Latest / Oxide and Friends / Intel after Gelsinger

Transcript

Intel after Gelsinger · Oxide and Friends

• 0:00
  Adam Leventhal: Hello, Bryan. How are you?
• 0:01
  Bryan Cantrill: I'm good. How are you?
• 0:03
  Adam Leventhal: I'm doing great. You have a good break. I
• 0:06
  Bryan Cantrill: Yeah, it was In Mexico, this is my mom. Yeah, that was a lot of fun
• 0:13
  Adam Leventhal: so And this explains why you were DMing me at 6.30 about a topic for tonight.
• 0:19
  Bryan Cantrill: So a little bit, a little bit. And actually my mom factors in a little bit here. So I was factor, yeah, I actually woke up even a little bit earlier than that. Had meant to get us set up for an episode yesterday, Which, you know, is trying to get them set up. Wow.
• 0:34
  A full hour hours in advance. Yeah.
• 0:38
  Adam Leventhal: Exactly. Well, a little bit of
• 0:39
  Bryan Cantrill: a tryhard there. And so I had woken up and I was getting, and it was good. I had an episode in mind. I had, unfortunately the people that I wanted to make it, couldn't quite make it, but I was getting it set up. And then I, at some point in there, I saw you tweeting out, or actually, or skeeting out, excuse me, the Gelsinger, your take on like Gelsinger quoting the Reuters piece, the Max Cherney piece from several weeks ago.
• 1:08
  Yeah. And my thought of that was like, man, Adam's like kind of behind on the news that Adam's just like reading this thing from like 2 weeks ago or whatever. And has finally got, I mean, you, like you do a very good job of like having, like, no, this is like on my list of things to do, and I've kind of been working through it all, and like, I've got to the list, you know what I mean? Like you actually said- I
• 1:27
  Adam Leventhal: didn't even realize that was from a couple of weeks ago. I was looking for like the most in-depth, I mean, it was so prescient. Like the fact that it doesn't mention that Gelsinger's out is kind of totally escaped me.
• 1:41
  Bryan Cantrill: So we, and we actually were going to do an episode on that piece. But for a variety of reasons didn't. That and Max Cherney at Reuters wrote that, Don Clark at the Wall Street Journal wrote another terrific piece. And we were going to do an episode on that piece. But I had this other idea for an episode and I'm kind of running it by you and you're like, yeah, okay.
• 2:02
  So, all right. I mean, sure. It's like surprising to me. You don't want to talk about Intel. And I was like, even at that point, I'd be like, I don't know what you're talking about.
• 2:10
  And then I was like, proud of myself for like, okay, I've got all I've gotten up early. I've gotten this early start. Like the sun isn't even up yet. I've, I've, I, you know, so proud of myself for having arranged our podcast an entire 11 hours in advance. Wow.
• 2:25
  Really impressive. And then I happened to go to Hacker news, thank God, because then it's a number of the story on hacker news. Yeah. And I'm like, oh, okay. And then the reason I say thank God is because then my mom messaged me like 30 seconds later.
• 2:44
  And I'm like, this is going to be mom telling me that galsingers out. And it was,
• 2:49
  Adam Leventhal: and if, and your mom does her podcast on Tuesdays, so you've got to get it in tonight. Otherwise she's going to scoop us.
• 2:56
  Bryan Cantrill: Oh, she is. I'm saying like, I came within 30 seconds of having this news broken to me by my mother, which is, I mean, there's no, there's nothing to be, there's nothing to be embarrassed about. That's, you know, that's really speaks highly of her own engagement, frankly. But the, I just was kind of impervious to this 1. Like, Oh, suddenly many things over the last half an hour making a lot worse.
• 3:16
  Adam Leventhal: It should be said we've been resisting doing an Intel episode. Yes. For, I mean, sure, a couple of weeks, but I feel like it like months. I feel like ever since I read chip war, I've been really eager to do it. And maybe you've been thinking about it even before then, but like it's been a while since we've wanted to talk about Intel.
• 3:34
  Bryan Cantrill: Yeah, I've wanted to talk about Intel for a while. I've got a lot of thoughts and a lot of opinions. It should be said just to the top that Intel Capital is an investor in Oxide, which I mean, that's, you know, that's been a factor.
• 3:47
  Adam Leventhal: A valued investor.
• 3:48
  Bryan Cantrill: A valued investor, a cherished investor. So that's kind of been a factor. And then there was the, the, the Reuter story and the Wall Street Journal story. I want to actually talk about both of those because those are, both of those stories are, and again, those ran, I think like on the order of 2 weeks ago, maybe 2 or 3 weeks ago, something like that. And wanted to do an Intel story after those stories landed, but we didn't end up landing on it for a variety of reasons.
• 4:17
  And, I'm really glad that we didn't because this is going to be a much better discussion because why,
• 4:23
  Adam Leventhal: can you imagine if we had done, well, that's it for Intel, nothing left to talk about 2 weeks ago or whatever.
• 4:30
  Bryan Cantrill: Totally. Right. So you've got that, right? You've got the, the, you know, we would have been exactly wrong. I, which the God knows not for the first time.
• 4:40
  The, but I also think that like, and part of the reason I was kind of relieved when that thing kind of fell apart, it's like, I think it would be kind of a negative episode about Intel and I think this is going to be much more positive in a sense in that we can actually be looking forward to the future of Intel and stop talking about where they're stuck. And now that said, we're going to talk about like, we're not going to gloss over the last 3 and a half years here. So this is going to be perfect timing. I'm glad we're doing this. And we've got a lot to say on this.
• 5:23
  I was being asked online. It's like, well, who do you who would be your candidates to run the company? And I'm like, you know, I don't know about running the company and we'll you know, we'll talk about this in more detail a little bit later, but I'm like, you know, the kind of you know, the kind of person they need, you know, like a George Hott's type not need, you know, you know George, Geo Hott's those of you who know George I don't know him personally definitely a bit of a live wire, bit of a nut, but kind of an interesting nut. You know what I mean? Like he's done, he famously jail broke the iPhone and definitely an interesting, like an interesting technologist, but also like bonkers.
• 6:02
  He was also the 1 that was gonna come in and save Twitter in 12 weeks and everything else. So, you know, then I went over to his, like, where is that guy? Like I went to go over to Blue Sky, which is actually where I'm looking for people. Are you finding that you're looking for people first down Blue Sky, Adam? Are
• 6:16
  Adam Leventhal: you? Yeah, yeah, I'm pretty much off twitter and then I don't have any hope of people being on mastodon so yeah I'm looking I'm looking at blue sky
• 6:23
  Bryan Cantrill: you know and this is actually a bit of a first for us this is an episode that we did not promote at all on twitter this has been only promoted on blue sky
• 6:31
  Adam Leventhal: yeah or on mastodon
• 6:33
  Bryan Cantrill: well or on mastodon we all I only put it some blue sky in part because well poor mastodon sorry mastodon But I, the I went to Twitter, I'm like, I don't do I really want to go deal with this. And, and then like Twitter, like the gelsinger thing is not even being discussed at all that thing is descended into such a truth social echo chamber that there's no even discussion of galsinger happening on Twitter feel like you know what like Ipother But then I'm like George Hott's is tweets have all been deleted Like what so yeah, yeah, Jude. So I like that's okay. What's what is going on? I'm like, did I?
• 7:09
  And I'm like thinking, like, am I on the right, like, page? Is this like a different way to find someone's tweets now? I'm like, no, it's like, his tweets all seem to be deleted. It's weird. And then I kind of go over back to to blue sky.
• 7:22
  And there is and I had I posted about this just because I thought it was so bonkers. The apparently his and this is Grace kind net on on blue sky posts in a strange turn of events George Hans has written an article advocating for expedited nuclear war and has also deleted all of his tweets. So like
• 7:50
  Adam Leventhal: Milkshake duck very hard
• 7:52
  Bryan Cantrill: there definitely Definitely don't shake duck that 1. Milkshake nuked that 1. So Yeah, so I don't think everyone's going to be forthcoming, but yeah. Okay.
• 8:03
  Adam Leventhal: What's your number 2 pick?
• 8:05
  Bryan Cantrill: Well, I've actually got a I've got a great pick. We'll get to him later, but I've got a great he won't do it I got but I actually have the person who should lead into I've got a couple of names I've got the personal I think should lead Intel There's no way they're gonna do it because there's no way he's gonna agree to it. But we'll kind of save that for the reveal. All right.
• 8:23
  Adam Leventhal: Very exciting.
• 8:24
  Bryan Cantrill: Okay, so that was like my, now that we've gone through my processing or misprocessing of this news this morning. You obviously correctly, you were up first of all early Pacific time as well. And you had obviously seen this news early.
• 8:38
  Adam Leventhal: Yeah, yeah, yeah, I got up, you know, went to the scroll and and saw the galsingers out and actually I think I might have gotten a text from my East Coast buddies, 1 of whom worked for Gelsinger. So they were chatting about it.
• 8:53
  Bryan Cantrill: That is interesting. That is interesting. So they were, the folks that have worked for them at VMware. That's right. Yeah, yeah.
• 9:01
  Yeah, interesting. Well, I would be very curious to know what their take is, because someone had posted on also Blue Sky that the pivotal slack is crackling. And I'm like, OK, I definitely want to know what's going on there. Because, so to give a, that's kind of rewind a little bit. So, Pat Galsinger came to Intel in 2021 and, I had in, in,
• 9:30
  Adam Leventhal: if they're just returned to Intel in 2021, I mean, just like, give him this due on this, right? Cause he, he was there, CTL, He was there for 30 years or something.
• 9:38
  Bryan Cantrill: He was. Yeah. I hate saying it that way because I think this was part of the problem.
• 9:44
  Adam Leventhal: Ah, the like returning savior, the like soul of the company That kind of thing. Yeah.
• 9:51
  Bryan Cantrill: Yes big part of the problem. Actually is the kind of return of the king and That this is the technologist. This is the engineer that's gonna lead us. This is the engineer from Intel's greatness. And that was a bit like a bit of a cocked eyebrow.
• 10:08
  Now you obviously, have you ever met Gelsinger? Cause you obviously, you know him well from the supercut.
• 10:15
  Adam Leventhal: Right, No, I've never met him. I only have like 2 friends who kind of met him personally, 1 because they're neighbors and lost a dog and the other from working at VMware.
• 10:29
  Bryan Cantrill: Interesting. And I mean, what I knew about the guy is that like people who worked for him liked him. Huh. Found him. Yeah.
• 10:37
  Adam Leventhal: I, well, I, I, I've heard that opinions vary on that 1. 0, interesting. I think that his, his very, sincerely held faith, I think, like some folks found a little overbearing.
• 10:53
  Bryan Cantrill: It's, it's pretty, and whenever he would land in hot water, people would rediscover the fact that this guy, I mean, he's very, he's very public about it. You know, he kind of tweets out Psalms every Sunday and so on, which I think is, is yet there was, there
• 11:07
  Adam Leventhal: was, there was that time when like the stock market was, was cratering and he tweeted out some Bible verse. And so I had a great tweet that was like, Hey, I'm, I'm new to the stock market. The CEO of Intel is praying. Like, is this a good sign?
• 11:25
  Bryan Cantrill: Right. And I understand, like, listen, I think it's great that you're finding, you know, inspiration from wherever you're finding it. I just don't think it's prudent. You know, I just think it's like there's a sense of like You know look at the optics of this and I just don't think it makes it. I just think it's a bad idea I think it's like I but you know, I'm with these old schoolers that things I also think the business and politics don't mix So that's clearly out.
• 11:47
  So yeah You know, I guess I'm a traditionalist in that regard. But I, so I didn't know very much about him, other than that your super cut in which you, you touched the hand of God, not to bring religion back into it, but you really did, I feel, touch the hand of God and that's super cut.
• 12:05
  Adam Leventhal: Thank you. I mean, you know how much work it was. And like we talked about last week on the, on the episode or a couple of weeks ago about episode about writing, I was also unemployed then. So spending a couple of hours or more than that, cutting up a video of Gelsinger and Andy Jassy from AWS debating whether it's on-premise or on-premises.
• 12:27
  Bryan Cantrill: Well, the beauty is that it's like, the debate is just right under the, it's under the surface. This is a, this is a, this is submarine warfare. Yeah. Andy Jassy is only saying it correctly on premises and Gelsinger is only saying it incorrectly on premise. And they're both repeating it back to 1 another.
• 12:45
  Love it. Really, Really genius. And I mean, the fact that like you like that, that this is, you know, today is the first day of reinvent. I guess it's like, do we give 1 another presence every day of reinvent? How does that work?
• 12:57
  I remind me of the tradition. We light a candle on the first day
• 13:00
  Adam Leventhal: of reinvent. Earing of grievances, I think.
• 13:02
  Bryan Cantrill: Earing of grievances on the first day of reinvent, the, the airing of new services, the, but the fact that it's the first day of reinvent, that was, I think that was a reinvent that was for sure from reinvent 2018. That's right. Yeah. Yeah. So, okay.
• 13:16
  So the, this is basically my experience with Gelsinger. I knew him from your Shipper Cut, knew him kind of vaguely as like, you know, the popular VMware-ish at VMware. And then I went and, you know, it's something we've talked a lot about here on Oxide and Friends, are the Computer History Museum Oral History Series, which I think are just absolutely superlative. And he has a 4 hour oral history with the Computer History Museum. Oh my God, this is great.
• 13:45
  And I mean, I think these are so valuable. Like this is perfect because, you know, we are just to kind of bring it home to Oxide. We are using the Tofino chip, Intel Tofino, which was a result of the barefoot acquisition. Our reticence, we like a lot of things about Tofino, our reticence about Tofino was that it was honestly an Intel product and we were concerned that Intel was going to kill it. So Intel spent a lot of time convincing us that they would never kill it.
• 14:13
  And that, ha ha. So I, it was extremely load bearing for us. Like, I want to understand who Pat Gelsinger is, what he's about. He's the new CEO of Intel. I want to get an audience with him and explain why we think this part is extremely important.
• 14:32
  So, I want to understand him. And so, I'm listening to kind of this oral history is that, you know, I'm washing the dishes and so on, as 1 does. And you know, it, I would say it's interesting in that he is, you know, he has the kind of accoutrements of humility without actually being humble. Do you know what I mean?
• 14:54
  Adam Leventhal: Maybe. Can you say more? That sounds...
• 14:57
  Bryan Cantrill: He's got this kind of folksy Midwestern humility, but it's actually kind of arrogance.
• 15:02
  Adam Leventhal: Interesting.
• 15:04
  Bryan Cantrill: And it is, and it's kind of this offshocks kind of arrogance. And I'm, so I'm listening to it. I'm like, you know, I'm getting kind of deeper and deeper into this. I'm like this guy, I'm kind of, because in particular, you know, he's talking about the history of the 486 and it is definitely, you know, are you remember Ian Batten? Yes.
• 15:22
  Yeah, who's a customer of ours at Sun at Fujitsu in the UK. It's terrific. And Ian would always talk about the, you know, he would talk about these memoirs of World War II from various folks, and they were all the how I won the war memoirs. And I kind of felt like I was getting a how I won the war from him on 486 in particular. And because he's talking about history that like, and obviously I wasn't an intel at the time, but it's definitely, you know, more complicated than Pat Gelsinger alone.
• 15:51
  And there were a bunch of other folks that he just wasn't mentioning that was kind of surprising him. He didn't, you know, on the 3 days. Anyway, so I'm like, I'm getting it's like, whatever, it's a normal history with him. So it's like, it's not totally unreasonable. And so, okay.
• 16:06
  But again, it's like, it's very like, aw, shucksy. So it's hard to like, it doesn't come across as off putting it kind of comes across as humble. But I'm like, if you actually take it apart, it just feels a bit arrogant. So you're gonna get through it and he starts talking about his departure from Intel. And he had a lot of misgivings.
• 16:26
  So he's going to be hired away by EMC. And he had misgivings in particular, because he was running Larabee at and you were every encounter with Larabee
• 16:38
  Adam Leventhal: No, Adam
• 16:39
  Bryan Cantrill: Larabee is a their GPU from it's like their 2012 era GPU. Okay, and Larabee maybe you have a plus or minus 2 years on that, but the, and he had this line about, and I'm, I'm paraphrasing, but, but like, I'm not that far off from what he actually said, that Larabee, that he knew that he had great misgivings about leaving Intel because he knew if he left Intel, Larabee would be killed. And as I'm washing dishes, I'm thinking, yes, that's correct because Larabee needed to die. Larabee had a lot of issues and a lot of problems. Like the beginning of some interesting ideas, but like Larabee was nowhere near where it needed to be.
• 17:23
  And in particular, the software story on Larabee was kind of an infamous wreck. So I knew if I left Intel, Larabee would be killed. Like, yes, Larabee was probably gonna die of its own weight. And I knew if I stayed at Intel, that Larabee would come to dominate the GPGPU market. And that, I bet I'm probably paraphrasing, but the thing he said almost verbatim is, and that Nvidia would be a quarter of its current size today.
• 17:54
  I'm like,
• 17:56
  Adam Leventhal: wow. You know, we're both baseball fans and everyone else who listens to the podcast is, I know, it always makes me think of like, Supermetrics and in particular, wins over replacement. And when you start talking about like, this is putting like just so many wins over replacement on your back. Like you are the most valuable player of all those valuable players times like a billion.
• 18:24
  Bryan Cantrill: This is like, I actually have a, I have a war of 88. It's like you have a war of 88, 160 games, 162 games in the season. I'm just, I'm just telling you, I would have taken a last place team and I would have turned it into last year's Los Angeles Dodgers.
• 18:40
  Adam Leventhal: That's right.
• 18:41
  Bryan Cantrill: Right. It's like, I, okay. Wow. And I mean, I was stunned by that. I'm like, I've really misheard him.
• 18:51
  You know what I mean? I'm like, I got to go back and replay that. I mean, it was so shocking to me that he, because it was, and so I went to replay it and it's like, no, that's that that is exactly what he said. And that machine. So this is this is a quote now I am now quoting from his oral history.
• 19:18
  Nvidia would be a fourth the size they are today as a company, because I think Intel really had a shot in that space. And it's like, you think, I mean, that is so, I mean, we're not going to give gensies more credit than It's like, I don't know what, I don't
• 19:35
  Adam Leventhal: know when this was recorded, but as of today,
• 19:37
  Bryan Cantrill: 2019, 2019.
• 19:39
  Adam Leventhal: Okay. So we're still talking about like, he thinks he could personally conjure a billion dollars from, from basically thinner.
• 19:47
  Bryan Cantrill: Tens of billions.
• 19:49
  Adam Leventhal: I always get, Oh, pardon me. Closer to a trillion. Excuse me. Yeah.
• 19:53
  Bryan Cantrill: Yes. Yes. That, that, that, that he, that his decision as a, frankly, a middle manager to leave Intel was for humanity had multi tens of billions of dollars of ramification. Yeah. It's like, wow, are we, do we actually believe that?
• 20:13
  Do you believe that? And I'm like at that, when I heard that, I'm like, this is the wrong guy for the job. This is the wrong guy for the job This is the wrong guy for the job because and I think it I'm actually like I also think this guy believes it that and that is really really dangerous because what we knew by 2021, and this is like not, you know, I'm not splitting the atom here. This is not a deep thought. Intel was in very deep trouble in 2021, very deep trouble.
• 20:44
  Intel was in very deep trouble in 2019. And he
• 20:48
  Adam Leventhal: joined in 2018. When did he join?
• 20:51
  Bryan Cantrill: Twenty-nine? Twenty-twenty-one.
• 20:52
  Adam Leventhal: Oh, he joined then. Oh, excuse
• 20:54
  Bryan Cantrill: me. He joined in 2021. But I'm saying like even 2 years earlier, Intel was in deep trouble.
• 20:58
  Adam Leventhal: Yeah, really struggling. I mean, struggling in a weird way, right? Like struggling with market dominance and yet a history of inability to execute.
• 21:09
  Bryan Cantrill: Well, this is the thing about semis though, is like you can look into the future and because you know like roadmap and you know it takes multiple years to do this. And if you act, you know, in 2019, we were evaluating what we wanted to do for a, for a CPU, whether it was going to be, and we kind of knew that, I mean, kind of knew, we, I mean, pretty strongly believed it was going to be AMD, but we wanted to really be careful about it. And so we did, and in particular, because like, I think, you know, I feel Robert and I knew that coming in, but we wanted to be sure that other people that, And Adam, I don't know where, because you were probably a little bit further away from where things were from. And-
• 21:50
  Adam Leventhal: Yeah, I didn't know the state of the art or whatever, but I mean, you guys made a very sexy spreadsheet. So of course I was interested.
• 21:57
  Bryan Cantrill: Well, listen, we know, you know, everyone's got the, everyone's got a path straight into their heart. We know like we get you. I color code a spreadsheet put in front of you. I'm not gonna get my way. I'm going to win your heart over.
• 22:10
  And it was it was actually know what you know, actually was not a color coded spreadsheet. Do you recall this? You made a color code.
• 22:19
  Adam Leventhal: That's
• 22:22
  Bryan Cantrill: not brand. I don't, but like this spreadsheet looks looks very good, but I think I can make it look extraordinary. Good. You you color code it. I'm like, yeah, that is just gorgeous because and I think this is why we wanted to do this right it wasn't it was for ourselves but it was also for our fellow technologists who maybe a step away from it it was for you know Steve it was for our customer it was for everybody right we want to make sure that like hey we've done our homework But when you do your homework on this, and this is in 2019, when we are comparing Rome to Cascade Lake, Rome being what was available at the time, Cascade Lake being still a bit forthcoming and they're like Rome was beating the pants off the thing that was still coming from Intel.
• 23:02
  And I'm like, I recognize this in terms of Intel roadmap. This is a spark roadmap. And these, I was remembering back to remember the, I mean, obviously you remembered the calls, the Panther calls. I mean, you're like, remember, I can't, I can't forget them. I've tried to, where you and I were both like, this would be a pretty good chip if it were released 2 years ago, right?
• 23:24
  If it were instead of being 1 year away, it was 2 years in the past. This company might have a prayer.
• 23:31
  Adam Leventhal: Yeah.
• 23:32
  Bryan Cantrill: And that was all struck for, and this is the same way. I'm like, this is the same thing. Like this, this part looking at cascade and then looking at sapphire sapphire lake and Be like yes sapphire is gonna be like Sapphire will be kind of barely competitive with Rome Milan. And then of course, we had the AMD roadmap too. So we were seeing like Rome Milan, Genoa, Torino.
• 23:57
  It's like we, these roadmaps are not comparable. And on any dimension, on cost, on performance, on price performance, on power, on performance per watt. I mean, just like pick your dimension. And this was in part because, And this is the thing I didn't realize until Robert and I got down there and we were talking with them. I did not realize the degree, because I felt by 2019, chiplets were kind of like the obvious direction for everything.
• 24:28
  And maybe I'd just been spending too much time with AMD. In fact, that's assuredly it because the AMD was so bought into chiplets, chiplets, this is through Silicon VIA. This is using, putting a smaller cores, smaller chips, onto a larger package onto a die. And the, and, you can, there's a bunch of advantages of that in particular, you get yield advantages, right? Cause you need less to yield.
• 24:51
  Yeah. It's like huge yield advantages. You need much of this area to be good, right? Yeah.
• 24:57
  Adam Leventhal: And I,
• 25:02
  Bryan Cantrill: I feel like the story on this is kind of yet to be told from an AMD perspective in terms of this move was such a good move in so many different directions. I think it was kind of by accident. My read on this is that this happened because they were spinning out global foundries, finally, a global foundries, which was AMD's fab. And, but that was gonna be on a memory service that was gonna be on a 14 nanometer node. And they wanted to use TSMC on a 10 nanometer node.
• 25:32
  Again, I'm numbers may be off here, but they wanted to use TSMC on a 10 nanometer node. Again, I'm numbers may be off here, but they wanted to use TSMC on a 10 nanometer node for their actual C. So they had, they were gonna use different processes for these things. So it's like, well, at that point, like we're gonna be on different, we're gonna have to use packaging innovation. And that's the end.
• 25:49
  And so, and then, so, but Intel like did not believe in chiplets and that I didn't realize like, oh, okay. Like, no, no, like we believe in tiles, not chiplets. I'm like, okay, interesting. And there are some advantages. Some folks are pointing out the chat.
• 26:08
  I mean, you can get better core to core latency across all cores when you've got this kind of monolithic die versus a chip, let's say you kind of go CCX to CCX on AMD, you can see some additional latencies. I mean, there are some like, there are some like technical reasons for it, but I think they're like way more than outweighed by all the technical advantages, manufacturing advantages, operational advantages, process advantages that you get from chiplets. So, I mean, like Intel was really, really far behind. And that was just incredibly clear to us in 2019. And I'm glad we went down there and did our homework, because that's how we discovered Tufino.
• 26:51
  And so you advanced, so we kind of knew that. And we were talking to people and they're just like, no, that's not going to happen. AMD is not going to displace Intel. And like, I know that seems unlikely, but we're just looking at the server share numbers. AMD probably had maybe 7% or 8% share at that point in the server space.
• 27:10
  They'd up from a low of below 3% when they were dead on the operating table. But we just know that this is what's going to happen. You know, this is basically locked and loaded based on the roadmap. So it was not at all a surprise when that's exactly what started to unfold. And in the next 2 years, Intel began to really struggle and got rid of Bob Swan, which was probably the right thing to go do.
• 27:43
  I would like to point out, because like Hacker News seems to have forgotten, They're like, these problems were all created by BK, Brian Kruzanich, who was not the CEO prior to Pat Gelsinger. The CEO prior to Pat Gelsinger is Bob Swan. So there's another CEO in there. You know, there's the Millard Fillmore of Intel. It's kind of being frapped.
• 28:05
  So Bob Swan was the CEO for 2 years in there. But Gelsinger, so Gelsinger was coming back to a company in trouble and a company that had lost its way in a couple of ways, but it actually lost its way quite a bit before that. And so I started a chip war, Adam, I have not finished it. I'm loving it, by the way. Does he talk about the Intel and their process node problems?
• 28:30
  I'm not sure if he elaborates on that.
• 28:33
  Adam Leventhal: As I recall, like a bit, but I think as you've probably inferred, like that to add a bit of a technical level of technical detail that's like a little too fine grained for the intended audience, but he does allude to falling significantly behind TSMC and Samsung. If significantly behind.
• 28:50
  Bryan Cantrill: Yeah. And so this was with their 10 nanometer node Canon Lake, and this had happened quite some number of years prior. So this is going to be in folks and look it up. That's going to be like kind of in the 2015 timeframe and Cannon like they it's actually still to me unclear How much money they ultimately lost on Cannon Lake because they basically never released a chip on the process note They did an entire process node that was that failed to clear the end of the runway. Wow.
• 29:21
  Which to me is like, Hey, if I'm an Intel investor, and this is like, we built an airport in the wrong city. We actually built an airport with like 8 concourses, 7 runways, you know, like advanced baggage system.
• 29:38
  Adam Leventhal: This sounds like something you'd find in North Korea or North Korea or whatever. Totally. So like the high rise with no elevator.
• 29:46
  Bryan Cantrill: No, but it's like, this is an extraordinarily expensive mistake, extraordinarily expensive. And Intel, I felt was honestly, And as an Intel customer, I'm like, what happened at Cannon Lake? And they were very tight lipped about it. It was getting delayed and delayed and delayed, delayed, and then it was basically dead. And there was very, There was no, and as a customer, it's like, I get it, it's okay.
• 30:22
  Things go wrong, that's fine. I, as a customer though, I've got a, there's a trust issue now and I want you to earn it back. I'm very predisposed to having you earn it back, But in order for you to earn it back, I need to understand what happened here. And that can be, we can understand that in the highest possible terms. But Intel never did.
• 30:45
  They would not say anything about it. What I think is, and they still have never spoken publicly about what happened with Canon Lake and the 10 nanometer node. What happened, and I think that, you know, I'm hoping that there's gonna be, there needs to be a book written on kind of this chapter of Intel. I don't know if there will be or not, but the Intel was using UV lift on Cannon Lake on the 10 nanometer when it definitely needed to be EUV is my understanding. I mean, again, we don't have an official statement.
• 31:17
  Adam Leventhal: 1 of the things that chipboard does go into is that Intel was a naysayer and late adopter of EUV. That they kind of didn't think that it could be made to work. And even as the evidence mounted, they were pretty reluctant to switch over to EUV.
• 31:31
  Bryan Cantrill: So, and that is Cannon Lake right there. That's where that's happening. And when they scrapped Cannon Lake that was them walking away from from UV and realizing they had to go to EUV. That's a very, very serious mistake. And, you know, it's like, okay, you can, okay.
• 31:52
  But, you know, we kind of need to know that you're not going to make that mistake again. And it's hard to know you're not going to make that mistake again when you're being so opaque about it, when you're being so resistant to actually describing it. And this kind of gets into like my, so here is the problem that Intel had when Gelsinger returned was a cultural problem. An acute cultural problem. And it needed someone who was going to rebuild it culturally.
• 32:29
  And That was the thing that Gelsinger was never gonna do. And they had a cultural problem because it was a culture that had allowed the failure, not just allowed the failure of Cannon Lake, but had failed at all to discuss it. It was a cultural problem that had, And I think this I'll give actually Intel a little bit more points on Spectrum Meltdown. Intel did a better job privately dealing with Spectrum Meltdown and getting ahead of subsequent issues. Spectrum Meltdown could have been worse, but it also could have been a lot better.
• 33:00
  And I also think that they could have been much more transparent and it would have served the industry for them to be much more transparent, much earlier about spectrum meltdown. So, so when you talk about
• 33:09
  Adam Leventhal: cultural problems, Oh, I definitely want to get into Optane again, but when you talk about cultural problems, like transparency, certainly, but you know, what, what, what other, were there, how would you describe
• 33:21
  Bryan Cantrill: the culture? Here's the
• 33:21
  Adam Leventhal: problem at Intel.
• 33:22
  Bryan Cantrill: And this has been the problem with Intel up until basically present day. You, when describing some of these massive problems in Intel, you'd be dealing with engineers at Intel who are very, very good and very thoughtful and thorough and they would agree with you and they'd be like, no, I absolutely agree with you, but the company's never going to change. And That's a cultural problem. When you have, when your engineers know what the right answer is, but they also feel that the right answer is culturally unobtainable, you have a cultural problem.
• 33:57
  Adam Leventhal: I wonder the degree to which this culture problem that you're describing is sort of deep in the DNA of Intel.
• 34:04
  Bryan Cantrill: Deep.
• 34:04
  Adam Leventhal: Because I, so I was, I listened to the acquired episode today on Intel and it's, and it's just kind
• 34:09
  Bryan Cantrill: of- Oh, did
• 34:10
  Adam Leventhal: you? Oh, interesting. It's kind of a vignette. It's actually, it was also recorded at sort of a interesting time during the pandemic. So some, some interesting background there, But- I
• 34:19
  Bryan Cantrill: have not listened to that. How was it?
• 34:21
  Adam Leventhal: It was good. I mean, all their episodes are good. But in particular, it talks about the transition from Intel being a memories company, a DRAM company, SRAM company earlier. Like that was it's, it's reasonable to the extra and then pivoting to becoming a CPU company, a microprocessor. Extraordinary.
• 34:40
  Bryan Cantrill: Yeah. Yeah. Honestly, 1 of the most amazing pivots I feel and kind of like in the history of the industry.
• 34:46
  Adam Leventhal: See, you might've heard of this 1. There was a great line in this where apparently Grove and Moore were meeting, and Andy Grove says, if the board kicked us out and they hired a new CEO, What would the CEO do? And Gordon Moore's like, well, they get rid of the memory business. And so they then ceremonially decide to fire themselves. They walk out of the office, they turn around, walk back in and they're like, okay, this is what we're going to do now.
• 35:16
  But apparently even after they had, they had, as an executive team made this decision, you know, found a president CEO, the factories like wouldn't stop. Like turning the ship was incredibly challenging.
• 35:29
  Bryan Cantrill: It was incredibly hard.
• 35:30
  Adam Leventhal: Like what would need to go to factories and like lay people off to get them to stop producing these chips
• 35:37
  Bryan Cantrill: Because I mean they were selling them right they had a market But they just realized like the market here is going to like the margin here is going to 0 We want to get out of this market.
• 35:44
  Adam Leventhal: That's right. I mean a commodity market that they already market, maybe not realizing it's, I mean, maybe it's hard to see that you're in a commodity market as it emerges because they certainly started as an innovative market, but then there were a bunch of competitors. And then you mentioned Optane, they started trying to sell value added memory, which was of course a dud. Right. Before figuring it out.
• 36:10
  But anyway, deep, deep in the DNA.
• 36:14
  Bryan Cantrill: Deep in the DNA. That's really interesting because, and it shows you mean like you got you know, you basically have you got 1 of the founders of the company and Or the founder you got like you got in Grove. There's always you got Grove I mean just like absent the and I don't know if Bob noise was still involved with the company or not at that point, but the, I mean, you, you have like the folks that I mean, and you know, Intel was was an interesting company in that it wasn't actually There were some cultural things that Intel had done that were really interesting in the noise era. That were more, I mean, I wouldn't quite go so far as to say kind of like a non-hierarchical management structure, but it was 1 thing that like encouraged much more encouraging of individual innovation. And so even in that culture, it was very hard to turn the boat, which is interesting.
• 37:06
  Adam Leventhal: I've been sounded very reminiscent of, of sun in some ways where like things would happen independent of, of where leadership was pointing. Have you read only the paranoid survive?
• 37:19
  Bryan Cantrill: I have not. It's, I would like to though. I mean, you're getting, Grov is an interesting cat, you know, I mean, in a lot of ways. And, I think noise is also incredibly, I mean, the more I really need to read a full biography of noise, because a really fascinating person, again, not a deep thought, but World War II being stressful and all. Yeah.
• 37:45
  But really, really interesting person. And in this case, you get to kind of like Intel at its, and so you have this, and I can't wait to get back to listening to that acquired, but you have this kind of extraordinary pivot into the microprocessor and realizing that like the microprocessor is where their future is, and it's not in memory. And then they just become so dominant and they executed so well, but they couldn't, you know, and maybe that episode was foreshadowing their inability to get beyond that market. And, you know, almost comically until I had tried to get kind of into these other markets and had said that never been able to do it. And, you know, we did an entire episode on obtained the which we called a funeral for Optane.
• 38:32
  I think that was where Tom Lyon had the line that it was a goldfish funeral, which I was not very funny. It's the image of Optane being flushed out of the toilet as like even the 7 year old doesn't is dry-eyed about it. But Optane was also, and I think Optane was interesting, but Optane part of Optane's failure to me is wrapped up in the opacity of it. Intel never described how it worked.
• 39:06
  Adam Leventhal: I remember that being a real frustration of yours that like you want your interest in deploying it. Just tell me what it is.
• 39:12
  Bryan Cantrill: Like just tell me what it is because it's like, and this is like, I need to write software for it. If you want me to write software for it, I need to know if it wears out or not. Like that can't be a state secret. You don't need to tell me, I'm not going to duplicate the physics of it. I need to know if it wears out.
• 39:27
  And so I need to know like, is it phase change memory or is it not phase change memory? And the answer is like, yes, it's phase change memory. But it's like, so does it wear out? And they're like, oh, hand wave, hand wave. No, no, we can't talk about it.
• 39:39
  And so then Optane was, it was very hard to kind of bet big on Optane. And it was, it was very, in part, and then I think it was going to be out kind of like at once the savior of the company. But by the way, we can't tell you anything about it. Cannon Lake is late. We can't tell you anything about that.
• 40:00
  And then meanwhile, like I'm having conversations with him in particular around our desire for open firmware And this is 1 of these and there were many such conversations But that you'd have conversations with Intel engineers around the need for for example The approach that we have taken with our holistic boot Adam and they would all oh, yeah. Yeah, that's absolutely what we should do. We're never going to do that. It's like all right. And you know when I went down to visit Brian Krasnich BK in maybe 2013 something like that, He wanted to be educated on cloud.
• 40:37
  And we had a joint board member that was also on the Intel board and suggested that he talked to me. So I went down there to go talk to him. Like, there's no way he's going to know that Joyant is an Intel capital portfolio company, which he definitely did not. But when I was actually laying out for him the vision for what is basically oxide, like this is what you need to go do. And he's like, I actually, I agree with you.
• 41:02
  That's, that'll be like that kind of holistic rack scale thinking. And this is at a point where AMD is dead. So it's like, nope, that's talking about or to AMD, we're just talking Intel. He's like, but I can't do it because I would be competing with HP and Dell, and I'm never going to compete with HP and Dell. I'm like, okay, I'm okay.
• 41:23
  I guess that's a constraint somewhere that you can't, I mean, that's going to limit what you can go do, you know? And, but that was there. So I just felt like even at the top of the company, they felt powerless. You know what I mean?
• 41:35
  Adam Leventhal: Yeah, or at least just hemmed in, right? Like it was very constrained and what that what they could possibly do. Which also makes sense for Optane, right? Like that's, I'm sure, to me, Optane was always a solution in search of a problem. But I'm sure it sounded great.
• 41:52
  It's like, it's this new space age polymer. No 1 else can build it. It could solve a bunch of problems if it cost a different amount of money.
• 42:01
  Bryan Cantrill: If it costs a different amount of money. Yeah. I mean, you could think like the kind of the, the, the bull case for Optane was going to be this thing is going to do to the industry where flash did. And you're going to get non-volatile memory at DRAM speeds with, with maybe better density. It's like, okay, that's interesting.
• 42:18
  But, womp, womp. That's my proudest point of saying.
• 42:22
  Adam Leventhal: Right, with none of the economic tailwinds of Flash.
• 42:26
  Bryan Cantrill: That's right. Yeah, as it turns out. And as it turns out, that was just kind of became a basket case. And then the other problem is that, and then Optane when Optane became really weird is when they started doing these hybrid sticks that had DRAM and then it was kind of like, D-RAM, D-stage to Optane. And they're like, are you interested in this?
• 42:46
  I'm like, no, no, like why? Because there's like, okay, what's
• 42:50
  Adam Leventhal: the firmware running on this? Like, oh no,
• 42:52
  Bryan Cantrill: there's no firmware running on this. I'm
• 42:53
  Adam Leventhal: like, well, okay.
• 42:55
  Bryan Cantrill: But this is, come on, you can't like go, no, no, no, no. Handwave, handwave, trust us. It's like, don't, no, stop the, no, I don't trust you. I don't trust you, I'm sorry. I don't like, drop the opacity please.
• 43:07
  And the, you've got to ditch this before we can actually build on anything here. And it's like, so we're definitely not interested in that. And this kind of dovetails into the other thing that Intel did not understand software. Just didn't. And I just don't, I feel that they made some inroads a little bit.
• 43:28
  They started doing some open source things that were actually interesting. Kind of in like the 2019-2018 timeframe, like I remember going to, they had an open source developers conference that they would host them like this is actually like there's some there's some incipient interesting stuff here. They were working on a replacement for KVM, which I thought was good. Open source. I mean, it's like there's like there were there were a little like there were some green shoots.
• 43:57
  But they had so much further to go And the company kind of knew that they didn't understand software, but didn't know how to, they didn't know what it meant to do anything about that. And I'm, did you ever make special like Intel RSD, the rack skill design stuff out of, no, not at all. Oh God. You'd find it really amusing to go look at some of their, because like, I mean, it's kind of the oxide problem, right? But they were going to solve it with only hardware.
• 44:22
  You're like, you don't, okay. My miss. Okay. That's not going to work.
• 44:30
  Adam Leventhal: I mean, I understand why they would love that architecture and I'm sure it would all be like Optane on CXL or whatever.
• 44:38
  Bryan Cantrill: Right, exactly. And you know, as they were, you know, they would kind of, and they would, I think they were good contributors to open source projects. I think they had done a lot in Linux and stuff like that. So I think it's like, that kind of, like they got a little bit, but they never really understood software. There was never really a true, they were never just bought into the importance of software.
• 45:04
  And I think they were also afraid to being bought into the importance of software, in part because they were kind of going, you know, in their world, Microsoft did the software, they did the microprocessor. And I think, you know, they weren't kind of slow to ultimately get the importance of Linux because they didn't want to alienate Microsoft. And which is just like, I mean, you're just, man, you're allowing Mike, like, why are you giving Microsoft is laughing all the way to the bank, right? Because they're getting all the margin. Right.
• 45:41
  And actually did remind in terms of like culture problems. So I mean, and you know, people were very upfront with Intel and especially like lately in the last like year or 2 years that Gelsinger was the wrong person for the job. But again, they'd be like, it's just not going to change. Like I'm powerless. I'm powerless.
• 46:02
  People at Intel feel powerless. Good technologists feel powerless. And that to me is a very, very deep cultural problem.
• 46:10
  Adam Leventhal: Yeah, and then compounded by the size of it. Right, because I think that sense of powerlessness and then, what is it, 130, 130 plus thousand employees, just when you pile on more folks into that morass.
• 46:23
  Bryan Cantrill: That's a lot of people. And so you have, so actually, you know what it reminds me of is it does remind me of bomber era Microsoft. And I don't know if you ever dealt with, with Microsoft in that era. Did your brother work with them at some point? I worked for
• 46:40
  Adam Leventhal: Microsoft in 1999.
• 46:43
  Bryan Cantrill: This I'm talking like 2000 and like 11. No. So like, Yeah,
• 46:49
  Adam Leventhal: actually my brother did work for Microsoft then. Yeah.
• 46:53
  Bryan Cantrill: So I think this was in 2012. I had a meeting with Scott Guthrie who was an exec, I think probably still is an exec at Microsoft. And this is, he was coming down to the node conference and, and the joint person arranged this meeting. It's like, Oh, this is like a very big deal. Like, okay, I'll like, I'll try to make time for it.
• 47:14
  I kind of don't care. I mean, I fine. And, you know, they, we kind of, I go into this room to meet with them I kind of had to spare whatever it was 30 minutes and they're like it's it's it was just you walk into the room and You can see the hierarchical problem, right? It's kind of Guthrie is there. He's got all the kind of the courtiers around him.
• 47:35
  And, but they start describing some of the things they've done with Azure. And in particular, he's like, I, I'm going to tell you something, but it's very, very confidential. I'm like, okay, well, this is getting, all right. And you got my attention. What's up?
• 47:45
  It's like, we have Linux VMs working in Azure. And I'm like, the fact that this needs to be a state secret, the fact that like, oh, no, I've got like something top secret. I, we have done something that customers want.
• 48:01
  Adam Leventhal: Oh, please not. In your plumbing.
• 48:04
  Bryan Cantrill: Right. We have delivered, cause at the time this is kind of amazing
• 48:11
  Adam Leventhal: to think, at the
• 48:12
  Bryan Cantrill: time Azure did not support Linux VMs. So... That's bonkers. Bonkers. And I'm like, that's okay, that's great.
• 48:17
  I mean, wow, how long have you been working on that? Obviously, part of the reason I wasn't taking Microsoft seriously in 2012 is because they didn't kind of get the fact that the world had moved long ago. The cloud world was all a Linux world But they'd I be I was like hearing what got 3 so I was actually like, you know what? I'm actually really impressed. I was I was kind of reluctant what it's been But then I'll like I can say like I am so impressed.
• 48:43
  I really have just 1 question for you all when are you gonna fire bomber And I have never seen anyone turn so red in my life. This guy, because he's bald and he's like entire, it's not a criticism. A bald is beautiful. Again, nothing gets bald. But his whole head went like this absolutely glowing red and everyone's like stammering and he's like, oh, okay, I've clearly hit on a nerve.
• 49:09
  Clearly everybody knows that Palmer is the wrong person to lead Microsoft. And, and then it was not that long after that, that finally Balmer was canned and they brought it in. It's not like it led this honestly, like pretty impressive Renaissance at Microsoft.
• 49:25
  Adam Leventhal: Yeah, really impressive. I mean, after like a lost decade, really.
• 49:29
  Bryan Cantrill: After a lost decade. And I don't know that that's quite possible. But I think that this is the reason that you can't count Intel out is a lots of people have now now Intel's got deeper problems, and they've got more of them. And the the person coming into this role, like 2021 was the time to really make some big changes. And you still had some I feel like you had just had a lot more altitude.
• 49:55
  They've got a lot less altitude right now. Yeah. So maybe it can we now get into the Gelsinger era? Yeah. Because can we, when do I get to talk about my issues with Gelsinger other than the, so I mean, issue number 1, so my kind of my foremost issue is that Gelsinger was coming into a cultural problem but he was not he didn't even treat the the country he felt like the company needed to be tweaked a little bit.
• 50:32
  And do you remember some of the things he did when he came back, by the way? No, I mean, I hope the AI PC.
• 50:40
  Adam Leventhal: Okay, no, I don't remember that at all.
• 50:42
  Bryan Cantrill: He had an ad campaign where we're gonna bring the PC back. And I'm like, Oh, my god, this is exactly the nightmare with this guy is that he was here during the dominance of the PC and he wants to go restore this PC era. So he's like, we have like, we've let go of the PC branding and we really need to talk about the PC again. You're like, what are, where are we right now? What are we doing?
• 51:04
  What year is it quickly? Because like, that's not at all what you need to do. Absolutely not. All of the wrong direction. Like personal computing is not the future of Intel.
• 51:15
  It can't be like the future of Intel has to be server side. It's got to be on the on the GPGPU. It's going to be on the network connectivity. It's like all this other stuff, like everything but the personal computer like Jesus.
• 51:25
  Adam Leventhal: Right. Right. I mean, even though there's a huge market there, but like you've lost it. You lost mobile and you're losing laptops and desktops are done, what are you doing?
• 51:38
  Bryan Cantrill: And fine, that's kind of the memory business. It's like dawn, you know what I mean? There are other things that you need to go into. And then so the thing that like, he should have done day 0, and the guy didn't have the guts to do it is he should have cut the dividend immediately. So Intel paid a dividend.
• 51:58
  And it was very mean, and this is like kind of a dated idea, right? This is the like tech companies. And I think you can argue that like actually the Googles of the world probably should pay a dividend, but they don't. I don't think they do anyway, right? Maybe Google did start paying a dividend.
• 52:13
  I don't know. Maybe they finally got, they said, but the, you know, the railroads used to be constructed whereby they would pay the dividend. It was the dividend yield of the company. That's what an investor would buy the railroad stock to get the dividend yield. In high tech, it's always been viewed as like, no, dividend doesn't make any sense because you are always 1 generation away from death and you just cannot afford, even for large companies, cash is king and you can't afford to give that up on a dividend and investors should make their money by selling the equity for a higher price than they buy it.
• 52:46
  And that seems to work. It's worked for a bunch of multi-trillion dollar companies now. So like it's worked for Apple, it's worked for Amazon, you know, worked for Nvidia. Intel was paying a dividend, like an absurd dividend, like a, like multi billion dollar dividend every year. And that had to be cut day 0.
• 53:14
  And I think that would have been, and he didn't want to do it because he didn't want to tank the stock price. And this is where it's like, I think this is what that's bought. Like, is this the return of the engineer or not? If this is the return of the engine, like, or is this like the second coming of Bob Swan? Because this is the return of the engineer.
• 53:30
  Like actually take a page from Jeff Bezos book and like don't care about the stock price here for a second and care about like the future that you need to go build and you've got to get in a different mode. But they didn't do that. And they ultimately, of course they did kill the dividend, but he killed the dividend 4 months ago, 3 months ago, and he should have killed the dividend. You know, 3 years ago. Yeah.
• 53:56
  So that I felt was like, absolutely the I think that was a huge mistake to continue to pay the dividend or to not cut the dividend. And then I think it was they needed to go really invest in a bunch of new efforts. And the thing that he wanted to invest in was all around process nodes. Was it 5 nodes in 4 years? I can't even remember what his like,
• 54:22
  Adam Leventhal: what is like a chicken in every pot kind of mantra.
• 54:26
  Bryan Cantrill: Chicken in every pot. It's exactly what it is. It is a chicken in every pot for fabs. We're going to have, we're going to have a node in every pot.
• 54:34
  Adam Leventhal: And also get it being in the Foundry business, right? Like, which was a big early initiative, right? Like, you know, catering to fabulous chip companies to compete with TSMC.
• 54:49
  Bryan Cantrill: Yes. Which to me is like, okay, that is a very new motion because what you're going to say is that like we have all the best of my knowledge, people are correct to me, by the way, the chat, okay, fine. All of these ridiculously profitable companies do work, did finally have dividends, Apple and video, Microsoft, all of the dividends. Those are things that they Google, but those are things that they had to be coerced into many years after being wildly profitable. And they, if those companies begin to tank, they should get rid of their dividend too.
• 55:20
  Adam Leventhal: This is not financial advice.
• 55:23
  Bryan Cantrill: Oh, excuse me. I thought I, but I, but then why is my finance, my, why is my financial advisor hardware engineer attorney? This is, this type of stock is
• 55:32
  Adam Leventhal: your attorney. That's right. Your, your post-mortem attorney,
• 55:37
  Bryan Cantrill: my, my, my post-mortem attorney, I did fire that guy. That guy I felt like I got a guy. I got thrown under the bus. But he's all right. So he was going to do a node in every pot.
• 55:47
  And then he's going to do, but he's going to have it be like, we're going to have the IFS, Intel Foundry Services. And that to me is like, you know, what you're kind of talking about is we're going to disentangle this thing that's been, has only had an internal customer and we're now going to have external customers. Like that to me is always a very ambitious thing. Don't you think it's a very ambitious thing for a company to take on?
• 56:10
  Adam Leventhal: Oh, tremendously. Yeah. Like to say that we have been the only customer, all of our process is around this to then sort of take an axe and sort of like make a big chop and and sort of like hope it works out very challenging.
• 56:26
  Bryan Cantrill: Very challenging. Like in some ways like more challenging than starting from scratch. In fact, you probably do want to start from scratch. And I mean, I think that like there were some interesting ideas there. But like if you're going to compete, because basically Intel is like Intel, the challenge that Intel had in 2021 is you are losing to the following 3 companies, TSMC, AMD, and Nvidia.
• 56:53
  And their idea is like, we're going to compete with all of them at once. It's like, you've not been able to compete with 1 of them. Like, And you're, I don't, I just don't understand how you're going to be able to go for all of them at once.
• 57:08
  Adam Leventhal: I'll just think, going to add 1 more to your list. Do you know what the market cap of arm is right now?
• 57:14
  Bryan Cantrill: No, What is it?
• 57:15
  Adam Leventhal: It's like 147 billion. It's like, Oh my God, 50% more than Intel. It's wild.
• 57:20
  Bryan Cantrill: That is wild. Yeah. So you compete with arm too.
• 57:23
  Adam Leventhal: Yeah. I mean, arguably.
• 57:25
  Bryan Cantrill: Yeah. And you're losing and you're losing to all of them. And, And, and then you know, you've got these kind of these other parts, as someone that Chad is pointing out, and the Chad is pointing out that also competing with Broadcom, which is definitely the truth. Also, yes, by the way, you're all well, and this is actually the thing is like, they were competing with Broadcom with Intel Tofino That was competing with Broadcom. And, you know, with Gelsinger's return, 1 of the things I was concerned about, it's like, I'm very concerned for Tofino.
• 57:56
  And it turns out those were concerns that were well-founded because Intel killed the roadmap for Tofino. And then Intel or the Shear killed Tofino and they killed Tofino really sloppily. And they killed it with very little time for last buys. And they kind of, they didn't follow their own process and it was a mess. It was a really, it was a big mess.
• 58:24
  And we were going to get to a better place with it because, and we're really excited for what we're doing with Excite Labs And we'll talk about what we're doing with Excite Labs and their X2. We've definitely had podcast episodes to come on that because we're, and when they launched it at the Open Compute Project Summit here, OCP Summit, we've like, we're excited, really excited for where the X2 is going. So we're going to actually land in a better spot. But man, the way they did this was bad. And it was really frustrating because it's like this is the engineer.
• 58:58
  The engineer just killed the thing that I actually think is like what it's only the only Intel piece of silicon we're using. I think it's actually as a piece of silicon, it's pretty interesting. I think it's an architecture. It's pretty interesting. I think there, it wasn't perfect, but there were some things that like, you know, Nvidia bought Mellanox in 2019, 2020, either way down to 2019 or early 2020.
• 59:22
  And at the time people were like, that doesn't make sense. And then you begin to realize like, no, no, no, that definitely makes sense because they understand what these like large training workloads end up really hitting the network hard. It's like, okay, so if you want to go compete with Nvidia, Nvidia bought Mellanox and you're killing Tofino. Like does this make any sense to you at all? And then this is where Adam, you got me off because you're just like, and then they wrap themselves in the flag, which just damn it is a piss me off.
• 59:55
  Adam Leventhal: Which flag the Intel flag or the American flag?
• 59:58
  Bryan Cantrill: The American flag.
• 59:58
  Adam Leventhal: Oh, because they're like, we're not going to fab it at TSMC.
• 1:00:03
  Bryan Cantrill: You know what? That would even be more coherent. No, they are, these are literally unrelated events. They are killing to Fino. You know, if they were killing to Fino out of an act of American patriotism, it'd be weird, but I would like give it some, not like, all right, that's very strange, but
• 1:00:16
  Adam Leventhal: you're saying like going in and Hat in hand for the chips act.
• 1:00:21
  Bryan Cantrill: Yeah hat in hand for the chips act yeah, and hat in hand for the chips act and America can't let us go out of business and we and it's like, you know And by the way, like I any interest in what your federal government customers think about you killing to Fino because, you know, they should probably go ask that question. Because they were, you know, among the, the Tufino was an interesting chip. Yes, it was a forward-looking chip and no, it would, no, it didn't have the ubiquity that they had kind of hoped for, but it was, It was a good idea that was on the right trajectory, but they killed him. And when they did that, they left a bunch of people high and dry and including the people that are, are supposedly the people that aren't going to let them go out of business. And you're just like, okay, this is like major cognitive dissonance that you and III, and I really do not like, I think that Gelsinger, among his many transgressions, I really do not like the fact that he wrapped himself in the flag.
• 1:01:30
  And I think it's ridiculously entitled. And as a taxpayer, it's like, we, as a taxpayer, we should not be bailing out your cultural problem. Like we should not be paying for your middle management load. I'm sorry, because you don't have the guts to actually make the cuts you need to make.
• 1:01:43
  Adam Leventhal: So I'm gonna agree with you. And I would- Right, we'll
• 1:01:47
  Bryan Cantrill: just say- You can cut it all out later.
• 1:01:49
  Adam Leventhal: Well, there we go, we can cut all of this. I would say having read chip war, I did feel like the, the patriotism associated with the, or the, the defense necessity or impetus around having domestic production of chips is interesting. I think like-
• 1:02:08
  Bryan Cantrill: Oh, I
• 1:02:08
  Adam Leventhal: think it's very interesting. I think if you set it, I mean, obviously you gotta set aside all of like all of the brokenness of Intel to see it as exclusively a good idea. But it's also the case that like TSMC benefit tremendously from national investment. Same thing with SMC. I think the same thing with Samsung, although a bit more obliquely.
• 1:02:35
  Anyway, and- But TSMC
• 1:02:36
  Bryan Cantrill: is also building fabs in Arizona, right?
• 1:02:39
  Adam Leventhal: I mean- Also true, also true. Although not like the-
• 1:02:41
  Bryan Cantrill: I believe that TSMC got their money under the HIPPS Act before Intel did.
• 1:02:44
  Adam Leventhal: True, but apparently also not, I think, on the newest process node, but you're right, still building fabs in Arizona.
• 1:02:52
  Bryan Cantrill: And I mean, I just think that is a real mistake. When, when companies start to wrap themselves in the flag and say, you can't let us go out of business that this is beginning to feel like the automakers in the seventies. And it's like, maybe we can let you go out of business. And I mean,
• 1:03:11
  Adam Leventhal: when it gets to that point, it's sort of too late, right? Like if you had a healthy, if you have a, like a healthy culture and are looking for that investment for the future, yes. When things are circling the drain and you're saying, prevent us from failing, trickier.
• 1:03:29
  Bryan Cantrill: Totally trickier. And I would say, well, how automakers in the 70s do you mean like the 2010s? I actually do mean the 70s because in the 70s, their products sucked. The 70s and 80s, the automakers, I mean, and you know, read about Lee Iacocca and Chrysler and the, and on the 1 hand it was, and the products got a lot better. They learned a lot from just getting their butts kicked by what was at first Japanese and then later Japanese and Korean competition.
• 1:03:59
  And I just think it's really problematic when you have, again, it's like we should be asking tougher questions. And it's like, okay, so what are you gonna, so what needs to be done? I mean, I think for starters, they, and I think that they're in a dire position, just flat out. They've got a huge debt load. They, the incoming CEO has got a lot of work to go do and you've got to cut a lot of folks ultimately.
• 1:04:26
  And that's what you're unwilling to do apparently.
• 1:04:28
  Adam Leventhal: But such an interesting position, right? Like they are, they feel so like just fractally broken and yet have such market share dominance. Like has there, are there examples in history of companies that, that feel like on such a downward trajectory and yet have, are such, are like the dominant player?
• 1:04:50
  Bryan Cantrill: I think that this does kind of remind me of the automakers.
• 1:04:53
  Adam Leventhal: Yeah, interesting.
• 1:04:55
  Bryan Cantrill: Where they were still dominant, but very clearly going to lose that dominant. I mean, this is the problem is like that dominance is looking in the rear view mirror. That's the problem. Yeah. And I don't necessarily, I don't think Intel is out on, but I mean, you sort of look for example on, he said really only dominant and kind of 1 market, right?
• 1:05:15
  They're not dominant on the GPU side. No. I think like Gaudi is interesting. You know, they bought Habana labs. They then totally screwed that acquisition up.
• 1:05:29
  And like ended up, well before, and the, that was like in, you know, whatever that was, 2012, and let that all kind of go to seed. And then they bought Habana Labs for 2 billion. And then they let that executive team, that executive team all left. And I think Gaudi is like interesting, is technically interesting, but it's really, I mean, as I told people, as I told the time, it's like, when you kill Tofino, I can't look at Gaudi. Do you understand why it's like, no, they do not.
• 1:06:08
  Like they don't like, how would these things possibly? It's like, because I've got a trust issue and I can't, like I would be, it would be negligent reckless for anyone to build out on Gaudi. Yeah. I feel the same way
• 1:06:23
  Adam Leventhal: about their foundry business where, you know, how do you, how do you bet your business on that? Which is really not going to be portable to some other foundry. I imagine to know like it's at the whim of whoever's in charge that
• 1:06:36
  Bryan Cantrill: day. That's right. And then the other thing is like they've also killed things that were then successful. Like Tofino had like, Tofino, yeah, I get it. I'm sure it wasn't paying for itself.
• 1:06:45
  I'm sure the P&L looked grisly, but you know what? It was a product that customers actually liked. There was also Spring Hill, which was this, Facebook was a huge Spring Hill customer. And when they bought Havana, they killed Spring Hill. Spring Hill was a skunkworks effort inside, I think it was a homegrown effort inside of Intel.
• 1:07:05
  When they bought Havana, they killed Spring Hill and Meta apparently, it was, was met at that point. And I think, Adam, we, I think we agreed that we are going to call the Meta as punishment for them changing the names. I read, I
• 1:07:15
  Adam Leventhal: can't cover our. Agreed. That's right. Punishment. That's right.
• 1:07:17
  Metapeties.
• 1:07:23
  Bryan Cantrill: You're remember your Metapeties. Yes. Metapeties.
• 1:07:26
  Adam Leventhal: Mine. The Mets. That's right. The Metamates.
• 1:07:29
  Bryan Cantrill: Wait, wait. Oh God. When Zuckerberg announced we were all going to be called now Metamates and you pointed out that this is actually going to be pronounced Metamates in the original Greek.
• 1:07:41
  Adam Leventhal: The Greek social networking site. Yes.
• 1:07:45
  Bryan Cantrill: But apparently Meta after that was like, we are never going to be an Intel customer. I don't know if that's an exaggeration, but I think that when they killed Spring Hill, it was really alienating to people that Metta was building out on Spring Hill. They spun out Solidon because they didn't want to be in the fight and they kind of viewed that, I think it's like, you know, this is like our tough stand and you know, we don't want to
• 1:08:12
  Adam Leventhal: be in the flash business, the SSD vendor. Yeah.
• 1:08:15
  Bryan Cantrill: Yeah. But like, that's actually like, that would be like dropping a bunch of capital on the bottom line right now. That was a mistake Yeah It's just like a mistake mistake mistake mistake mistake mistake and then you got 18a which is you know the 1 1 of the nodes in 1 of the pots. The age feels, because this is now we're down at angstroms, which is like now this is like we're acknowledging that this stuff just doesn't mean any, has no relationship to the physical world, it's just a coincidence. But 18A, do you remember, there was this thing where 18A's pipeline is growing week over week by like billions of dollars.
• 1:08:55
  But they won't name who the customers are. And I've got reporters who are calling me asking like, do you know anything about them? I'm like, they're just saying things. And I still don't know anything about 18. In terms of like, is 18 a gonna deliver?
• 1:09:07
  Is it not going to deliver? Absolutely no idea. We we know very little about and maybe folks in the chat know more about the process know, but I feel like we don't know a lot about it. And I feel that like I don't feel that Gelsinger being finally ejected is a positive sign for 18 a, because Gelsinger has been saying a lot of positive things about 18 a and it's very unclear to me that. So I
• 1:09:34
  Adam Leventhal: gotta say at the top, you talked about your optimism for Intel. And I gotta say when you were talking about your optimism, I keep hearing so much about your optimism for Intel. You know, I feel like I've been through this a bunch. Like I had a like over a couple of beers at at a solid I'm conference actually discussion about how just I see no path forward this few months ago, no path forward for Intel and someone fighting me tooth and nail saying, you know, long Intel would invest in Intel. It believes in the future of Intel.
• 1:10:05
  And I just think, you know, Gelsinger's out, apparently with relatively short notice, right? Like this does not feel like the, like a carefully coordinated orchestrated.
• 1:10:18
  Bryan Cantrill: And out with a lie. I mean, I do think it's appropriate that Gelsinger, that, that, that Gelsinger's tenure as CEO ends with the kind of bullshit that, that it, the, The idea that he's retiring, it's like, come on, come on. And I mean, with no succession planning, it's a bit like, oh, I think he might've been fired. It's like, oh, you think he might've been? Yes, I think he might've been fired.
• 1:10:42
  Yes, of course he was fired. Yeah. Of course he was fired. By the way, the board should fire themselves next because the board bears from, I mean, the board made this call the higher this guy and
• 1:10:55
  Adam Leventhal: it's like him and to
• 1:10:57
  Bryan Cantrill: and to stick with him
• 1:10:58
  Adam Leventhal: and to fire him with no succession plan. And I love that despite all of this, you're like, I I'll take the abyss, give me the leap into the abyss over, over this, over whatever was in the cards, over the note. I never thought.
• 1:11:16
  Bryan Cantrill: No, it's like, I would rather like the abyss can't be worse than this. So I do think that like, I think there's a degree to which that's true, because I think that Gelsinger was, and I think also Gelsinger was like surrounded by this middle management layer. And again, I hold him responsible for it. I don't hold like the, the, the, the gooey layer around him responsible for it. Cause the gooey layer is just doing what the gooey layer always does.
• 1:11:46
  And the, you've got to cut through all of that. And I gotta say, I gotta say, you kind of need to go founder mode on this thing. All right. You got me. Like there's a company that has been locked in manager mode and you know, and this is like, maybe you kind of need to go founder mode, not in the way that we kind of recast in founder mode is like trust and clarity mode or just kind of an oxide way of thinking about it.
• 1:12:14
  So I don't know, I think you may need to go like medieval founder on this thing. I think you need someone who can come in and, and be like, no, no, sorry. We're not cutting 15%. We need to have 1 fifth of the current employees here.
• 1:12:30
  Adam Leventhal: I mean, chat said, Elon for CEO. And I feel like that's what you're saying.
• 1:12:36
  Bryan Cantrill: You know, I kind of scared myself here a little bit. I think, but I think you need that kind of, you need to like, I also think that it's a false dichotomy to think that you can't do that. You that you're doing that in a way that's devoid of empathy. Because 1 thing I will say about Intel, by the way, Intel, there are plenty of people at Intel who have been making a nickel off of the demise of this company. They have been because Intel has been such a bad place to work for so long, at least at these kind of and when you see kind of this high profile person that goes to Intel, and they're there for 2 years and they leave, it's because they got paid.
• 1:13:17
  People were making a lot. And I, I have had 2 people, I mean at Oxide, like the compensation thing is like pretty clear, right? Pretty clear about our conversation. I've had 2 people, 1 before we disclosed it, 1 after we disclosed it who have walked away from oxide over comp both coming from Intel The person Intel and we weren't even at the point of an offer we were at the point like okay, we're having a conversation It's kind of interesting. Like I think this is like, you know, do we want to You know what we're exploring kind of going to the next steps and this person's like well we need to talk about compensation.
• 1:13:49
  I'm like, well, that's probably like no problem because we've already talked about compensations on the blog. So, you know, I know that's what you say, but I actually, I know the real story is that you're paying everyone differently. It's like, what really? Okay. And he's like, and by the way, like, I need to be making like 400K a year.
• 1:14:05
  Adam Leventhal: And also that, that values thing about transparency, like I know that's also bullshit.
• 1:14:10
  Bryan Cantrill: Well, like those are all lies. So what's actually, and I'm like, wow, okay. And also someone who was like, not someone we were like unequivocal on someone who was like, I don't know, maybe it's worth reaching out to. It's like, this is actually like, you are doing me a huge favor. I got very helpful.
• 1:14:25
  Adam Leventhal: Yeah, that's great.
• 1:14:26
  Bryan Cantrill: Super helpful. And I really appreciate it. And then another person is actually part of the reason we publicize the comp is because we had another person who we actually did want to make an offer to and his salary demand was 3 X the in order to meet his Intel salary, we'd have to pay them 3 X the oxide salary. And I was like, Oh my God. And I know there are other folks in there that are making that is so bad made a lot of money.
• 1:14:54
  And I would I mean, I think the new CEO I've already CEO I would start with a spreadsheet that has got everyone's name and everyone's salary next to it and start slicing that way. But that's the kind of disposition you need. Like, yeah, I don't care. You're the highest performer. You're great.
• 1:15:11
  Okay, that's great. You're not making 800k a year here. We're a company that is like in crisis.
• 1:15:18
  Adam Leventhal: Yeah, you need the Pat Gelsinger who can conjure tens of billions of dollars.
• 1:15:23
  Bryan Cantrill: That's right. I need people who can make billions of dollars in their money. No, but you got to like go through and really reset the company. And I do think that, so this is the thing that I think I am opposite, where is this? Is this the optimism you've been talking about?
• 1:15:38
  I think that like, although actually that said, I hate the way they're messaging it because like what they actually need to be doing, sorry to be graphic here, is you need Gelsinger's decapitated head on a pike at an all-hands at Intel and Being like that's the last guy things are gonna change here and like you don't need to be bought into it. And like, you may want to leave, but like, this is a company that is in crisis. We need to do some radical things. And if you, and it got, I really, I'm kind of sounding like Musk here. I'm a little scared about myself here.
• 1:16:19
  This is like, but I know, I think it's like, it's actually not unreasonable to ask people to really buy into something that is like an exciting rebuild, but you like you need to be like AMD was dead it was dead when Lisa Su became the CEO it had been left for dead and That is where we that that's that's kind of where Intel needs to think of itself Intel if you don't change trajectory, you will die. And that is the kind of level of urgency people need. Sorry about the, all right. Maybe I was a little bit too graphic there, but the, III think that that is the level of urgency people need.
• 1:17:07
  Adam Leventhal: Yeah. And you need, I don't know if they're going to be capable of it. Right. Like I,
• 1:17:11
  Bryan Cantrill: I don't know. I don't know.
• 1:17:13
  Adam Leventhal: I, again, I mean, I don't want to make this about sun again, but like that's the tough love that we needed, right? Like, yes, we, the CEO, he would say, like, we are on a trajectory to death, to death. And instead, we got, you know, Jonathan Schwartz's consensus and kind of like not visiting customers and kind of slow burn,
• 1:17:40
  Bryan Cantrill: you know, business as usual kind of stuff. And you needed something a lot more radical. And I think that you could do and again, I think you can do something that's more radical and I think you can do it away with empathy. I know when you're it's like the same person just said like the word head on pike. I was also talking about empathy.
• 1:17:52
  You hear yourself. I know. But I, I mean, I think that you but the problem is like, folks, you are all living in an unsustainable economic vehicle. And that's the reality. The like, and you've got to be able to really convey that.
• 1:18:14
  And I agree with you. It's like, and I'm not even sure this is like, you know, with Jared Diamond, I'm a little bit concerned that I've been kind of Malcolm Gladwell on him. And now Michael Lewis a little bit too. Yeah. You know what I mean?
• 1:18:28
  Like you must have very, I mean, Michael Lewis, the whole like blind side thing, I'm now feeling like, oh God, like I can't believe anything that I read anywhere right I Loved Jared Diamond's collapse. I think it is too a little too reductive, but according to Jared Diamond in collapse the the mean that the So the Norse are in Greenland and go extinct in part because they refuse to eat fish. And you had the Inuit around them are eating fish and are surviving. People are surviving in the Arctic. You have peoples that are surviving in the Arctic, but they're doing it by eating fish.
• 1:19:02
  And meanwhile, you've got the Norse that are trying to like grow grass and feed goats and eat livestock. And it's like, yeah, you're going to die. And they died. They did die. So, Yeah, Intel may choose to not eat fish, in which case, you know, if only heathens eat fish then yes, you will die.
• 1:19:21
  And so that's kind of the question. If you can have someone come in with that level of urgency. And so this does get us to like, but That said, okay, so you can, you kind of, you know, again, transport yourself back to AMD. Like you can go do interesting things. Like we're gonna go do chiplets because we have to, but by the way, chiplets are really interesting.
• 1:19:44
  And I think that like, I think that 1 of the things that would be super interesting is that you can go get like in the Foundry business. And I understand it's not clear to me, but it sounds like they can't actually spend the Foundry business out as part of their chips act money. I don't know. I mean, what did you get? I don't know.
• 1:20:01
  I don't have a sense of that. So I'm not sure that they can, but 1 of the things that they can go do is that they can actually go rethink the way a foundry business looks. And so in particular, they're, do you have to listen to Invest Like the Best?
• 1:20:18
  Adam Leventhal: Sometimes. That's a great podcast.
• 1:20:20
  Bryan Cantrill: It's a great podcast. That is Steve's favorite podcast. I mean, I love it too. I know. But, I mean, Steve is rightfully an ardent listener to Invest Like the Best.
• 1:20:29
  And that's like the best thing is great. So they had a very good invest like the best recently. I got to get the analyst with a semiconductor analyst, but they were talking about the semiconductors, which is now like with AI is really exploding with interest. Like there's a lot of things that are interested, but some is now and he was the Sean is he's asking him. Why is it so expensive to have 1 of these startups?
• 1:20:57
  And I'm thinking like, okay, this is going to be really interesting. It's like, obviously, like you and I have not worked on an ASIC and I've got my own sense of that, but I'm super curious to hear someone else's sense of that. And I got to say my own sense of that did actually match what the first thing says, like we have to understand the software is all proprietary and really expensive. The tooling is very expensive because you're buying this broadly cadence software that is a fortune and the it's like well that's software and intel could like a desperate company could give that away.
• 1:21:39
  Adam Leventhal: You're right.
• 1:21:39
  Bryan Cantrill: So what if like Intel to like be on Intel boundaries? Like, no, no, I know, I get it. We're not TSMC, But our software is all free by the way, and it's all open source. How about that?
• 1:21:48
  Adam Leventhal: Yeah got it like pin it all on an open source Tool chain. Yes cost of getting in there like is is very very low
• 1:21:58
  Bryan Cantrill: That's right. And then you say like okay, and then hey government. I want you to take our chips act money. I want you to take a billion dollars of that. I want you to earmark it for startups that are gonna be that invest in startups that are gonna use our tool chain, which by the way is open source.
• 1:22:14
  Adam Leventhal: Yeah. Step 3 oxide applies for that money. I see where you're going.
• 1:22:18
  Bryan Cantrill: Listen, I was going to leave step 3 as an exercise. It's like, so I'm like, you know, we all know that we're you, we need to do our own trust at some point. This is the only, this is the ad. We also know that I refuse to be a cadence customer. So, you know, this
• 1:22:33
  Adam Leventhal: is a real, real nice secret plan we had there for a second.
• 1:22:37
  Bryan Cantrill: Listen, this is the podcast is the giant sneak attack, but like that could be interesting, right? Like that gets kind of like, okay. And then you can also imagine like okay. We've cut the company way down But by the way now we put this beacon out there and like now you can attract some pretty interesting people to the company Yeah, we're like hey, I'm kind of like yeah, I like even if this is like a, you know, yeah, I think this is a long shot But this is like an interesting long shot. Like this has now become fascinating Someone said earlier asked about a silicon photonics silicon photonics another thing I think is like super interesting that Intel had was doing really interesting things in.
• 1:23:18
  And I think they've killed that as well, unfortunately. So that's another area that is interesting. I think that like the AI, the stuff they've done with the 1 API stuff, I think is interesting. But they, again, they've got to find like, the market doesn't trust them there. So it's like, that's going to be really hard for them.
• 1:23:39
  Adam Leventhal: So I love all of this, but to what degree does this play to their strengths? No, like to what degree do we like? I think it's a great plan. Like you're hired as, as a, shareholder of Intel.
• 1:23:50
  Bryan Cantrill: I want you to dial down. So I understand this, this paper mache head that you've been working so hard on. We want you, we want to dial that down a little bit.
• 1:23:58
  Adam Leventhal: It's looking great. Let me just, I just want to, I want to reemphasize that point.
• 1:24:05
  Bryan Cantrill: And I can see that you've worked very hard on it.
• 1:24:08
  Adam Leventhal: That is a lot to you.
• 1:24:10
  Bryan Cantrill: And I know, I understand that what you did with the balloon and you told me about that, you told me about that a couple of times actually. I know that's, I can see, you know, okay. We need to stop talking about it. Okay, let's put the head down. Yes.
• 1:24:20
  And?
• 1:24:21
  Adam Leventhal: But like to what, you know, a good software, foundry business, we talked about those are, are like potential vulnerabilities, not potential strengths. And you're saying like, you need to bring in not just the CEO, but then the person that that CEO can bring in that can really execute on those things in a company that resists change that like manufacturing memories when Gordon Moore is like fucking locking the doors
• 1:24:50
  Bryan Cantrill: to the foundry and telling them to kick them out. Okay, I know. I hear what you're saying. I hear what you're saying. And I think we should use the paper mache head in the all hands.
• 1:25:00
  Yeah, I hear you're saying. No, I agree with you. And I think it is a challenge because you need a CEO who can be, who can really at once go total chainsaw out on this thing. Yeah. And also really inspire people with a vision towards the future and then execute on all that.
• 1:25:20
  Adam Leventhal: Well, yeah, and not get sued by shareholders who see this as a huge role of the dice. Right. And I think that that is a big problem with in general with corporate culture, where it makes it hard to do the only sensible kind of bet that has any kind of chance for a future for the company as opposed to saying look we we used to have a 400 billion dollar company now I have a 100 billion dollar company and I don't want to turn it into a $10 billion company. So I need to sell the pieces or milk when there's left.
• 1:25:52
  Bryan Cantrill: I agree with you though. I think Cheryl does pretty much sent the message today that like a roll of the dice actually sounds kind of interesting right now.
• 1:25:59
  Adam Leventhal: Yeah. Yeah. I mean, I think like a role, but I think like,
• 1:26:03
  Bryan Cantrill: it might not like that.
• 1:26:05
  Adam Leventhal: Yeah. I kind of like that. Like kind of, can't you just like do the thing that has been working for the last 40 years?
• 1:26:12
  Bryan Cantrill: No. Yeah. Right. We can't do that Because that is the path of death, unfortunately. And no, we can't do that.
• 1:26:19
  And this is why the board should fire themselves, because we can't do that because we made many mistakes on the way. And we made mistakes, many mistakes over the last 3 and a half years. We made many mistakes on the 2 years prior to that. And it's like the mistakes have compounded. And so, no, unfortunately, we can't do that.
• 1:26:40
  We've got to actually go, we've got to think much more radically. And so, you need someone who can actually, and this is why, And you get to like, all right, who are the candidates? Because I think the other problem is like, you know, the person that you would want for this job isn't necessarily gonna take it. This is not necessarily, this is gonna be really hard. You're gonna have so much resistance.
• 1:26:58
  You're gonna have resistance from shareholders, the board, employees. I mean, like, why? Right? And so you, it's gonna be hard. It's gonna be hard.
• 1:27:10
  I do think so. Litbu Tan was the founder on or with a founder, certainly very early CEO of cadence, no longer the CEO of cadence on Intel's board was on Intel's board and supposedly put his hat in the ring. And, when they were considering Gelsinger and he would have been a much better hire than Gelsinger. No doubt in my mind.
• 1:27:34
  Adam Leventhal: And, and resigned from the board, relatively recently. Is that true?
• 1:27:38
  Bryan Cantrill: Yes. We're signed from the board because the board refused to do what they needed to go do. And I think that the, lip, boo would be interesting. Laboo is definitely not like if it moves to CEO, think you can throw out my plan of the open source Foundry. I think we can safely say that.
• 1:28:01
  Okay. So that's on the, on that's on the back burner. Then do I understand that correctly? Oh, okay. Oh gosh.
• 1:28:05
  Go fuck myself. Got it. So I think that that, is, and it's kind of like the Intel board does have the stomach for a 4 year turnaround. Gelsinger was not on a trajectory to turn the company around, folks. They don't have the stomach.
• 1:28:22
  They have finally summoned the courage to get rid of a failing CEO. This is not on. This was a turnaround that there is no sign that this turnaround was working. So the idea that like, oh, they just don't have like the courage, just, it's like, no, nope, don't buy it, sorry. Don't buy it.
• 1:28:39
  They, or like, if you want to be, if you show me that you can turn this thing around by getting your revenue per employee back in line and go make some really tough decisions. And that does not mean cutting 15, 000 people. Sorry, just doesn't. Or do like go cut salaries. And that's the other thing you can go do, which I think it would be an interesting thing to go put on the table.
• 1:29:00
  Yeah. So lip-boo would be good. I think in that liberal would be interesting. And if I don't think you would take it right now, but that boo is not who I think is the person for the job that I would say if I could pick anybody.
• 1:29:12
  Adam Leventhal: Okay. Who would you pick?
• 1:29:14
  Bryan Cantrill: I would pick Navin route and sorry, Navin. So I don't know if it means a listener to the podcast or not. It's not inconceivable that he is. If he is, he just like, you know, whatever beverage he's drinking just came out his nose. Sorry, Naveen.
• 1:29:31
  Naveen is the, so was the founder of Nirvana that sold the Intel, was at Intel for a couple of years and really tried to get Intel on the right path with respect to AI. Gave up out of frustration, started a company called MosaicML and MosaicML was bought by Databricks maybe 2 years ago. So he's probably not, I think he's pretty bought in the way Databricks is doing. And just to fly away. But Dev, you're meant to be.
• 1:30:00
  Adam Leventhal: Never.
• 1:30:01
  Bryan Cantrill: Devine's great. Formerly, like at Intel way back in the day started his career at Intel. When actually to our alma mater to get a PhD in neuroscience, because felt that like, I, this is like a domain I want to understand much better got a PhD in neuroscience started Nirvana sold Nirvana to to Intel and then it became and I like super interesting guy and I think that like that's the kind of person you need, you know, now there's just, I just don't think there's any way that maybe it would do it. The, but that would be kind of my, cause you're going to have to, like, I don't know that you, But that's the kind of the complexion of the person that you want is going to be someone who can really affect a massive amount of change and can do it with a way that's going to be inspiring for folks and still do the really, really, really tough stuff that they need to go do. Yeah.
• 1:31:02
  And the something in the chat, like, I'm not sure the darker future exists with a $7 billion inbound from chips. $7 billion is a drop in the bucket for this company. That's how severe this company's problems are, by the way.
• 1:31:13
  Adam Leventhal: So What about the darker future of being acquired, of selling off parts of it, of private equity? What do you think the answer of those things?
• 1:31:24
  Bryan Cantrill: I don't know. I don't know. I think that those are kind of like, I kind of think that's what should happen for Gaudi. Honestly, I think Gaudi gets more interesting as a private equity play, and that's something that they probably could be bought by someone. I think that would be kind of like fascinating, you know, for like, and that would be, I mean, actually, so, okay, now we're just in like crazy land, but like, what if meta were to buy Gaudi?
• 1:31:52
  Yeah. And because in that has already proven that they're I mean, they're actually I kind of like, respect to meta for like, I mean what they're doing with llama has been with llama 2 has been huge, right? And the The willingness to be like to really kind of disrupt what ai looks like by making it available and like meta buying Gaudi would be kind of fascinating. Yeah, interesting. Or Snapchat.
• 1:32:30
  It'd be kind of interesting. Like if at some point someone's going to be like, Hey, you know what? I've actually done the math on the GPUs we need to go by. And I've done the math. I'm buying Gaudi from Intel and it's going to be cheaper to buy Gaudi.
• 1:32:46
  And so like, I think that is, I don't know. What do you, what do you, what, yeah, What do you think? I don't
• 1:32:55
  Adam Leventhal: know. I just, I mean, it's, I feel like it's gonna kind of take this long path to cratering. I don't know what it looks like, but it just, I don't know how they stay aloft. And I don't think,
• 1:33:09
  Bryan Cantrill: you know- But how do you, so who's the CEO that you bring in? You're on the CEO's search committee. You have woken up and say, they should do a Choose your own adventure from the Intel CEO search committee.
• 1:33:23
  Adam Leventhal: Yeah. I mean, I wouldn't be surprised if it was an insider to,
• 1:33:27
  Bryan Cantrill: Oh, that'd be such a mistake.
• 1:33:29
  Adam Leventhal: Agreed. But an insider to, to navigate more closely what the board wants, which is just like preserve value, sell off the things that are worth something, make some deals, have a dedicated federal division that is for hoovering up chips act money.
• 1:33:50
  Bryan Cantrill: Yeah. If you're going to do that, like if you're just going to, like, I could do that. I can go be a butcher. Like you see what you're looking for is a butcher is what you say.
• 1:34:01
  Adam Leventhal: I think so because I, a real change agent, like why do you take that spot? Like, a position that might've been interesting 3 years ago is not interesting now. And, And I think that the kind of person for whom it would be exciting to have this historical company, this unique market position, would have the wrong profile for a risk averse board. Again, I just don't see them rolling the dice.
• 1:34:34
  Bryan Cantrill: Yeah. When you think of like people like, as you look at like technologists to help lead this transition, I mean, Jim Keller has already been in and out. Raj has been in and out. I mean, a bunch of the folks that you would look to have already been in and out. So I don't know, maybe you can convince, although they're 10 stories straight, but raised a bunch more money.
• 1:34:50
  So I don't think that's going to happen. This is the other thing is like, somebody's are hot right now. So like, why, if you, why would you leave your like interesting startup or interesting company to that's the other problem. Yeah.
• 1:35:03
  Adam Leventhal: Like, so
• 1:35:04
  Bryan Cantrill: yeah, I don't know.
• 1:35:05
  Adam Leventhal: It would not be fun, right? Like, or are there very few people for whom this sounds like a good time?
• 1:35:12
  Bryan Cantrill: I think it would be, You got to be a little bit warped. But you, you would have to see something there. You would have to know the company well enough to see some real promise. And I'm not sure. The problem is in like, this is the people like coward that have been in and out I mean or I should say like our because I want to speak for him for Jim But the if you talk to people who are ex Intel, it's always pretty they're pretty frustrated about Intel That's what the Havana like why did the Havana folks leave like go ask them that and You why did the beam leave and like there are some pretty, like, concrete cultural answers to that.
• 1:36:00
  But Like, I don't, the alternative can't be just, you know, it just can't be business as usual, I think. You know, I think you
• 1:36:08
  Adam Leventhal: gotta- I hear you, like I agree. And I think, I don't know that there's a path forward that's not business as usual and like the slow descent and then like lobbying for keeping them alive. You know, we're not going to end up with a single source on CPUs in the same way that AMD wasn't allowed to go out of business. So I think
• 1:36:31
  Bryan Cantrill: Or Blackberry, which is probably the better analogy at this point. Like Blackberry, you write down to the co-CEO. I do love the fact that they've named co-CEOs. I go, that is awesome. Please print business cards with co-CEO on it.
• 1:36:47
  The dumbest title ever. I mean, it's just like, it's not a good idea to have co-CEOs. That's what I have to say. It's like, you need to have a CEO, unless you're going to fire both of them at the same time when you, it's time for the next CEO. Oh, so actually on the Scott Guthrie thing, when I was telling you, I was like asking when they're going to fire bomber.
• 1:37:09
  The thing that I have embarrassed Intel employees by saying it's like, yeah, I'm looking forward to making the case for Tofino to Intel's next CEO. Like, what do you mean Intel's next CEO? I'm like, well, actually, so now I've said that so many times. I'm like, I'm not sure now actually it's like, actually now I'm actually not interested in doing that. I said I was going to make the case that does next year.
• 1:37:27
  I'm actually not sorry, you waited too long. I'm actually not going to make the case for me. No, we're just going to move on. Unfortunately. But yeah, I don't know.
• 1:37:37
  And maybe I, you know, certainly some strategic things are going to be off the table for like there are going to be some things that they're not going to be allowed to do in terms of folks they sell to. I mean, they can't, you know, they're not selling it to Huawei, right? So it's like, there's going to be some limits, but I think that there is great possibility here, but I agree. So what do you think that, so you think that just like that die is cast, the, they can't stop making the memories, even though, according to Ford,
• 1:38:10
  Adam Leventhal: I mean, I feel like that is a very, like pessimistic outlook, But I also, it's so hard to steer these kinds of massive tankers. And that's with someone who's like an eager captain at the helm. And I just don't see either, like, I don't see overcoming either of those obstacles. You've got a board that has proved itself to be risk-averse. You have, I think, in general, a market where Intel's not a sexy place for 1 of these kinds of change agents to operate.
• 1:38:43
  You have a bunch of people who are comfortable making a bunch of money, maybe not wanting to make waves. And that's kind of the Intel way. And I just don't see a path out.
• 1:38:54
  Bryan Cantrill: Well, and also the thing like in terms of like spinning these things off and selling them, though, like who wants to buy these cultural problems? Yeah. Right. I mean, you have like the, and someone earlier in the chat had said this, I've got kind of Boeing vibes and I'm like, yeah, like post McDonald Douglas, 1 buys, like who wants to buy their McDonald Douglas by buying something from Intel. You know what I mean?
• 1:39:14
  And I, why is like, Why is Broadcom not just gonna let them go to business? I'm not interested in
• 1:39:21
  Adam Leventhal: that. I mean, maybe I was thinking actually Broadcom because Hocktan has such a, I don't think he would allow McDonald Douglas to survive in his Boeing. Like I think he would have the paper mache head of Gelsinger as a warning to anyone else who wanted to speak at a turn.
• 1:39:41
  Bryan Cantrill: I think we're gonna have to get Ian up on stage because Ian does have a suggestion for a CEO. Okay, so who's your thought on a CEO candidate?
• 1:39:51
  Ian Grunnert: I cannot pronounce his last name, but Johnny Sruji from Apple is the Senior Vice President of Fargway Technologies at Apple and is a reasonable candidate because of the size of the organization that he's currently in charge of as well as being at Apple through the Intel to Apple Silicon kind of transition. I feel like there's just not a lot of candidates out there that have the background that Intel requires as well as the kind of remit over a large enough organization to understand the difficulties and kind of aligning people. So yeah, I mean, I think that would be the candidate that I'd look to.
• 1:40:44
  Bryan Cantrill: I think
• 1:40:44
  Ian Grunnert: they were also considered in 2019 during that search. So seems like a reasonable kind of bet at this point.
• 1:40:54
  Bryan Cantrill: I like it. I don't know if he's going to go for it because I think it's going to be, you're going to candidates you've passed over in 2019. They're going to be like, I don't know. Like the, okay, so the other 1 that was interesting was apparently, so Sanjay Gia was on there. That's interesting.
• 1:41:13
  And then Renee James, which that would be fascinating because she, I was at Ampere, right? I think she's still the CEO of Ampere. So that's gonna be kind of a wild 1 for, but that's another 1 that might be. So that's interesting. And I think, so the thing I like about that is that the You know you because you kind of need someone who's like I'm a little bored You know, I'm that for I like yes, I got the Apple thing going.
• 1:41:48
  Apple is like a crushed it. You know, I've already got like you find like I've made plenty of money. And but now I this would be interesting. I could do go do something really fascinating. There are yeah, there is absolutely no way the, the, JT, who is the, the CEO of Arista JT is great.
• 1:42:13
  Have you ever seen her speak? Adam? No, no, really compelling. Yeah, she'd be great. I would be stunned if she if you can get her.
• 1:42:21
  Sure. I think she's going to be off the table. I think she's not going to have 0 interest. I think she will laugh you out of the room, which I think is going to be true for a lot of the folks you'd want. But.
• 1:42:31
  Well, I have to come back here when they are. So how about Sam Altman? Like this, I want to look like, give me some crazy ideas to people. I want some crazy ideas. Like why not Sam Altman?
• 1:42:42
  Adam Leventhal: Why not?
• 1:42:43
  Bryan Cantrill: Oh, no, no, no. I actually, that's a rhetorical question. There we go. But I think that, look, look, the last, the last person that I, that I wanted had to delete all of his tweets because he became, you know, he wanted to
• 1:42:59
  Adam Leventhal: read no bad ideas. There are no bad ideas.
• 1:43:02
  Ian Grunnert: Well, I feel like the likelihood of George Hotz lasting more than 3 months before getting frustrated with the realities of dealing with such a large organization and attempting to shift, like a fair change in a large organization. I Don't think it's a great
• 1:43:23
  Bryan Cantrill: I don't think it's very fit I think and then I appreciate your kid your candor here in terms of your feedback and I agree with you No, I know I he's obviously terrible a terrible fit, but Yeah, I think it's gonna be it's gonna be tough but I think it's potentially Interesting is what I gotta say Potentially interesting. I think there are some possibilities. But it's gonna and I think that but the board is also going to need to change its ways. It's what I have to say. So, you know, I don't know if you have any other thoughts in any other, any other, other crazy folks out there,
• 1:43:59
  Ian Grunnert: Howard Schultz, former CEO of Starbucks.
• 1:44:05
  Bryan Cantrill: I like this guy. Howard Schultz.
• 1:44:07
  Adam Leventhal: Now you're getting crazy.
• 1:44:08
  Bryan Cantrill: Yeah. Now you're getting, this is what we do. Okay. Now the juices are flowing. Now we got OK, now now we can start.
• 1:44:13
  Howard Schultz would be great. That's I mean, no, I mean, obviously terrible, but like, getting like, what about, okay, but this is an interesting category of like, no, you get someone from a totally different industry. I think that that would be like, the John Scully theory.
• 1:44:31
  Adam Leventhal: Is that the idea?
• 1:44:33
  Bryan Cantrill: The John Scully theory, or I was going to say, you know, I was, I, I had, I was in Japan and I was talking to a very nice elderly woman who was, and we were talking and she's from Minnesota. And she said, Oh, it's very exciting in Minnesota because there's a prize fighter who is running to be the governor. And I become out of the news. I'm like, there's a prize fighter. It's like, his name is Jesse Ventura.
• 1:44:57
  And I'm like, ma'am, that is Jesse the body Ventura. And that is not a prize fighter. And so there's a, I got an idea like, Oh, let's have Hulk Hogan run Intel. I mean, like, okay. Like, you know,
• 1:45:09
  Adam Leventhal: I feel like the next administration has like a hammer lock on all the true crazy.
• 1:45:14
  Bryan Cantrill: The next administration. I do think you're going to be fighting for? I, yes, I think like, no, I, sorry, Hulk Hogan is already our secretary of commerce, though. I'm not sure. Like I'm not sure he's kind of spoken for.
• 1:45:25
  Yeah. I mean, I feel like kind of bonkers spin the wheel folks are, are all going to be. But I think howard schultz is kind of interesting getting someone who's got I don't think I I mean I don't think it's gonna work pierre lamond song's creative pierre would but no pierre's not gonna do it although that would be very entertaining
• 1:45:48
  Adam Leventhal: oh yeah
• 1:45:49
  Bryan Cantrill: that'd be fun be fun pierre would also spit out a strike if you're saying
• 1:45:55
  Ian Grunnert: a, Intel acquired by Oracle and Larry Ellison getting vertical integration with Oracle cloud. It's another,
• 1:46:05
  Bryan Cantrill: scary thought. That's getting a little chilling. I, yeah, that would be, I, that would be interesting if a regulator would... I don't think a regulator is going to allow Oracle to buy all of Intel. But yeah, maybe a piece of it.
• 1:46:25
  I think that IBM, someone's suggesting IBM buying Intel, that also seems a little far-fetched, but who knows?
• 1:46:36
  Adam Leventhal: Are we remembering who's regulating things like coming up pretty soon?
• 1:46:40
  Bryan Cantrill: I you know what this is true This is it is it so you know what? Congratulations on your new acquisition Mary Ellison Please let us know when we are allowed to have a CPU. It's like, okay, the microcode update has a license manager now. It's like I'm getting a license on it on my microcode. Too real.
• 1:46:59
  All right. Well, I think we're going to need to do an update a little bit when when I mean the co-CEOs crew are not going to last so we're going to need to do an update at some point. Yeah. Speaking of updates Adam as we kind of close out, we've done something truly radical for this podcast, something that is unprecedented.
• 1:47:16
  Adam Leventhal: That's right. Put down your drinks, everyone. Sit down,
• 1:47:19
  Bryan Cantrill: put down your drinks. And yeah, you may exactly be have all sharp objects away from you because this is really kind of shocking. We actually have, we've actually done 2 things unprecedented back to back. We have scheduled not just 1 episode in advance, but 2 episodes in advance. And we have dtrace.conf coming up in a week and a half.
• 1:47:40
  That's going to be that we have gone all online with that conference. Really excited about that. So that's going to allow everyone to attend for free and in part to get ready for detrace.com Next week we're going to talk about conferences in tech, which I'm really excited about and we're going to be joined by Stephen O'Grady and Theo Schwarz-Nichol.
• 1:48:00
  Adam Leventhal: Yeah, those guys are really looking forward to that
• 1:48:02
  Bryan Cantrill: which is gonna be great and But wait, there's more Okay, and I know this is like overwhelming we have planned the next week And I know that this is is this is too much Have we have we have we overdone it is this the because you know, we have planned 2 episodes in advance, you know, something earth shattering is gonna happen between right now. But Paul Frazee, who is the CTO of Blue Sky, is gonna be joining us on scaling blue sky. And I am stoked for that conversation.
• 1:48:36
  Adam Leventhal: Yeah,
• 1:48:36
  Bryan Cantrill: that's going to be, I think it's going to be great. I think Ben, we've been really excited to see all the progress on blue sky. And I mean, Paul's been obviously at the actual absolute epicenter of it all. And really excited that he's going to join us. It's going to be great.
• 1:48:53
  So that's going to be a great conversation. So we've got 2 amazing conversations coming up the next 2 weeks, and then we're going to close it. Actually, Adam, We actually know, we actually know 3 episodes in advance. Yeah, exactly. Because we know what we're going to do for the 23rd.
• 1:49:08
  We're going to have our, our year end closeout.
• 1:49:12
  Adam Leventhal: Well, and we know 4 episodes in advance. We know, we know the first episode of the year is going to be our predictions episode. So I 23rd, bring your, your fondest memories of the year to share with us. You know, on Oxide and friends and then predictions in, in January.
• 1:49:31
  Bryan Cantrill: At this point, Oxide and Friends has got a more established roadmap than Intel. We know the future of Oxide and Friends more than we know the future of Intel. This is amazing. So definitely bring some of your, be thinking of your kind of favorite moments from the past year We can that we can talk about at our year-end episode This is an idea we got from the changelog folks about kind of doing our the year-end wrap-up, which I'm really excited about And then we'll And then it's predictions. Yes, very exciting stuff.
• 1:50:06
  And then we're gonna have another great January for you. But this is, it's exciting. Yeah. Well, and it all starts, thank you, Pat Gelsinger for giving us the gift of content.
• 1:50:19
  Adam Leventhal: That's true.
• 1:50:21
  Bryan Cantrill: And, really, I think passing conversation, Adam, we're going to have to see how it works out. I'll be, and when, when Howard Schultz is named to the CEO of Intel.
• 1:50:31
  Adam Leventhal: Yeah. You heard it here first.
• 1:50:33
  Bryan Cantrill: You heard it here first with Hulk Hogan hired by the Foundry, by an IFS. All right. Thanks everyone. Talk to you next time.