The neXt Curve reThink Podcast · 2026-05-27 · 40 min
Key moments - from our scoring
Substance score
48 / 100
Five dimensions, 20 points each
The episode examines three pivotal industry announcements shaping the semiconductor landscape. Leonard Lee, Karl Freund, and Jim McGregor analyze Qualcomm's confirmed ByteDance partnership for millions of ASICs targeting agentic AI inference - marking Qualcomm's breakthrough into hyperscale deployments with rack-scale solutions expected to be announced at Computex. The discussion reveals concern about geopolitical entanglement similar to what NVIDIA and AMD face, though inference workloads and Chinese government restrictions on competitor hardware may create regulatory gaps. Huawei's 1.4nm Kirin chip announcement generates skepticism about claims versus reality, but the experts stress Huawei's genuine innovation capability in 3D stacking, novel interconnects, and packaging technologies that sidestep traditional geometric scaling limitations. Finally, Cerebras' IPO success validates wafer-scale computing's market potential, though the hosts explore fundamental challenges around on-die SRAM memory constraints, MemoryX external streaming for training, and inter-wafer IO bandwidth bottlenecks that define the architecture's competitive positioning.
It marks Qualcomm's first confirmed hyperscale customer for rack-scale ASICs after years of limited data center traction with the underpowered AI 100. The deal signals a pivot toward agentic AI inference with substantial CPU and accelerator customization, with additional hyperscaler announcements expected at their June investor day.
The 1.4nm figure represents an aspirational time-based target around 2031, not a current geometric achievement. Huawei is constrained by lack of EUV and High-NA lithography, but compensates through 3D stacking, novel interconnects, and packaging innovations - focusing on performance-per-watt for domestic Chinese markets rather than competing globally.
It puts Qualcomm on government radar for China-facing AI accelerators, similar to export restrictions NVIDIA and AMD face. However, inference workloads may fall outside current export control frameworks designed for foundation model training, creating regulatory uncertainty.
Wafer-scale designs offer fast on-die computation but face two constraints: insufficient SRAM memory relative to model sizes (requiring MemoryX external streaming for training) and severe IO bandwidth bottlenecks, since internal dies lack the 'shoreline' to route IO like edge dies do.
Innovation now spans four pillars: transistor design (2D to FinFET to Gate-All-Around), materials science, lithography, and packaging/3D stacking. Geometric node scaling has slowed, but lithography still impacts design constraints; packaging now drives performance gains in three dimensions rather than two.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode mixes a handful of genuine data points (Cerebras token throughput, Infiniband growth, Nvidia ARM server share) with substantial padding and surface-level reactions. The Cerebras layer-parallelism explanation is one substantive technical insight, but many topics are touched at headline depth before moving on with 'we'll have to wait and see.'
they take each layer of the network and they parallelize at that level. So the inner layer communications aren't that demanding. So That gets them around their IO problem.
80% of that growth is roughly is NVIDIA
Most takes are standard industry-analyst consensus: Huawei is innovative (well-established view), memory is hot, hyperscalers diversify silicon. The mild reframing of Moore's Law and the inference-vs-training export control nuance show some independent thinking but nothing genuinely contrarian or first-principles.
really changing, Moore's Law from millimeters squared to millimeters cubed
a lot of the policies are geared more toward, concerns around, foundation model training versus inference
All three participants are independent semiconductor/AI analysts with genuine domain depth and industry access (Freund called the Cerebras CEO directly), but none are operators who built or scaled the products being discussed; this is a panel of knowledgeable watchers, not practitioners.
I called Andrew Feldman, the CEO, and said,'How'd you do this?'
I've got a tracker now, and, like at least once every two weeks, there's a new announcement...I think I've got over 60 names on there already, and that's just over the past three years
The episode has a better-than-average density of named figures, company references, and hard numbers - Cerebras throughput, Infiniband 4x growth, Intel packaging revenue, memory vendors' trillion-dollar caps, Nvidia CPU TAM figure - though many are dropped in quickly without source attribution or fuller context.
running Kimi 2, 2.6.
that's a trillion parameter model...doing it 1,000 tokens per second...about six and a half times faster than the fastest GPU cloud
The host's questions are largely open-ended topic launchers ('what do you guys think?'), and genuine pushback is rare; the two moments where guests correct each other (GlobalFoundries history, ACIE growth rate) are the episode's best conversational moments but are brief and underdeveloped.
it's growing faster for Nvidia. I don't know that it's growing faster overall. I mean, still, the vast majority of the, the money, the CapEx that's going into spending for AI data centers is still by the hyperscalers.
I would warn you- Yeah that, you have to remember that they made a similar commitment to GlobalFoundries a couple years ago and pulled out of it.
Computed from the transcript - who did the talking, and the words that came up most.
Send us Fan Mail Silicon Futures is a neXt Curve reThink Podcast series on AI and semiconductor tech and the industry topics that matter. This month, the AI narrative has rediscovered its mojo with hyperscalers continuing to double down on their CapExc on AI infrastructure, while realizing unrealized "other income" from their investments in AI startups and labs as the likes of Anthropic, OpenAI, and xAI prepare to go public. We didn't cover it in our discussion, but yes, Elon Musk lost his case against OpenAI. Too bad. Regardless, another action packed month in May 2026. In this episode, Leonard, Karl and Jim talk about some of the top headlines from May of 2026. ️ Qualcomm's mystery hyperscaler customer? ️ The geopolitical risk of playing the AI game ️ Huawei teases "1.4 nm" chip and novel architecture ️ Cerebras finally IPOs! ️ NVIDIA's Q1 FY27 results and new reporting structure ️ Memory vendors are now $trillion market cap ️ Will Apple use Intel Foundry for Apple Silicon? ️ The state of Intel Foundry business ️ Alex Katouzian joins Intel to head up Client Computing Physical AI ️ AMD proving they are more than AI accelerators ️ Is the AI infrastructure boom driving IoT revival?
Transcribed and scored by The B2B Podcast Index.
1 - > Karl Freund: Next curve 2 - > Leonard Lee: Welcome everyone to, uh, this episode of Next 3 - > Curve's Rethink podcast, where we break down the latest tech 4 - > and industry events and happenings in the world of 5 - > semiconductors, and Carl's favorite topic, AI, AI, AI, and 6 - > AI, into the insights that matter. 7 - > I'm Leonard Lee, executive analyst at Next Curve, and I'm 8 - > joined by the illustrious Carl Freund of Cambrian AI Research. 9 - > And we also have agentically modified and augmented Jim 10 - > McGregor of the famed and infamous, a- and, fantabulous, 11 - > right?
12 - > That's another good word to describe- Terius Research. 13 - > Yes, Terius Research. 14 - > Welcome, gentlemen. 15 - > Jim McGregor: Th- this is not the Jim McGregor you're looking 16 - > for.
17 - > No, just kidding. 18 - > Leonard Lee: Yeah, you're in a purple metaverse. 19 - > In this, episode, of Silicon Futures, we're gonna be talking 20 - > about the headlines for the month of May, which has been 21 - > completely insane. 22 - > And before we get started, remember to like, share, react, 23 - > and comment on this episode.
24 - > Also, subscribe here on YouTube and Buzzsprout to listen to us 25 - > on your favorite podcast platform. 26 - > Opinions and statement by my wonderful guests here are their 27 - > own and don't reflect mine or those of Next Curve. 28 - > We're doing this, for informational purposes only, to 29 - > provide an open forum for discussion and debate on all 30 - > things, AI and silicon. 31 - > So with that, wow, g- guys, it's just an- I mean, this is, like, 32 - > r- really nutty.
33 - > It's getting out of- 34 - > Karl Freund: You say that every month. 35 - > Leonard Lee: I know. 36 - > Karl Freund: That's the new norm. 37 - > Leonard Lee: I'm gonna kick us off here with, um, this, 38 - > announcement by Qualcomm.
39 - > I'd love to get your reactions to this. 40 - > There's this report, from Bloomberg, that mystery, uh, 41 - > hyperscaler that, Qualcomm mentioned on their earnings call 42 - > this month, is ByteDance. 43 - > Karl Freund: Mm-hmm. 44 - > Leonard Lee: looks like there is an alleged, uh, deal for, 45 - > quote-unquote,"millions of ASICs", uh, to support, uh, AI 46 - > requirements for their social platform, so this is largely 47 - > inference, right?
48 - > agentic AI. 49 - > Uh, so what do you guys think? 50 - > I- is this- 51 - > Karl Freund: I think it's fi- finally happened. 52 - > I mean, Qualcomm's been in AI on the device for a long time.
53 - > eight years in production, I think, maybe nine. 54 - > Leonard Lee: Oh, yeah. 55 - > Karl Freund: and they dabbled in the data center. 56 - > It got some ODM traction with Dell and HP, with the Qualcomm 57 - > AI 100.
58 - > But it was really underpowered for today's AI. 59 - > Mm-hmm. 60 - > it was designed for yesterday's AI. 61 - > Now Qualcomm has what we all believe they will announce 62 - > probably next week at Computex the next generation of that, 63 - > which is rack scale.
64 - > Mm-hmm. 65 - > if you wanna start fresh with new AI models for agentic AI and 66 - > a lot of CPU power, that seems to be where they're heading. 67 - > Mm-hmm. 68 - > And ByteDance fits 69 - > Jim McGregor: Well, and they showed off their first rack 70 - > scale solution at, Mobile World Congress.
71 - > it's a huge rack. 72 - > it's a stan- it is a standard Really tall, right but it's 73 - > really tall. 74 - > It's standard than, what I would consider a normal rack bait is. 75 - > It is an industry standard still.
76 - > So no, it, it's gonna be interesting. 77 - > first off, they, they've kinda hinted that they're talking to 78 - > more than one hyperscaler or w- more than one- Yeah large 79 - > customer. 80 - > So, uh, I would expect that this is probably the first in many 81 - > announcements, and they planned on making their first 82 - > announcement about their first customer at, their AI inve- 83 - > their investor day, which is later in June. 84 - > So this is kind of, uh, uh, leaking, the first announcement.
85 - > But I'm also a little concerned because, this also kinda puts 86 - > them on the government radar. 87 - > they could very easily all of a sudden be pulled into some of 88 - > those geopolitics that Nvidia and AMD have been pulled into. 89 - > Mm. 90 - > I'm hoping that's not the case, 'cause I really hope that, we 91 - > get rid of some of those geopolitical issues.
92 - > but I think it's a great I- if it is true, I think it's a great 93 - > win for Qualcomm. 94 - > Karl Freund: Yeah. 95 - > Did they announce where this would be stood up? 96 - > Jim McGregor: They have not, and that's 97 - > Karl Freund: the interesting part.
98 - > I was wondering if maybe they're gonna do, like Singapore or 99 - > something like that, that would allow them to, perhaps sidestep 100 - > government intervention. 101 - > So I 102 - > Jim McGregor: The government already opened up the prospect 103 - > of selling H200s and, AMD products That's true but the, 104 - > the Chinese government has said, "Absolutely not, we don't want 105 - > them in the country." 106 - > Leonard Lee: Yeah. 107 - > Jim McGregor: Yeah.
108 - > Leonard Lee: And I think a lot of the policies are geared more 109 - > toward, concerns around, foundation model training versus 110 - > inference, right? 111 - > True. 112 - > And, and the systems- are different. 113 - > And, and, you know, if you recall, when they first started 114 - > the bans, and NVIDIA introduced they couldn't package HBM into 115 - > them, right?
116 - > And so the memory content is different. 117 - > I think the whole, let's call it security control policies, might 118 - > have to evolve, Mm-hmm if inference is as much of a 119 - > concern as foundation model training. 120 - > And, if you read some reports, y- H100 is great for model 121 - > training, and so where does that put the older generation, GPUs, 122 - > right? 123 - > And those can fall under the radar, or at least the current 124 - > one, right?
125 - > Uh, from a policy perspective. 126 - > So yeah, it's interesting. 127 - > I guess the bigger question for me is what do they mean by ASIC? 128 - > And, is it NPU or, is it gonna be a custom job?
129 - > that was the thing that I thought was, a head scratcher 130 - > from the announcement. 131 - > Jim McGregor: Qualcomm has indicated that they are 132 - > interested in doing custom- Custom CPUs and custom 133 - > accelerators. 134 - > And that they, they hope to be doing both. 135 - > So y- it could be one or it could be- One of those one or 136 - > both.
137 - > We'll have to wait and see. 138 - > Leonard Lee: so yeah, it's, that was an interesting announcement. 139 - > And then, of course, I think the other one that caused quite a 140 - > stir this week was Huawei's 1.4 nanometer Kirin chip, or at 141 - > least it's been characterized as 1.
4 nanometer in the headlines, 142 - > but that's, that's a claim that they might be able to achieve an 143 - > equivalent, if you will, in, in 2031 or something like that. 144 - > Jim, what were your impressions of that? 145 - > Jim McGregor: I take all of that with a grain of salt at this 146 - > point in time. 147 - > I still think that they're gonna be limited on how far they can 148 - > go with their current lithography technology.
149 - > and they're still gonna be several years behind at best, 150 - > even with that, and not having, especially not having EUV and 151 - > High-NA. 152 - > So it's gonna be a challenge. 153 - > But with that said, they are a very innovative company. 154 - > They continue to find ways to be competitive, and they always 155 - > have.
156 - > So, you know, even if they go to electron beam or some other 157 - > technology, I'm sure that they will come up with something 158 - > that's gonna at least fit their needs and make them competitive 159 - > in the marketplace. 160 - > And you have to remember that Huawei's not gonna be selling 161 - > these chips to the global market. 162 - > They're gonna be selling it in China, and they're gonna be 163 - > using it themselves, selling it in China to other vendors, and a 164 - > select other countries.
165 - > it's not necessarily a global competitor to what we already 166 - > have from the Western companies. 167 - > Karl Freund: The government's made it a little bit easier for 168 - > them and their compatriots to, have a competitive platform in a 169 - > country where there's really no outside competition coming in, 170 - > right? 171 - > Mm-hmm. 172 - > And the bar is lowered, however.
173 - > Sound- sounds like they are making, good progress- Yeah and 174 - > will satisfy the needs of domestic Chinese-based 175 - > hyperscalers. 176 - > Leonard Lee: I think a lot of the media is latching onto this 177 - > 1.4 n- nanometer, data point that was shared by the head of 178 - > HiSilicon. 179 - > but it's this idea that you're not looking at geometric 180 - > scaling, you're looking at time scaling, right?
181 - > That's 182 - > Jim McGregor: exactly it. 183 - > Leonard Lee: So this whole tau- We 184 - > Jim McGregor: stopped geometric scaling, or we stopped tying the 185 - > numbers to geometric scaling about a decade ago. 186 - > Leonard Lee: Yeah. 187 - > Jim McGregor: So- 188 - > Leonard Lee: And this is what I think is really interesting.
189 - > I, and'cause, a lot of people still think that Huawei just 190 - > steal stuff, they're not innovative, they're not... 191 - > they're copying everyone else. 192 - > I still hear this all the time. 193 - > No, they are a very innovative company I think it's extremely 194 - > dangerous, yeah, extremely dangerous to assume that these 195 - > guys are a bunch of idiots.
196 - > they're not. 197 - > And, they're also not constrained to, legacy 198 - > architectures, which- 199 - > Karl Freund: Mm-hmm 200 - > Leonard Lee: you could argue, the current semiconductor 201 - > industry is stuck with a legacy right now. 202 - > As you mentioned, geometric scaling has been, is pretty 203 - > much, um... 204 - > Now you're essentially stacking anyways, right?
205 - > And so these guys are doing, 3D stacking and some novel, I don't 206 - > know if you wanna characterize it a networking or interconnect, 207 - > to, address tau, right? 208 - > reducing the signaling, latency, to get more performance, right? 209 - > They're looking at the problem more holistically and finding an 210 - > alternative, approach that ecosystem will support. 211 - > But the question is, how long will it take for them to 212 - > actually scale this thing economically and in terms of 213 - > production as well, right?
214 - > Jim McGregor: you- That's the big question you have to 215 - > remember that there is a lot of innovation in semiconductors 216 - > still to be had. 217 - > we traditionally had three main pillars of innovation- Mm and 218 - > those were the transistor design, which we've gone, from 219 - > 2D planar to 3D FinFET to,, Gate All Around, technology now. 220 - > there's materials technology, which continues to change. 221 - > They're basically- Yeah using the entire periodic table to 222 - > come up with- Yeah new solutions, and it varies by 223 - > generation to generation.
224 - > Matter of fact, that's one of the things that kind of tripped 225 - > Intel up starting about the 14 nanometer generation. 226 - > but you also have the lithography technology, and that 227 - > was the geometric scaling, really tying it to the gate 228 - > length, which we stopped doing that. 229 - > But it still imp- we are still scaling it, and it still impacts 230 - > your design constraints in terms of how you design, features 231 - > within the chip. 232 - > And that is- Yeah it is still scaling.
233 - > I wanna make that clear. 234 - > It is, we are still scaling out with each generation, just not 235 - > the same rate that we used to be. 236 - > Yeah. 237 - > And now we have a fourth one, which you mentioned, and that is 238 - > packaging technology, where we're stacking- Yeah die.
239 - > and we're doing things with the substrate. 240 - > We're doing things on the base wafer. 241 - > We're doing, all kinds of very innovative things that are 242 - > really changing, Moore's Law from millimeters squared to 243 - > millimeters cubed. 244 - > To your point, Huawei is a very innovative company.
245 - > They've always been highly competitive. 246 - > And you have to remember that it's not just Huawei. 247 - > It's a Chinese ecosystem. 248 - > They have government support.
249 - > Yeah. 250 - > They've got, semiconductor foundries with SMIC and others 251 - > over there. 252 - > SMIC, right, yeah. 253 - > So it is an entire ecosystem that's supporting this.
254 - > It's not just Huawei. 255 - > Leonard Lee: Yeah. 256 - > And it, it's surprising that they're gonna be, targeting, 257 - > chips, based on this process, for the P90 smartphone- Mm-hmm 258 - > which I thought maybe they would go for something less demanding 259 - > than a smartphone SOC. 260 - > But yeah, yeah, 261 - > Jim McGregor: Smartphones are always on the bleeding edge- 262 - > Mm-hmm along with the, PC processors now, the data center 263 - > processors, and the- Yeah AI accelerators.
264 - > All of those are pretty much on the bleeding edge at this point 265 - > in time. 266 - > Leonard Lee: So, Cerebras. 267 - > Karl Freund: Yeah. 268 - > It's, it's 269 - > Leonard Lee: ideal Oh, you got to have something to say about 270 - > Jim McGregor: Oh, yes.
271 - > Karl Freund: Well, of course, I've been a big fan of Cerebras 272 - > since- First off, congrats. 273 - > Leonard Lee: Huh? 274 - > Karl Freund: First off, congrats. 275 - > Yeah.
276 - > Exactly. 277 - > Congrats to the whole team, and their customers. 278 - > They've really, grown enough that the public market was very 279 - > receptive to this IPO. 280 - > Mm.
281 - > Stock skyrocketed. 282 - > it's fallen back somewhat, since then. 283 - > It's up a little bit more today. 284 - > But bottom line is th- they're the only guys in town that have 285 - > wafer scale computing.
286 - > Leonard Lee: Yeah. 287 - > Karl Freund: And there's, there's good things and bad 288 - > things about wafer scale. 289 - > I think most people get the good stuff. 290 - > you, you- That's big eliminate all the switching and all the 291 - > rack-based architectures to get a really fast, solution to 292 - > market.
293 - > But there's a couple of issues you have to deal with, and it's 294 - > interesting to see how well Cerebras are dealing with those 295 - > issues. 296 - > So one of the issues is not enough memory, okay? 297 - > It's all SRAM, kind of like Groq. 298 - > But instead of Groq, it's like Groq, right?
299 - > so you get a lot more SRAM per mo- for the model to run on die. 300 - > Yeah, I guess it's still die on wafers. 301 - > But once you go beyond the size of the memory, you need to 302 - > either offload the weights to a streaming device, primarily for 303 - > training, and that's called MemoryX. 304 - > also you have to go off chip more efficiently, If you think 305 - > about, I- IO is a, peripheral problem, right?
306 - > And pun intended. 307 - > you have to use the shore line of the chip to send wires out 308 - > for IO. 309 - > Well, what about all the chips that are inside the wafer? 310 - > Well, there is no shoreline.
311 - > Zero. 312 - > Jim McGregor: Right. 313 - > Karl Freund: And so a lot of people complain that,"Ah, 314 - > Cerebras gonna have a problem scaling wafer to wafer, just 315 - > because they don't have enough IO bandwidth." Well, that's 316 - > absolutely true, but there are ways around it.
317 - > Leonard Lee: Mm. 318 - > Karl Freund: and Cerebras has done a pretty good job of 319 - > getting around it. 320 - > In fact, last week, they came out and said,"Hey, look, we're 321 - > running Kimi 2, 2.6."
Well, that's a trillion parameter 322 - > model. 323 - > Leonard Lee: Mm. 324 - > Karl Freund: and they're doing it 1,000 tokens per second- in 325 - > enterprise trials right now. 326 - > Well, that's about six and a half times faster than the 327 - > fastest GPU cloud.
328 - > Leonard Lee: Yeah. 329 - > Karl Freund: And it runs Claude Op- Opus at 10X 330 - > Leonard Lee: faster. 331 - > Karl Freund: So it's like, wow, they seem to have solved the 332 - > problem. 333 - > So I called Andrew Feldman, the CEO, and said,"How'd you do 334 - > this?"
And they said,"Well, you know, we got some smart 335 - > engineers, and we figured out a way around the problems of not 336 - > having enough memory." And what they did is they take each layer 337 - > of the network and they parallelize at that level. 338 - > So the inner layer communications aren't that 339 - > demanding. 340 - > So That gets them around their IO problem.
341 - > And since they're just having one wafer, excuse me, one layer 342 - > of the neural network on a wafer, they get around the 343 - > memory problem. 344 - > So it's pretty impressive. 345 - > Ve- very impressive 346 - > Jim McGregor: And you have to remember, there's still a lot of 347 - > innovation going on. 348 - > First off, it proves that there's not one size fits all 349 - > for AI because- Yeah the models are different, the applications 350 - > are different, the model sizes are different.
351 - > So I mean, that's gonna lead to a lot of different 352 - > architectures, especially for inference processing. 353 - > Yeah. 354 - > But also the fact that, we're seeing a lot more innovation, 355 - > and this is interesting'cause a lot of it was all around the 356 - > models and, going to- especially to- towards transfo- going from 357 - > RNNs to CNNs to transformers, blah, blah, blah. 358 - > Now we're seeing a lot of innovation in algorithms to run 359 - > those models.
360 - > Simple things that they can do and how they run those models 361 - > that can speed them up two, three, four, even 10X faster. 362 - > Matter of fact, even working with some of the startups we've 363 - > seen recently, and NVIDIA, some of the stuff that they're doing. 364 - > They... 365 - > Just through that software layer innovation, they're being able 366 - > to do a lot more than they ever have before, and I would suspect 367 - > a lot of that's also coming into play with Cerebras as well.
368 - > Mm-hmm. 369 - > So I I think that there's still a lot of innovation to be had 370 - > there, and it's not just tied to the hardware. 371 - > A lot of it's still in the software. 372 - > Leonard Lee: When you look at- the size of the models that are 373 - > being used in, let's say, edge AI type scenarios or inference 374 - > in general, the smaller models.
375 - > Mm-hmm. 376 - > you have the whole, tiny ML or tiny AI or smaller AI movement 377 - > going on. 378 - > And so there's that as well that I think changes, the technical 379 - > requirement, situation, for these accelerators, right? 380 - > The demand profile is, always changing or the requirements are 381 - > always, shifting, toward, more for less actually.
382 - > you get more out of less compute, thanks to innovations 383 - > on, model compression and, the performance improvements that 384 - > we're seeing with smaller models, for inference, a lot of 385 - > the requirements are domain or application specific anyways, 386 - > right? 387 - > you don't need a world model to, do certain, specific edge 388 - > functions or, to support, you know, certain edge applications. 389 - > And then also from a safety perspective, you're seeing a lot 390 - > of isolation anyway, so it's nicer if you have a capable 391 - > small model that you can deploy into a container and use, the 392 - > least amount of resources in ex- executing that, that 393 - > intelligence, if you will.
394 - > So- 395 - > Karl Freund: Mm-hmm. 396 - > Leonard Lee: No, that's cool. 397 - > That's good insight. 398 - > Karl Freund: And I think it's also important to remember that 399 - > they're not just one customer shop anymore, right?
400 - > Yeah, yeah. 401 - > G42. 402 - > They have a large deal outside of G42 with AWS and with, 403 - > OpenAI. 404 - > Well, that's pretty good.
405 - > Leonard Lee: Yeah. 406 - > Karl Freund: mean, if you're gonna pick two customers, those 407 - > would probably be two of the ones you'd wanna go after. 408 - > Leonard Lee: hyperscalers are th- are the thing, right? 409 - > I mean, if we're- Yeah gonna talk about Nvidia's results that 410 - > came out, the top hyperscalers are still 50% of their data 411 - > center business- Yep right?
412 - > If not even, if you consider the Neocloud's hyperscalers, which 413 - > they really are, it's even more, right? 414 - > they put hyperscalers in this weird bucket called- A- ICE or 415 - > something, A-C-I-E? 416 - > Karl Freund: ACIE is, is, is for the h- neo- neo clouds and 417 - > enterprise- Yeah and for, Industrial, right? 418 - > In- industrial, right.
419 - > A- as well as c- as well as, the, you know, country-specific 420 - > op- Sovereign stuff data centers being set up The sovereignty 421 - > sovereign data centers. 422 - > That's also in the I- I- AICE- Yeah bucket. 423 - > But they, it's, it, the thing is that's growing so fast now. 424 - > Mm-hmm.
425 - > It's, the, Nvidia's, I think, doing a good job of segmenting 426 - > the revenue so they can show things that they, that counter 427 - > the narrative against them, which is they're just selling to 428 - > five guys. 429 - > they're saying,"No, no, no, no, we're, we're selling to major, 430 - > data centers for sovereign AI, for enterprise AI, and for the, 431 - > neo clouds." Yeah. 432 - > And that's growing really fast, growing faster than the, than 433 - > the hyperscalers.
434 - > And that, the, to me, that was- Yeah the highlight of the 435 - > earnings call. 436 - > Jim McGregor: Yeah well, it's growing faster for Nvidia. 437 - > I don't know that it's growing faster overall. 438 - > I mean, still, the vast majority of the, the money, the CapEx 439 - > that's going into spending for AI data centers is still by the 440 - > hyperscalers.
441 - > Karl Freund: Oh, yeah. 442 - > Jim McGregor: it does point out that, the hyperscalers, if 443 - > nothing else, they point to a key trend, and I think that's 444 - > one of the key things people focused on while, Nvidia's 445 - > facing more competition with, ASICs and everything else, as 446 - > well as the hyperscalers doing their own silicon. 447 - > There's not a single hyperscaler out there that's using one 448 - > solution. 449 - > So I mean- Yeah.
450 - > Yeah I think it's really important to note that they're 451 - > still scaling up Nvidia, they're still scaling up, AMD. 452 - > Now they're using Cerebras, they're using their own silicon 453 - > solutions. 454 - > They're scaling multiple solutions for different AI 455 - > platforms. 456 - > Leonard Lee: Yeah.
457 - > Mm-hmm. 458 - > Mm-hmm. 459 - > Yeah. 460 - > And, the other thing to recognize here with the 461 - > hyperscaler bucket that they have is most of the growth is 462 - > actually networking, so it's- Yeah 3X.
463 - > Karl Freund: Infiniband was up 4X. 464 - > Leonard Lee: Infiniband That's scary. 465 - > Yeah. 466 - > Yeah.
467 - > Infiniband- Yeah. 468 - > And, um- 469 - > Jim McGregor: Can, can you imagine what Nvidia would be if 470 - > they were a memory vendor? 471 - > And, and, and, and I, and I kid you not when I say that. 472 - > I mean, that's, that's one of the other major things in May 473 - > that we have to talk about, and the fact that now all three of 474 - > the major memory vendors are in the trillion,$1 trillion- 475 - > Trillion dollar 476 - > Karl Freund: cap 477 - > Jim McGregor: mark- yeah, market cap, you know, gang That's crazy 478 - > And they all hit it in the same month.
479 - > Yeah. 480 - > So, I mean, that's huge. 481 - > I mean- Yeah Micron, SK Hynix, and Samsung are all, you know, 482 - > above a trillion dollar market cap now, and because of memory. 483 - > And as, as we look at memory, it's funny'cause, middle of last 484 - > year we were thinking, oh, there might be some oversupply and 485 - > everything else, and then just things went crazy.
486 - > And now we're looking at 2028 or beyond before- Mm-hmm we ever 487 - > catch up with memory demand. 488 - > Leonard Lee: Yeah. 489 - > Um, yeah. 490 - > A- and I think, and I always like to point out that 491 - > valuation's one thing, revenues are another thing.
492 - > it's- 493 - > Jim McGregor: But it points to demand, and the demand is key. 494 - > And that's what's really driving it. 495 - > Karl Freund: Yeah exactly, Jim. 496 - > Leonard Lee: Or the perception 497 - > Karl Freund: of demand- That, that's producing a longer- is 498 - > driving the valuation it's producing a longer cycle.
499 - > Yeah. 500 - > So instead of a three-year cycle that memory vendors have 501 - > typically operated in- Yeah two to three years, now we don't 502 - > know how long this cycle's gonna be. 503 - > Some people- Yeah are saying it could be five-year cycle. 504 - > Yeah.
505 - > And so that reduces some of the risk on their stock prices- 506 - > Right as, as well as risk of an impending i- implosion. 507 - > And I tell you, the, financial analysts keep raising targets on 508 - > these guys. 509 - > Jim McGregor: Oh, yeah. 510 - > Karl Freund: They're all expected to basically double 511 - > Yeah, and then- Again, and they've already tripled 512 - > Leonard Lee: Yeah, and I think that's also creating risk 513 - > outside of the data center or the AI- Absolutely 514 - > infrastructure segments, and that's where, we saw that 515 - > announcement, I think, or I don't know if it's a rumor, 516 - > Apple talking to or, inking some deal or coming close to inking 517 - > a, an agreement with Intel for their foundry services, right?
518 - > There's no indication on, which Apple Silicon series, product 519 - > would be the first candidate. 520 - > But, I think that's where even the likes of Apple, which 521 - > historically, or at least for the last couple of, a decade and 522 - > a half, has really, dictated the, the conversation around, 523 - > advanced, manufacturing. 524 - > Now they're in a position wh- or, or in a situation where they 525 - > might have to diversify, right? 526 - > Go back to, the days before they committed to TSMC.
527 - > Jim McGregor: Well, I would warn you- Yeah that, you have to 528 - > remember that they made a similar commitment to 529 - > GlobalFoundries a couple years ago and pulled out of it. 530 - > So, Mm I take everything that comes out of Apple, especially 531 - > in terms of new suppliers, with a grain of salt, because I've 532 - > seen them just tank people and leave people high and dry- 533 - > Mm-hmm with the crystal display, with the Sapphire crystal 534 - > displays- Yeah with the, GlobalFoundries with the foundry 535 - > capacity.
536 - > Apple can be your best and worst customer all at the same time. 537 - > Leonard Lee: Oh, yeah. 538 - > Yeah. 539 - > but y- y- given the situation, don't you think that the 540 - > circumstances are different for them?
541 - > I mean, they're getting crowded out a, a, a bit, right? 542 - > Yeah, but TSMC- If you look at TSMC 543 - > Jim McGregor: results- is adding a lot of capacity too. 544 - > So, I mean, TSMC is adding a lot of capacity. 545 - > They've got several fabs- Mm one at least, already in operation 546 - > in Arizona, another one going up.
547 - > Matter of fact, we know that, NVIDIA's gonna be producing 548 - > products here in the States as well at that facility. 549 - > Amkor is in, in, adding additional packaging capacity, 550 - > as well as TSMC in Arizona. 551 - > So, I mean, there's a lot of capacity going in place. 552 - > And, I - First off, I think it, it would behoove, Apple to 553 - > definitely diversify into other foundries.
554 - > It's not a cheap process. 555 - > It's a very- Yeah complicated process to be able to d- support 556 - > multiple foundries, but I think it's in their best interest to 557 - > do that in the long run. 558 - > But we'll have to wait and see. 559 - > I wouldn't count TSMC out by any means, and if Apple decides that 560 - > they just wanna continue on TSMC, they can, and I believe 561 - > that they would if it comes down to it.
562 - > So it'll be interesting to watch. 563 - > On the other hand, though- Yeah Intel is doing fairly well. 564 - > we expect to see by the end of this year some announcements 565 - > around 18A, and/or 14A customers. 566 - > Yeah and we already know that their packaging group is doing 567 - > exceptionally well.
568 - > They noted north of a billion dollars in revenue just from 569 - > advanced packaging. 570 - > Leonard Lee: Yeah. 571 - > Jim McGregor: and that's out of, Penang, Malaysia, and Rio 572 - > Rancho, New Mexico. 573 - > So that side of their foundry is already cranking, and they won't 574 - > break it out, but I believe they're actually already making 575 - > money through packaging.
576 - > Leonard Lee: And, and there's a lot of news that came out of 577 - > Intel this month, right? 578 - > Mm-hmm. 579 - > One of which w- is, Alex Katouzian is the new- Yeah head 580 - > of, yeah, consumer- I was- consumer and physical AI, right? 581 - > Yes.
582 - > Yeah. 583 - > Consumer computing and, or client computing and physical 584 - > AI. 585 - > He's gonna kill me for getting that wrong. 586 - > Jim McGregor: I'm gonna tell him you got it wrong.
587 - > Leonard Lee: Oh my God, no. 588 - > Karl Freund: For those in the 589 - > Leonard Lee: audience- Please don't. 590 - > He works out who might 591 - > Karl Freund: know who Alex- 592 - > Leonard Lee: he works out Alex- He's very buff. 593 - > Karl Freund: He's very buff.
594 - > Leonard Lee: So. 595 - > Karl Freund: for those who don't know, Al- Alex has been with 596 - > Qualcomm forever. 597 - > Leonard Lee: Oh, 598 - > Karl Freund: yeah. 599 - > And, getting Intel capturing his, his talent and leadership 600 - > s- skills, was a real coup for Intel.
601 - > Leonard Lee: Yeah. 602 - > Karl Freund: I mean, if you look at the Qualcomm team and you had 603 - > to say who's got the most, talent- 604 - > Leonard Lee: Mm 605 - > Karl Freund: on the team, I'd say, I've always thought it 606 - > would be Alex, and sure enough, there he goes. 607 - > Leonard Lee: Yeah. 608 - > No, he's a great guy.
609 - > all the best to him. 610 - > he's, got big shoes to... 611 - > He, well, he has big f- shoes to fill. 612 - > maybe that's not the way of putting it.
613 - > he has, a very interesting challenge ahead of 614 - > Jim McGregor: him. 615 - > he's walking into 616 - > Leonard Lee: a very 617 - > Jim McGregor: challenging environment And 618 - > Leonard Lee: so let's give a little love to AMD. 619 - > Anything on the AMD front? 620 - > Anything else you guys wanna talk about?
621 - > Jim McGregor: Just the fact that, they, uh, we're not 622 - > talking financial, we're not financial analysts here, but- 623 - > Yeah when they came out with their Q1 results, or their Q4 624 - > results, you know- Who? 625 - > they kinda, AMD, they kinda said- Oh, okay. 626 - > Yeah that, we're worried about memory and everything else, and 627 - > I- Yeah kept looking at this and saying,"Guys, I'm not worried 628 - > about AMD. 629 - > I think they're cranking."
And, after seeing the Q1 results, 630 - > obviously they- they've kinda revised that. 631 - > They are cranking. 632 - > They're doing very well. 633 - > And the thing about AMD, and I- I think that this kinda gets 634 - > lost in the a- in the whole AI discussion, a lot of times, it's 635 - > not just about selling AI accelerators.
636 - > They are doing well. 637 - > Karl Freund: Yeah. 638 - > Jim McGregor: They are making headway with the MI300 series, 639 - > and a lot of interest in the MI400 series. 640 - > But they are also very, very competitive, and I would argue, 641 - > and continue to gain market share in server processors.
642 - > So they are no longer, I think, the second child in that race 643 - > anymore. 644 - > they are doing exceptionally well, and Epic has favored in a 645 - > lot of those AI platforms. 646 - > and they continue to do well. 647 - > They've got great partnerships with, with TSMC and 648 - > GlobalFoundries.
649 - > They're doing very well in, especially in desktop PCs. 650 - > they're doing well in embedded segments. 651 - > You gotta remember that they invested heavily w- in all 652 - > these- Yeah industrial embedded applications with Xilinx, and 653 - > they continue to execute very, very well. 654 - > Leonard Lee: Yeah.
655 - > And you know what? 656 - > speaking of embedded, one of the things that I'm noticing is the 657 - > IoT for AI infrastructure is starting to turn pretty hot. 658 - > Last year when we were at Cadence LIVE- Mm we heard a lot 659 - > about digital twinning. 660 - > NVIDIA had this whole collaboration with Cadence, and 661 - > I'm sure they had it with Synopsys as well, about, 662 - > instrumenting, and modeling AI factories or, AI infrastructure 663 - > data centers, right?
664 - > But, you can only do that if you have, um, sensor devices and 665 - > embedded, technology out there providing that perception edge 666 - > around that data center or for that data center. 667 - > And so I, I don't know if you're noticing, some movement there, 668 - > but it's interesting to see that, some of the, the 669 - > non-processor,'cause you mentioned, the accelerator. 670 - > We- we're seeing some of the IoT-ish, chip guys- Mm-hmm out 671 - > there doing pretty well now, you know?
672 - > Jim McGregor: a- absolutely. 673 - > I, I think we've s- we've turned the corner on that whole supply 674 - > glut that we had, you know- Yeah coming off of, COVID and 675 - > everything else. 676 - > Leonard Lee: stopped, yeah. 677 - > Jim McGregor: And we've always said this about IoT, and it 678 - > extends, to your point, to the data center.
679 - > Sensors are extremely important. 680 - > Mm. 681 - > wireless communications are extremely important. 682 - > RF, and, analog solutions are important, obviously, as part of 683 - > that.
684 - > And then you also have power. 685 - > Power. 686 - > Just managing power is becoming critical in just about every 687 - > environment. 688 - > Mm.
689 - > So it is, a... 690 - > We, we definitely see the market turning and all of those guys 691 - > doing well. 692 - > Matter of fact, I still think that sensors is the underrated 693 - > area, and I think that we're gonna see a lot of... 694 - > We've seen some transitions.
695 - > NXP sold off some of their sensor, st- some of their sensor 696 - > products and technology. 697 - > we've seen On Semiconductor buying, we've seen TI buying, 698 - > we've seen other companies buying, or ST, sorry, ST buying. 699 - > They bought the NXP stuff. 700 - > it's gonna be interesting.
701 - > I think we're gonna see a couple of, major sensor vendors emerge 702 - > out of this over the next decade. 703 - > Leonard Lee: yeah. 704 - > And then just going really quickly back to AMD, one of the 705 - > things that I sensed in, and have sensed in their reporting 706 - > for the past few quarters is, CPU and traditional data center 707 - > stuff, i- is, really a key driver- Mm not only of profit, 708 - > but, a significant driver of revenue as well. 709 - > They put CPU in front of GPU.
710 - > and the odd thing is also, during Arm's, earning call, 711 - > Rene, I think, he mentioned th- their CPU pipeline has doubled 712 - > since, Arm Everywhere. 713 - > and then we have NVIDIA. 714 - > What did NVIDIA... 715 - > NVIDIA announced, or, I think Colette mentioned that, uh, CPUs 716 - > will be, what, 20, is it 20 billion?
717 - > Karl Freund: 20 billion. 718 - > Leonard Lee: Yeah. 719 - > Karl Freund: Yeah 720 - > Leonard Lee: what do you th- 721 - > Karl Freund: what do you 722 - > Leonard Lee: think? 723 - > Karl Freund: I think NVIDIA's kind of underrated by many 724 - > people in terms of the strength of their CPU business.
725 - > If you look at the Gartner, server report, it shows Arm 726 - > growing fantastically, and 80% of that growth is roughly is 727 - > NVIDIA. 728 - > and so While NVIDIA's two biggest competitors are both 729 - > x86, Intel and AMD, the pole position for Arm servers is wide 730 - > open, and it looks to me like NVIDIA's winning it. 731 - > Leonard Lee: Yeah. 732 - > Jim McGregor: I would agree on the Arm side, but you also have 733 - > to remember a lot of those Arm CPUs, are displacing what used 734 - > to be other risk-based CPUs for storage- Yeah for networking, 735 - > for other platforms within the data center.
736 - > So, but it is interesting as you start looking at the numbers 737 - > for, the likes of NVIDIA, AMD, Intel, et cetera. 738 - > It is data center and then everything else. 739 - > Leonard Lee: Yeah. 740 - > I think everything else is only, like, 7% of NVIDIA's business.
741 - > It's, like, insane. 742 - > Jim McGregor: Yeah, gaming is now everything else. 743 - > Karl Freund: Everything 744 - > Leonard Lee: else. 745 - > Who would've thought 746 - > Jim McGregor: that?
747 - > Leonard Lee: Yeah, they didn't even bother talking about 748 - > everything else. 749 - > It's like, well, we have two, two parts of our business, 750 - > hyperscaler and this ACIE thing, right? 751 - > I mean, it's amazing how much things have changed in just one 752 - > year. 753 - > I, um, tuned into Dell Technologies World.
754 - > One of the things that came out of there that I thought was 755 - > really interesting is they're pivoting toward warm water 756 - > cooling, Lenovo was one of the f- two or three pioneers in this 757 - > particular area, and one of the things that's really interesting 758 - > about warm water cooling, it uses facility water, so y- it 759 - > reduces requirement for, chillers, right? 760 - > and that has pretty huge implications in how you design 761 - > your data centers, which is interesting because now we have 762 - > f- you know, generation over generation, the data centers 763 - > themselves are on a advancement or architectural, 764 - > diversification curve.
765 - > Jim McGregor: Well, and that, that was a big focus of Open, 766 - > Open Compute last year. 767 - > Yeah. 768 - > Leonard Lee: Yeah 769 - > Jim McGregor: I just don't see the immersion taking off, and 770 - > the warm water is much more appealing. 771 - > And quite honestly, the directed chip is still a much more 772 - > appealing solution than I think immersion ever will be.
773 - > Yeah. 774 - > But yeah, it is... 775 - > You're right. 776 - > and that's an area where you've got companies like Vertiv, and 777 - > Flex- Yeah and LiquidCool, and you know, that continue to 778 - > innovate around there.
779 - > Leonard Lee: Yeah. 780 - > And the other thing, um, AI workstation. 781 - > Yeah. 782 - > More and more people talking about it.
783 - > And Jim, remember you and I, when we were at Computex, was it 784 - > last year? 785 - > Mm-hmm. 786 - > We did our little bit, and we talked about how AI workstation 787 - > is probably gonna be that beachhead for enterprise AI. 788 - > Yeah.
789 - > And lo and behold, we you know, we have OpenClou that got 790 - > everyone excited about putting, agents locally on device. 791 - > And, yeah, there was a lot of talk about that. 792 - > And then I think, the, the work that Cisco's doing in helping to 793 - > fortify some of these open, agentic frameworks on device i- 794 - > is pretty, pretty interesting. 795 - > I'm tracking that pretty closely, and I'm gonna be at 796 - > Cisco Live next week, so I'm gonna be delving into that.
797 - > And Jim, you still owe us That, readout from your team on 798 - > OpenClaw Yes, we 799 - > Jim McGregor: need, we need to get Damian 800 - > Leonard Lee: on here- And NemoClaw 801 - > Jim McGregor: because he's- 802 - > Leonard Lee: Yeah 803 - > Jim McGregor: He's been working on, first off, the, especially 804 - > the different types of workstations, from a full-on 805 - > workstation down to the mini workstations. 806 - > And it really is going, the workstation market's going 807 - > through a renaissance with AI because it is so much more 808 - > efficient, or I should say, economical- Cost efficient, 809 - > Leonard Lee: yeah, economical 810 - > Jim McGregor: economical to actually be working on those 811 - > than sending everything to the cloud all the time.
812 - > Yeah. 813 - > So that's going through a renaissance, and then OpenClaw 814 - > just exploded everything, where everyone's, looking for new ways 815 - > to create personalized AI agents, sometimes in a very 816 - > dangerous fashion. 817 - > But it is- Yeah it is just exploding. 818 - > It is amazing- Yeah what's going on.
819 - > So we should really have, Damian- We sh- and/or Kevin on 820 - > one of these calls to talk about- We should what's going 821 - > on. 822 - > Karl Freund: It'll be interesting to see if Nvidia 823 - > makes that a highlight next week at Computex. 824 - > Jim McGregor: it will be 825 - > Karl Freund: I think it will. 826 - > And Jensen was this morning announced, significant increase 827 - > in his investments in, in, in Taiwanese companies.
828 - > and I think a lot of that's gonna be around the PC. 829 - > Who has the best platform for running OpenClaw? 830 - > Leonard Lee: Yeah. 831 - > Karl Freund: It's probably made- I mean, that's- It's probably 832 - > Nvidia 833 - > Jim McGregor: Yeah, and you should, we should probably 834 - > highlight that and the fact that Nvidia, now that's basically 835 - > printing money, should It has become one of the largest VCs in 836 - > the tech industry.
837 - > Yeah. 838 - > Mm. 839 - > Matter of fact, I've got a tracker now, and, like at least 840 - > once every two weeks, there's a new announcement of, about an 841 - > investment or partnership, not always the financial- Yeah 842 - > details, but, something Nvidia is doing with somebody, and it 843 - > is, it's ridiculous. 844 - > I think I've got over 60 names on there already, and that's 845 - > just over the past three years.
846 - > It is insane. 847 - > and they are, they have... 848 - > I think Jensen's always had, a preference towards the Taiwanese 849 - > companies and working with the Taiwanese companies and raising 850 - > up the Taiwan, 851 - > Leonard Lee: Yeah 852 - > Jim McGregor: visibility and, tech ecosystem. 853 - > But even beyond that, Nvidia is investing across the 854 - > Leonard Lee: spectrum.
855 - > Yeah. 856 - > I think he's- 857 - > Jim McGregor: You know, from networking to EDA tools to, 858 - > models to, you name it. 859 - > it is incredible the amount of investment that this company is 860 - > pouring back into the industry, Yeah through its success. 861 - > Leonard Lee: Yeah.
862 - > and it's other income 863 - > Jim McGregor: Yes, in its other income 864 - > Leonard Lee: Which is interesting. 865 - > We're seeing a lot of these players, report other income 866 - > and, these are their investments. 867 - > Th- they're- Mm-hmm they're reporting their unrealized, um, 868 - > gains on these, investments that they're making, in startups and 869 - > such. 870 - > So yeah, that's a...
871 - > That was another interesting thing, thing that came up in 872 - > this, last, earnings season for the tech guys. 873 - > I think it was Google and AWS that reported really big 874 - > numbers. 875 - > So, but then, um, yeah, anything else? 876 - > Want any things that you wanna get off your chest and share 877 - > with our audience?
878 - > Jim McGregor: I would just say that- I think we covered 879 - > Leonard Lee: a 880 - > Jim McGregor: lot already, 881 - > Leonard Lee: but- 882 - > Jim McGregor: myself and Damien will both be at Computex, so 883 - > we'll be reporting live from there. 884 - > Awesome. 885 - > Damien will also be at, Automate, Industrial, Robotics 886 - > Automation Conference in Chicago, in June. 887 - > And then I will be at the Qualcomm, Investor Day also in 888 - > June.
889 - > So a busy month. 890 - > And Kevin will also be at N- Nebius Inflection, so it's gonna 891 - > be a very busy 892 - > Leonard Lee: month for us, yes. 893 - > Really? 894 - > Okay, cool.
895 - > Yeah, I'll be there. 896 - > I have eight events next month, so it's- Eight? 897 - > Eight. 898 - > Jim McGregor: Wow.
899 - > Karl Freund: Only four weeks, so you're doubling down. 900 - > Leonard Lee: I don't know what happened. 901 - > Jim McGregor: Well, and there are a lot more events- Yeah on 902 - > our calendar. 903 - > That's just the ones that we've committed to.
904 - > Leonard Lee: Yeah. 905 - > Carl, any last mentions or sharing, 906 - > Karl Freund: I think we should all watch what happens in 907 - > Taipei. 908 - > I think Computex... 909 - > a lot of companies I work with are saying,"Wait till Computex.
910 - > We'll tell you all about it." 911 - > Leonard Lee: Mm-hmm. 912 - > Karl Freund: I really do think that, it will be a big event for 913 - > all the Nvidia wannabes to come up with new technologies, new 914 - > announcements, new customers. 915 - > So, I think it's- Not gonna miss out bears watching.
916 - > Bears watching. 917 - > Jim McGregor: and this is important because Computex 918 - > traditionally was The PC show. 919 - > The largest PC show in the world. 920 - > now it is really transformed, and it's funny'cause at one time 921 - > they tried to tran- transform it into mobile, and that really 922 - > didn't work well.
923 - > But now there is a whole section, especially one whole 924 - > floor that's dedicated to embedded computing. 925 - > So industrial- Yeah automation. 926 - > Yeah robotics, those types of applications, and we're seeing a 927 - > lot more, especially since AI is so dominant in the data center, 928 - > we're seeing more innovations and announcements and everything 929 - > around the data center and AI- Yeah at Computex as well. 930 - > Leonard Lee: Yeah.
931 - > and I think you're right, Carl. 932 - > This might be an AI PC, AI workstation year for Computex. 933 - > Karl Freund: Yeah. 934 - > Leonard Lee: Yeah.
935 - > I 936 - > Karl Freund: think so. 937 - > Leonard Lee: Yeah. 938 - > Karl Freund: find out till next week, yeah? 939 - > Leonard Lee: Yeah, we will.
940 - > So gentlemen, let's call it an episode. 941 - > What do you say? 942 - > Karl Freund: Yeah, 943 - > Leonard Lee: it's an episode. 944 - > Carl's gotta go, so yeah.
945 - > Karl Freund: go. 946 - > Leonard Lee: Thanks everyone for, tuning in. 947 - > We really, appreciate your viewership, and, thank you, 948 - > gentlemen, for sharing your knowledge and, your, insights. 949 - > Jim McGregor: I didn't get to use the slap icon, but we'll use 950 - > it next week.
951 - > I'll save it- Yeah for then. 952 - > Leonard Lee: did you invent it? 953 - > Did you actually vibe code that? 954 - > Jim McGregor: I did, but we'll save it for the right time.
955 - > Leonard Lee: Oh, okay. 956 - > All right. 957 - > Send it to me, okay? 958 - > Send it to me in a zip file.
959 - > But hey, everyone, make sure to reach out and follow Carl Freund 960 - > @CambrianAIResearch www.cambrian-ai.com. 961 - > He's also on Substack and Forbes.
962 - > also LinkedIn. 963 - > So- Connect with him, tap into his research. 964 - > It's great stuff. 965 - > And also reach out to and follow Jim McGregor and the Terius 966 - > Research Team, who are, the foremost authorities on, 967 - > everything that has a chip in it, at www.
teriusresearch.com. 968 - > And, also please subscribe to our podcast, which will be 969 - > featured on the Next Curve YouTube channel. 970 - > Check out the audio version on Buzzsprouts or find us on your 971 - > favorite podcast platform.
972 - > Also, subscribe to Next Curve Research Portal at 973 - > www.next-curve.com for the tech and industry insights that 974 - > matter, and we'll see you next month. 975 - > Take care, gentlemen.
976 - > Cheers. 977 - > Karl Freund: Safe travels.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Stay Curious: Jerrod Lew on AI VideoImagine AI Podcast · on ByteDance55 / 100
How SoftBank Vision Fund Reshaped Venture CapitalThe CEO Diary with Fexingo · on ByteDance46 / 100
Trump Nominates Bowman for Fed Vice Chair, Klarna Replaces Affirm at Walmart, Oracle Eyes TikTok takeover, & Highlights From Day One of CBA LIVEBanking on Disruption Daily · on ByteDance40 / 100Edge Continuum: Where AI belongs from sensors to cloud (with Azita Arvani)86 / 100Edge AI Foundation at Sensors Converge 2026 (with Pete Bernard)55 / 100Silicon Futures for April 2026 - The AI CPU craze, Qualcomm's custom AI, Google TPU 8 explained80 / 100The Blueprint for Agentic Security (with Raj Chopra)75 / 100The Unknown Unknowns of Agentic Security (with Raj Chopra)80 / 100