And I know you've been chomping at the bit to talk about embedded world because that was a really big event, but mm-hmm. Obviously something that really. resonated with you this year. Yes. so I take it we're not gonna talk too much about GTC, right? Because we already have a separate podcast on that. So folks, if you, I quite

Jim McGregor: 3:09

honestly, I think we should talk about some of the most recent, things that have happened. the intel learnings, the shift of manufacturing by Apple to India. the biggest thing to talk about right now is the tariff scramble. Yeah. Yeah. And that's the way I phrase it because everyone's scrambling. Nobody knew how to really prepare for it. So everyone right now is trying to figure out, how's it going to impact them, because it's changing every day. And how can they, expand their supply and manufacturing resources so that they can avoid the tariffs or limit the impact of the tariffs.

Leonard Lee: 3:43

Absolutely. Yeah. I mean, I think, that and a lot of the geopolitical mm-hmm. events that have transpired in the last two months, it's pretty overwhelming actually. Right. And, that's probably the most top of mind thing in the industry. and the sourcing. The security, resilience, all these topics are, are really coming into play and, I. So, yeah, we can definitely touch on that. So did you want to expand on that right now or do you want to, do you want to we, I think it's a great

Jim McGregor: 4:19

place to start because I mean, really that is probably the most critical thing going in our industry right now is adjusting to. This tariff, this trade war that has erupted over, the US tariffs and not understanding really how things are gonna, and I've actually had this debate with some people and they're like, oh,, it's gonna be minimal'cause it's just this amount and blah, blah, blah. I'm like, no, it's not when you start factoring in, you got, 25% tariffs coming into the US on steel and aluminum. That impacts almost every industry that you have to talk about. Rare Earth materials, which are a huge, huge leverage point that China has when they control 85% of the rare earth material manufacturing at this point in time, or processing, I should say. and some people have said, okay, well yeah, but there's plenty of rare earth materials in the US and other parts of the world. It can take three to five or even 10 years to bring that to process. Right, right. So I think that the worst thing we have going on right now are those tariffs, and especially this battle between the US and China. We've spent decades. Creating a global ecosystem and leveraging the expertise, the labor resources, the manufacturing capability, all the resources that go into building a very robust electronics industry. and now that's being torn apart and it's gonna be very painful.

Leonard Lee: 5:44

Yeah. I think there's more going on here than just, simply tariffs, right? you mentioned trade war, and you also mentioned the rare earths, We're also starting to see, some movement on the services front, right? I think it's very difficult for any, company to really navigate what's happening right now because the level of volatility, in terms of the policy, how it's being expressed, it flip flops, dramatically, literally day to day. So. not a easy situation for the, industry at all.

Jim McGregor: 6:20

Well, and I think the. Policy makers, and I'll leave it at that, are underestimating the impact. Yeah. Obviously you have it on the final product. You have it on the supply chain. Yeah. You have it on, legal environments. it could impact, IP protections. Across borders, now you even have it going out to consumers. We're already seeing this where Canadian consumers are trying to avoid any US products. so I mean that, it goes from the heart of our industry all the way out to the consumer and it's gonna have a huge impact. and even as you mentioned, services. are companies willing to use a service if it's headquartered in the US or if it's headquartered in China, or even if they have offices in those countries? it's going to be a very difficult situation and. It's going to be a difficult regulatory situation to navigate. once again, things changing day by day, but also knowing that, there's likely to be other issues with technology transfers or IP sharing or any type of interaction or transfer of data. So it's going to be a challenge. going into 2025, I thought it was the most exciting time for our industry. Yeah. just the direction we were going with AI and AI is still huge and still going millions of miles an hour. We have a whole new forecast for that, and it's blowing away anything that we did before. the demand is insatiable, but. At the same token, where everything seemed to be moving in the same direction. think of the wireless industry. 5G was the first wireless technology where from day one there was a single standard for the entire world. From day one, and we've had that going forward and it's really propelling the industry. And now we're threatening to break all this effort apart. Yeah. Over, this trade battle over these IP issues, over, all these other political issues. the electronics industry unfortunately has become a pawn of the political world.

Leonard Lee: 8:22

Yeah, absolutely. And I'm glad that you brought up the whole. Topic of ip and at the end of the day, what's really being, threatened. If not compromised in the process of being compromised is that level of global trust. Right. You mentioned 5G, and this is all based on trust. This is about literally industry and regulators as well as government supporting. a standard, I'm not gonna say open. It's more of a standard for technologies so that these technologies can scale. And so going forward, depending on how the tariff conversation and the trade relations play out and the tech relations play out, we might be looking at a very fragmented, trust landscape. In fact, that's probably already in play right now. I think not only the semiconductor industry has to ask itself, but the broader TMT,'cause we have to bring content in here as well, right? these are all, industries that have benefited from this global standardization, markets in order to scale and, deliver on this, ubiquitous benefit. for consumers as well as enterprises, right? What happens when that all becomes dismantled? And I agree with you, this is something that, policymakers around the globe need to contemplate and are starting to contemplate. it's just a very, very complex topic and it's also something that's going to be very difficult to, navigate. all the critical considerations, because the landscape is shifting so quickly, right? But, unfortunately it looks like things are moving toward distrust. Versus trust. And as you, I think most people know you can lose trust in a second, right? It takes years, if not generations to build trust. And I think this is, number one, a big. Risk for the industry, because the dismantling of trust is something that, is not trivial. And we may have to start thinking about, well, what is this trust landscape going to, look like? What, where is it moving towards? And how well new networks of trust be built, because that's where you're gonna see pockets of value, market opportunities. And so these tams that everyone was looking at are all of a sudden becoming very fragmented, and this is just gonna be really tricky for, the entire industry. And I don't think there's a, solution at the moment, right? Everyone is just trying to figure out how they need to react or not react or maybe, take their time to examine, what a sensible play might be, in an otherwise, super volatile, situation.

Jim McGregor: 11:23

Well, and you mentioned trust. Yeah. That trust in the technology, that trust between technology providers, that trust of the consumer, all results from our industry working together holistically. Yeah. our industry has overcome issues with, obviously with security, with safety, with everything else. Think of the fact that we've never had a major disaster around a semiconductor fat. That's a huge accomplishment, especially when you, someone that knows some of the chemicals they use, like hydrofluoric acid, it is a huge accomplishment. So if there's one message, and I really, really hope that we get this podcast out to, government regulators around the world. Mm-hmm. And say that, listen, our industry and innovation thrive when politics stays out of it. Yeah. And that is a key point. granted, the electronics industry is so important to, the political realm because it drives GDP, it drives growth, it drives innovation, and it should. The best way you can do that is to support it as a whole and to support it through, incentives and everything else. But when you start putting restrictions on it. You really hamper everything that we do. quite honestly, every government should learn their lesson by now that when you start hampering technology innovation, you really threaten the technology development in your own country.

Leonard Lee: 12:47

Yeah. Especially when you, can argue that, you're in a leadership position and you are somehow, now succumbing to this notion that you're not, I think. You're making a great point there because. while TSMC might have, leadership on the fab side, the manufacturing side of things, from a design perspective, systems a technology perspective, it, it still the United States with all of the big brands, that we all know so well, that have amazing technology portfolios, but are also exhibiting, product leadership. And market leadership through, leadership products. Right. Nvidia. Perfect example, right? so I think discounting these things can be really, detrimental because as you start to look at what your policy needs to look like, you have to be cognizant of whatever leadership. Is in play at the moment. And I think a lot of this is also relevant to what's happened with 5G, and six G, where, the policy makers may not be appreciating the importance of this technology because it's gone through a hype cycle. I say that has been detrimentally disregarded, but um, China is definitely not easing on the throttle on mobile technologies. Right. In fact, they're doubling down. and so they're looking at whatever kind of advantages that they can, lean into and maybe invest in and accelerate, to have a competitive, advantage, going forward. And so, yeah, I agree with you. one of the things that has been a huge benefit for the industry is global competition.

Jim McGregor: 14:39

I'd call it coopetition

Leonard Lee: 14:41

because it's cooperation and competition. Right, right. You're, you're right. and, to the credit of, many of the Chinese companies that have engaged over the years in, this global electronics, industry and its advancements, they have contributed. I I think the myth is that they haven't. I think it's well documented that they have contributed and have been, active participants, in the advancement of, the electronics industry as a whole and also, the semiconductor and AI sectors, so I think those are all important things to recognize. as a policy maker, otherwise you can shoot yourself in the foot. And then also, considering all the things that you mentioned earlier, Jim, about supply chain and, the structure of, the electronics industry as well as the semiconductor industry, which is a subset. being very, very clear on that. so that, as you're looking at developing your policies, you're doing it a thoughtful and strategic way, right? That helps you achieve whatever goal that you have, whether it's, resiliency, supply chain, resiliency, Or, national security. that all has to be thoughtful because at some point, if your industry and your industry, champions, have their market shrinking. that doesn't, help advance any, objective I think even national security.

Jim McGregor: 16:12

a message to the. Industry itself. Vendors need to think globally, obviously not just in terms of their customer base, but in terms of their engineer resources, their manufacturing and support resources. I think a lot of the companies are, we've seen even some of those results of the past month, apple shifting, smartphone manufacturing to India. That's a huge announcement. also, TSMC. Introducing fabs in the us playing on introducing fabs in Europe and in Japan in their future. So I think that we're seeing that and that that's really an important thing. I think the next step a lot of these companies need to take is they need to think about expanding beyond their traditional markets. And we've seen some of this, obviously in automotive. Automotive is a great example where the whole value chain is thrown into disarray because now you have semiconductor providers not just providing chips. But providing the boards and complete system level solutions. Yeah, and I think that's almost what we need, what a lot of companies need to think about. Because we don't know how all these trade tensions are gonna play out, but it's really important that you think about. Not just where you playing the value chain, but you know where you can expand and enhance that value chain. Nvidia is a great example. It went from being a chip company to a board company, to a systems company, to a software company, to a services company. I mean they pretty much provide the entire stack. For AI solutions from embedded applications like robotics, all the way through, the data center. For the high-end AI factory data centers. I think it's really critical. And I think the last thing that these companies need to do, and I'll be honest with you, I don't think enough of'em are thinking, even in the technology industry, are thinking enough about it. Not just how they can use ai, but how AI changes their business model. Yeah. and some companies, like I asked Jensen Halling that this Ed GTCI said, how do you, where do you see. Nvidia in 10 years. He's like, Jim, I wish I knew. that was a great answer because his business model is changing constantly around ai. He likes to say that they innovate at the speed of light and they really are. And it's not just their products and services, but for the technology it's also their business model. And I think a lot of technology companies, as well as other companies throughout other industries. Need to think about how their business model changes. and I really think that that globalization, that, change in business model, and change in where you play in your industries is really gonna drive a lot of change and hopefully overcomes some of the issues we're seeing today with the political environment.

Leonard Lee: 18:55

Yeah. And well, I might. I might beg to differ a little bit. I generally agree with what you just said, Jim. but I also think that, companies like Nvidia and others have been a victim of the hype as well. I can appreciate the advancements that have happened with quote unquote ai, but more specifically, we need to be talking about. Generative AI and what's happening there? I think, beyond some of the prototypical use cases and applications that it's being leveraged for, such as, these chatbots like chat, GPT and some of the reasoning models that are now starting to come out, these are still, we're still in early days in terms of these technologies being mature and I think. One of the things that has victimized the industry is the hype that, and sort of these misconceptions that somehow, AI innovation is specifically tied to the number of GPUs or the kind of GPUs that you have access to. I think another big news item that came out. this in the past two months was Huawei introducing their AI cloud matrix. You know, I think it's like 3 84, right? Mm-hmm. Where it's an all optical, system, right? no copper, apparently. And, it can scale out larger than a Blackwell. MVL 72, according to, our friends at semi analysis that did a study on this. So, you could be chasing the wrong rabbit if you don't understand these things at a system level. But more importantly, understanding. Well, where are we with this technology and what is it and what sort of impact is it really having? AI has always been difficult. it's always been costly. It's never been easy. And so some of these, aspirational use cases and applications that everyone talks about are difficult to achieve, right? And so I think, rather than, the policymakers. simply subscribing to, let's say, the fantasy and the fear. Of these technologies, there needs to be more of a grounded, strategic as well as a tactical, assessment of what they're looking at, when it comes to quote unquote ai. And that needs to be unpacked very, very well. And it's not always the case that you're married to whatever buzzword is prevailing at a particular time. because about a year and a half ago, there was all this talk about the GPU, right? And as you and I have talked about, and Carl have talked about on this show, it's no longer about the GPU or the chip. It's about the system, and we talked about the importance of interconnect and networking. This is something that Chinese, in particular, Huawei is very good at. And guess what? At a system level, they have been able to innovate and do some very interesting things on top of some very novel approaches to the actual AI software. that is not part of necessarily what we're seeing in the portfolio and the approaches that, AI companies, I hate using that term, AI compute companies, but you know, the AI ecosystem in the west is using, in particular the United States. And so, there's a lot to navigate here and a lot of complexities. And the funny thing is we're you see a lot of policy makers still fixated on quote unquote, the chip. You still see it.

Jim McGregor: 22:40

Right. and Jensen Wong probably made the most accurate statement that the new univ compute is the data center. So that system you talk about isn't just the chip. It is everything that goes into, it's even the power. It's the cooling. It's everything cooling. Yeah. Yeah. It is in giant. It is a giant server, if you will. And you're right. It is. And policymakers, I think, really can't understand the grasp of a how quickly this technology is moving. All of the technology, and the demand that really exists out there for it. What We let the rabbit out of the box. It's not slowing. And you're right, it's still immature. We're still not seeing the adoption at the enterprise level that I think we should have, but. The consumer level, the pro user level, it's off the charts. We projected a nine, we just finished our, forecast update. We projected a 19 x increase from, 2024 to 2024, or during 2024. Mm-hmm. In just token consumption. Mm-hmm. It was over 500 x. Over 500. That's crazy. And that's just tokens. That doesn't include the image and video creation. and not to mention the, the TCO, the cost of these systems and to run these systems in these solutions. Once again, I'm not gonna add these solutions. it's huge. It's massive. heard a stat the other day that was talking about chat, GPT. And saying, just people saying thank you to chat, GPT is costing millions because it's doing inference processing every time you type something in. So consumers and users are finding ways, very much like the internet, finding ways to use it and really driving that level of innovation, driving demand just through the roof. there is no stopping this. And I think anything that they do to get in the way is really gonna hamper that.

Leonard Lee: 24:39

Yeah. I agree with you. But again, going back to the point I made earlier, the association of the GPU. Okay. In very simplistic terms, to somehow, this idea that is the sole factor in influencing quote unquote AI leadership. Okay? And the reason why I bring this up in simple terms is because that's how most people think about these things, including policy makers. That is not good enough. I think that's really the point we're trying to make here. Right. And like even, you're mentioning here of the exponential growth of tokens. Well, yes. You're gonna see that, especially as you see this transition toward reasoning models that are using long thinking, it's, they're going to generate. A lot more tokens because they're, theoretically thinking, for a particular request, they're gonna this stuff doesn't come for free. It's not a one shot inference that is doing it's inferencing for a very, very long time through this auto regressive process. and that's also separate from compute. And that's the other thing that the folks at Nvidia reminded us, right? not every token is, created equal. depending on the system that you have, you're gonna be using either more or less compute for a token, depending on, what sort of application you're running. So there's a lot of, Layers to peel back and understand. And that's one of the biggest challenges, I think, not only for policymakers but for the industry as well. Because to your point, this stuff is moving so quickly. There are folks in the industry that don't even really know where we are with things. You know what I'm saying? Other than maybe those leading edge folks, who have esoteric knowledge about what's going on, who are some of the folks that we bump into and talk to all the time. But that unpacking outside of that community, that small, group of. Folks, it is not a trivial, task. And I think what I'm trying to do here is highlight, in particular to policymakers, the scale and the level of complexity of, what they need to do. To, arrive at good policy, right? You do have to navigate, through all this complexity and then synthesize things back to, what are those key drivers that are gonna shape your policy? And so for anyone to sit there and say, Hey, I understand this stuff, and I can come up with a great policy based on these numbers that somebody's shown me, or a couple of bullet points that someone's, Present to me. I don't think that's enough.

Jim McGregor: 27:24

I have that be very, very thoughtful of this stuff. I have the answer, and you can use a disclaimer on this one. We're talking about how you can use AI to make you more productive, productive and or replace jobs. We use AI to replace the policy makers. Well, that's what they're looking at doing in, UAE, right? Absolutely. Let's get rid of'em all together.

Leonard Lee: 27:47

Yeah. Web, what do they call that? it's a crypto concept. It's a Dao, right? Disaggregated something organization. Yeah. Yeah. Yeah. All this stuff is tied together in a weird way though. I remember. no, seriously, back in 2015, I forgot the name of that one, company. This is when I, where was I? I was in, Barcelona and this was when blockchain was a really big deal. and, all the crypto stuff was starting to go ballistic. I forgot the name of the company. but there was An ai, humanoid, what do they call it? animatronic, doodad that, a robot was going crazy, right? Yeah. Early robot. Yeah. but I, I recall, the inventor, Mentioning that tokens was going to, were gonna be the basis for ai. And so it's really interesting. I, I'll have to add, that individual on the notes, but, yeah, that, that was pretty good foresight on his part. So yeah, a lot of the stuff is related and, there's no, it shouldn't be really that much of a. a stretch to, associate Nvidia with a lot of these trends as well, because, GPUs were used for were, I would say, or maybe even still used for, crypto mining, right? Mm-hmm. So. And then you had the metaverse, and then now we have all this generative AI stuff. So no, now we have Agentic ai agentic. Okay. You and your agentic ai. So

Jim McGregor: 29:20

reasoning ai. Okay. Let's call it reasoning

Leonard Lee: 29:22

ai. Yeah, I think that's more, look, I got really confused, at GTC and, NVIDIA's use of Vient ai. I think it means something different. For them then what you're hearing outside of, uh, GTC or the NVIDIA ecosphere?

Jim McGregor: 29:40

Well, agentic AI should mean not only the ability to reason, but also the ability to make decisions.

Leonard Lee: 29:47

and then actions. Right, right, right. Because, that action part is really where A lot of folks heads are at with the agent ai, right? Well, and that you're closing the loop, you're trying to close the loop and have autonomous automation. I think, the AI community, especially the one around, Nvidia, thinks of it more in terms of the long thinking.

Jim McGregor: 30:11

well, it's also for physical ai, it's going to enable the autonomous machines. It's a critical step for especially humanoid robots, right? Yeah. Which we're seeing a huge amount of innovation in at this point in time. So that's a key message here is the fact that our industry is working together. Our industry is working effectively, and these, this trade war is just inhibiting, innovation. And it's gonna inhibit GDP, it's gonna inhibit markets, it's gonna inhibit consumers. It's a bad situation.

Leonard Lee: 30:44

Yeah. But you know, Jim, I think we do have to recognize, when we do look at the geopolitics where we do have conflicts going on around the world, national security, is a legitimate concern, right? Mm-hmm. It's, I think what you and I are trying to say is that, just make sure that you don't shoot yourself in the foot and that you don't, buy into. Unintended consequences, which oftentimes happens if you don't go through that process that we outlined earlier, right? Where you're unpacking things to a degree that you should, and you're doing due diligence, but then you're also synthesizing that back down into, key factors and drivers that should. help you shape a good policy or end a good strategy, right, as a

Jim McGregor: 31:35

policymaker. And the only people, or the only way we can increase security, not just for us, but for everybody around the world, is through our industry. Our industry is the only one that can achieve that. And. I think we've actually done a very good job of that. we still have hackers. We still have continued evolution of threats. and, we're even entering the quantum era where we're gonna have new types of threats coming out. But, our industry has continued to increase in addressing or risen to address those threats as a whole. So, I still think it should be a hands off policy.

Leonard Lee: 32:13

Okay. All right. I think it should be balanced. Maybe you'll agree. Hands off is like, to me, I don't know if that's going to be something that, you know,

Jim McGregor: 32:22

really

Leonard Lee: 32:23

subscribed

Jim McGregor: 32:23

to. I know to walk away from a policymaker when they put a hyphen in semiconductor, or when they say silicone instead of silicon. Yeah. Um, you know, or silicon,

Leonard Lee: 32:37

you know, who would've known that there were so many ways of pronouncing silicon?

Jim McGregor: 32:41

not silicone.

Leonard Lee: 32:43

Silicone is something different. Something very, very different.

Jim McGregor: 32:45

we need, hopefully our industry can influence the policy makers, but at the same time, our industry needs to react and understand that, things are going to change. And, that globalization, that partnership is still very important to our industry. Absolutely. Yeah. And, cooperation and collaboration. So, I really truly hope that this, I mean, the six G working group just started its efforts, About four or five weeks ago. Yeah. So I would really hate to see that dissolve into regional

Leonard Lee: 33:17

conflicts. Which, and, but a lot of it could be just, driven by, leadership, technological leadership. Yes. And, actually market leadership. So when you look at all the different scenarios that you would use 5G advance for the market that is leveraging the technology and expressing the standardized technology in the most progressive way, I think is gonna be the country, that shapes what the next generation's gonna look like, and whether or not it, the whole project, starts to fragment and bifurcate that's a different, problem that has to be solved. But I think, as GSMA and all these, and 3G PP, look forward, and even the ITU, how do you anticipate the, fragmentation that may happen and, I. Reconstitute this, commitment and understanding of the importance of, the global standardization of technologies, right? I think that's gonna be a big challenge going forward, because there's so many forces that are moving toward breaking everything apart, right? So, but hey, I. We got so serious. This is what happens when you talk about tariffs and we're supposed to be a fun show, Jim. Hey, this is the most critical topic in our industry today. Have to be serious on us. Right? Okay. Yes. All right. But I know that you were really excited and happy About something that you saw embedded world. So I'm gonna give you an opportunity to talk about that really quick.

Jim McGregor: 34:53

Actually. embedded world I thought was, a great show. It really highlights the entire. Semiconductor ecosystem from top to bottom and from, the smallest embedded applications using, a really small microcontroller all the way up to, server-based type applications. There was a lot of innovation going on there. Obviously AI was a huge topic, but even seeing companies like tis getting back into microcontrollers and claims that they have the smallest micro control in the world. A and d on the other end is introducing their turon. Epic platform into their embedded segment only six months after they introduced it to the commercial server segment. And They plan on matching that in the future. So it's gonna be not just for. The standard servers or enterprise class servers or data center servers, but it's gonna be available for embedded applications from day one with the extended temperature ranges, with the extended life cycles and everything else. Really bringing, if you want to call it embedded, you want to call it IOT, whatever you want to call it. Bringing all this stuff up to being very, very important. and, startup there in a terra had a quantum. Pro or, sorry, a neuromorphic processor. Very impressive. It, was capable of about a 40 parameter model in the size of about a microcontroller. Very, very small. Yeah. Very impressive. Just synaptics. Microchip, Silicon Labs, on and on and on. Just the level of innovation we're seeing because of new technologies. Yeah. in terms of semiconductor technology, in terms of chips, in terms of AI software, all these solutions. it's going to be, fun to see because we've talked about AI at the edge. And it's kind of been slowly moving there, but I gotta tell you man, it is reaching a fevered pace and the level of innovation that's enabled both from the semiconductor technology and the software technology, especially in terms of ai. Is really there now and 2025 is going to be exciting year. And we're gonna see, it's funny'cause we're trying to do our forecast for, what you need to do AI processing and everything else. And let's face it, there are like 7 billion smartphones in our handsets in use around the world. Today, small fraction of them can, do on device ai. But think about by 2030 the, you're gonna have the vast majority of those devices. And not just those, but your other consumer devices, your PCs, your pretty much everything capable of doing ai. So, within the next five years we're the level of innovation we're going to see in. The plethora of electronics applications we have out there is gonna be phenomenal. And, seeing the semiconductor ecosystem rise up to be able to support that is really incredible.

Leonard Lee: 37:38

Yeah, no, that's cool. I like to remind the audience, or tell the audience if they don't know that there's been a lot of effort, an investment going on in, AI or what they call Edge AI these days. you might be familiar with a tiny ml, tiny ml in the foundation. now it's called Edge AI Foundation. this has been an area of development RD and innovation for quite some time. And, it's good to see that you're seeing a lot of progress. yes. because I think power efficient compute, scalable, I. compute, meaning scaling down and the model scaling down has been a, those have been the primary pursuits within, that particular sector, the embedded sector. And, I think what we're seeing, with the work that's been happening in mobile as well as in the data center now. a lot of, new possibilities for quote unquote broader edge computing. So, edge infrastructure,'cause you've mentioned what AMD's doing, but we also, know what Qualcomm's doing there as well. In terms of bringing AI accelerators. to edge infrastructure, but then also across, all varieties of, endpoint devices and so, yeah, a lot of exciting stuff. And so it's good to hear, that you are witnessing a lot of, advancement and traction there.

Jim McGregor: 39:01

yeah, and, a lot of the focus on the solution. Obviously Qualcomm was very excited, mobile World Congress about their acquisition of Edge Impulse. So bringing that expertise that Zach and his crew have from Edge impulse into Qualcomm. really enabling a full solution, hardware, software solution for edge applications, that's gonna be critical. and it is incredible. I think we still have a few hurdles in terms of compatibility, everybody Developing their own tools and not really creating standardized solutions out there in the market, but I think that's gonna work its way out. And just the ability to use and leverage open source models is driving a lot of innovation.

Leonard Lee: 39:43

yeah, my recommendation to the industry, ultimately, at the end of the day, middleware, don't be afraid of it, it's just what it is. Don't wait for standards. they ironically can. Stymie innovation, application level innovation. It's like if you need something to happen, and you need things to talk to each other, middleware systems, innovation. Make it happen. That is what happens. And it's just a matter of how much it's gonna cost, but, if the solution's worth it, it'll work itself out. the middleware as well as the integration work will pay for itself.

Jim McGregor: 40:23

I think we're seeing a lot of innovation in a lot of these iot applications. We're seeing, obviously, innovation in consumer applications. Mm-hmm. We're seeing innovation in automotive. Unfortunately, it's all in China. I'm serious when I say that, unfortunately. Yeah. That is sad to say that the US and the other Western countries are so far behind. we're seeing innovation obviously in, not just what we're doing with AI and the data center, but the data center construction itself. Going to more modular solutions. And anything from the racks to the, power solutions. To, even portable data centers where you can actually just have a portable building and just multiply buildings to create a data center. and so we're seeing a lot of innovation across the industry and it's, it is an exciting time.

Leonard Lee: 41:09

Yeah. Lenovo bringing, their Neptune, warm water cooling architectures and systems to, edge appliances. I thought that was interesting.

Jim McGregor: 41:18

Yeah, actually, there was also a company, and I'm gonna have to go back and check the name, but, it was at the Intel booth. It was one of the Intel partners at Mobile World Congress that was showing off a water cooled ran, so a poll based radio access network that was fully water cooled.

Leonard Lee: 41:34

Yeah. I think it, the name. Was that at the HP e booth? I know. No, it was at I it was at the IBM booth or the Intel booth. The Intel booth, sorry. Intel booth, right? Yeah. I bumped into a company, three years ago that was doing that kind of stuff. is it direct cooling or is it like that, semi immersive. No, it was direct.

Jim McGregor: 41:57

It was two chip. It was, and it, you don't think about it, but you know, here's systems that have to sit out in the elements all the time and they have to deal with the dust and everything else. Well, if you have liquid cooling, you can create a fully encapsulated, sealed environment. Yeah. You don't worry about all those conditions. So it, it is amazing, the level of innovation that's going on in the industry and we're seeing. Some of that, innovation at the edge go up to the data center and some of that innovation in the data center comes down to

Leonard Lee: 42:23

the edge. Exactly. you can see it happening and that cross pollination, I think is a super interesting dynamic. that, we get to witness,

Jim McGregor: 42:32

we should have embedded world and super computing at the same time.

Leonard Lee: 42:38

be too big of a conference. Right. we might get a bit of that when we, go to Taipei, next month. Comtex, right? Comtex. Because, believe it or not, they have a. Pretty wide variety of stuff there, including a lot of fans. Lots of fans.

Jim McGregor: 42:58

Yeah. you seem to

Leonard Lee: 42:58

be a fan of fans. Okay. Wonderful. So I think with that, we'll call it a wrap.

Jim McGregor: 43:04

Is that good? I would just like to say that there's a lot of stuff coming up, you know? we had TSMC Symposium this week we're gonna have Intel Direct Connect. So you got two major foundries talking about next generation, semiconductor process technologies. TSMC was talking about their a 14 process. This week we're gonna have IBM think next week we're gonna have cadence live coming up. We're going to have RA security conference. we're gonna have compute, text, a lot of stuff going on in the industry, a lot of things coming up. So look for more of that. From us together. Yes. And individually. And, we developed a brand new, ai, data center inference t forecast and TCO model that we're gonna be introducing here the next couple weeks. there you go. And we're gonna be doing our best for everyone out there to cover this. I know we got a little serious on this one, but I know, so serious. We, we had to, if you know a regulator, please send this link to them and tell them to stop messing with our industry.

Leonard Lee: 44:08

yeah. If anything, learn more about it and, be thoughtful. that's, what I would say. But, yeah, definitely. I think they should engage with us. Yes. We're gonna help policymakers quite a bit. so, yeah, with that, Hey Jim, real fun. I think we need to bring levity to our next episode, and hopefully we'll have Carl joining us, but, geez, its so hard to, well,

Jim McGregor: 44:34

we're gonna be in compute text, a couple margaritas in. I'm sure we'll have a lot of levity to bring.

Leonard Lee: 44:39

Yes. And Taipei. Taipei is the perfect place to get a margarita. Ironically it is. Okay. So, yeah, lot of exciting stuff coming up folks. So, make sure that you continue to, tune into our podcast here. So, remember to like, share, comment on this episode. Subscribe to the Rethink podcast here on YouTube and Buzz, and, take us on the road on your jog. And listen to us on your favorite, podcast platform and also, check out, Jim McGregor and the folks at TEUs Research, good partners of ours here at ncur. And, yeah, follow them. Jim is one of, if not the leader, in the analyst community. And, just a brilliant, partner to have, but also a great friend and a tremendous resource for the industry. you can go ahead and say it. Just say the Jim's the uber geek. you are, a super nerd. Okay. Happy now. Okay. Yeah. and then also, subscribe to the next curve research portal at www. next dash curve for the tech and industry insights that matter. We'll see you next time.