IoT Coffee Talk: Episode 262 - Sharding Your AI (A COMPUTEX 2025 Special)

Oh yeah. Oh yeah. Oh yeah. IoT coffee Oh yeah. Oh yeah. Oh yeah. IoT coffee talk. talk. talk. Yes. Welcome. Yes. Rob is driving. I'm Yes. Welcome. Yes. Rob is driving. I'm Yes. Welcome. Yes. Rob is driving. I'm not driving. I promise I'm really not not driving. I promise I'm really not not driving. I promise I'm really not driving as far as you know. Virtual driving as far as you know. Virtual driving as far as you know. Virtual background. Are you in the omniverse? background. Are you in the omniverse? background. Are you in the omniverse? I'm in the omniverse. In the welcome to I'm in the omniverse. In the welcome to I'm in the omniverse. In the welcome to my underground layer. my underground layer. my underground layer. Yeah. Yeah. And I assume then that is a Yeah. Yeah. And I assume then that is a Yeah. Yeah. And I assume then that is a uh synthetic data that's running around uh synthetic data that's running around uh synthetic data that's running around in the background. It's all Yeah. It's in the background. It's all Yeah. It's in the background. It's all Yeah. It's like a video. It makes it kind of like like a video. It makes it kind of like like a video. It makes it kind of like when you watch movies and you see cars when you watch movies and you see cars when you watch movies and you see cars in the back, you know, the back window in the back, you know, the back window in the back, you know, the back window and Yeah. It's like that. It's like and Yeah. It's like that. It's like and Yeah. It's like that. It's like video. Yeah. It's a creepy movement. video. Yeah. It's a creepy movement. video. Yeah. It's a creepy movement. Yeah. Yeah. So wonderful. Yeah. Yeah. So wonderful. Yeah. Yeah. So wonderful. Did you just fly back? Yeah. Just Did you just fly back? Yeah. Just Did you just fly back? Yeah. Just like home, dude. From Taiwan. Oh my god. like home, dude. From Taiwan. Oh my god. like home, dude. From Taiwan. Oh my god. How was it? Oh, it was um it was well How was it? Oh, it was um it was well How was it? Oh, it was um it was well worth the investment. Um, I mean, I kind worth the investment. Um, I mean, I kind worth the investment. Um, I mean, I kind of I decided to go a little bit last of I decided to go a little bit last of I decided to go a little bit last minute, right? It was about a month, minute, right? It was about a month, minute, right? It was about a month, right? I was thinking, right? I was thinking, right? I was thinking, man, this has got to be an interesting man, this has got to be an interesting man, this has got to be an interesting year to go to Comp again because Comp year to go to Comp again because Comp year to go to Comp again because Comp Computex last year was a really weird Computex last year was a really weird Computex last year was a really weird one for the conference because, you one for the conference because, you one for the conference because, you know, a lot of the um analysts who go know, a lot of the um analysts who go know, a lot of the um analysts who go there all the time will tell you it was there all the time will tell you it was there all the time will tell you it was primarily a PC

primarily a PC primarily a PC um event. and Brady Wong uh who is with um event. and Brady Wong uh who is with um event. and Brady Wong uh who is with Counterpoint Research. He was a good Counterpoint Research. He was a good Counterpoint Research. He was a good buddy of mine, the Gartner days. He's buddy of mine, the Gartner days. He's buddy of mine, the Gartner days. He's over there now. He's a senior director over there now. He's a senior director over there now. He's a senior director uh covering the semiconductor industry uh covering the semiconductor industry uh covering the semiconductor industry in Taiwan. He lives there. Was telling in Taiwan. He lives there. Was telling in Taiwan. He lives there. Was telling me is he used to not go there because it me is he used to not go there because it me is he used to not go there because it wasn't worth going to, you know. Did Ray wasn't worth going to, you know. Did Ray wasn't worth going to, you know. Did Ray leave his own company? No. Brady. Brady. Brady Wong. Okay, No. Brady. Brady. Brady Wong. Okay, gotcha. Yeah, it freaked me out there gotcha. Yeah, it freaked me out there gotcha. Yeah, it freaked me out there for a second. Constellation. I said for a second. Constellation. I said for a second. Constellation. I said counterpoint. Not counterpoint. Not counterpoint. Not said Ray Wong and I was like, "Oh no, he said Ray Wong and I was like, "Oh no, he said Ray Wong and I was like, "Oh no, he left left left Constellation." Oh no, I'm not going to Constellation." Oh no, I'm not going to Constellation." Oh no, I'm not going to see him on Fox see him on Fox see him on Fox Business or Business or Business or mostly on Fox. What's his name? Uh from Futura. Fox. What's his name? Uh from Futura. We're both on Fox Business in the We're both on Fox Business in the We're both on Fox Business in the mornings a lot. Yeah. Daniel. Daniel.

mornings a lot. Yeah. Daniel. Daniel. mornings a lot. Yeah. Daniel. Daniel. Daniel, he's on Fox Business a lot, you Daniel, he's on Fox Business a lot, you Daniel, he's on Fox Business a lot, you know. So, just getting getting the going know. So, just getting getting the going know. So, just getting getting the going with the market in the morning, right? with the market in the morning, right? with the market in the morning, right? Wall Street. Yeah. Wall Street. Yeah. Wall Street. Yeah. So, okay. So, last year was mostly about So, okay. So, last year was mostly about So, okay. So, last year was mostly about PC. Was it like was last year like the PC. Was it like was last year like the PC. Was it like was last year like the AIPC thing? Yeah. And that's what drew a AIPC thing? Yeah. And that's what drew a AIPC thing? Yeah. And that's what drew a lot of the big um semiconductor lot of the big um semiconductor lot of the big um semiconductor companies like Qualcomm, Intel. So last companies like Qualcomm, Intel. So last companies like Qualcomm, Intel. So last year I was for Intel's tech tour and um year I was for Intel's tech tour and um year I was for Intel's tech tour and um you know they flew a bunch of us out to you know they flew a bunch of us out to you know they flew a bunch of us out to hang out for the week before Computex hang out for the week before Computex hang out for the week before Computex and then they they and then they they and then they they um I don't know how many times they they um I don't know how many times they they um I don't know how many times they they gave us a lot of technical details like gave us a lot of technical details like gave us a lot of technical details like the dirty details of Lunar Lake and um the dirty details of Lunar Lake and um the dirty details of Lunar Lake and um as well as Grand Rapids and Um, and um, as well as Grand Rapids and Um, and um, as well as Grand Rapids and Um, and um, oh she g jeez. Uh, Sierra oh she g jeez. Uh, Sierra oh she g jeez. Uh, Sierra Forest I don't know memorize all these Forest I don't know memorize all these Forest I don't know memorize all these things, dude.

things, dude. things, dude. Does it feel like all that stuff's going Does it feel like all that stuff's going Does it feel like all that stuff's going to make a difference? to make a difference? to make a difference? Um, yeah. You know, I Um, yeah. You know, I Um, yeah. You know, I think I mean, yeah, it it makes a think I mean, yeah, it it makes a think I mean, yeah, it it makes a difference for Intel because if they difference for Intel because if they difference for Intel because if they don't if they don't continue advancing don't if they don't continue advancing don't if they don't continue advancing forward with their products, then forward with their products, then forward with their products, then they're I mean, AMD is going to bulldoze they're I mean, AMD is going to bulldoze they're I mean, AMD is going to bulldoze them over, you know? So, are one of them over, you know? So, are one of them over, you know? So, are one of these lakes going to be built with a new these lakes going to be built with a new these lakes going to be built with a new 2 nanometer whatever or 3 2 nanometer whatever or 3 2 nanometer whatever or 3 nanometer 18A process? Oh, the 18A. nanometer 18A process? Oh, the 18A. nanometer 18A process? Oh, the 18A. Yeah, Lunar Lake is uh 18, not Lunar Yeah, Lunar Lake is uh 18, not Lunar Yeah, Lunar Lake is uh 18, not Lunar Lake. Um, Panther Lake, the thing that's Lake. Um, Panther Lake, the thing that's Lake. Um, Panther Lake, the thing that's coming out later this year, um, probably coming out later this year, um, probably coming out later this year, um, probably toward Christmas time is 188. So, that's toward Christmas time is 188. So, that's toward Christmas time is 188. So, that's good news. Hey, you know, it is good good news. Hey, you know, it is good good news. Hey, you know, it is good news cuz, you know, Intel's been too news cuz, you know, Intel's been too news cuz, you know, Intel's been too incremental every year, you know. Yeah. incremental every year, you know. Yeah. incremental every year, you know. Yeah. They need a le they need a giant leap They need a le they need a giant leap They need a le they need a giant leap forward that really because you know forward that really because you know forward that really because you know year after year you look at new laptops year after year you look at new laptops year after year you look at new laptops and PCs. You really can't tell the and PCs. You really can't tell the and PCs. You really can't tell the difference in performance.

difference in performance. difference in performance. Oh, and guess what? You're really not Oh, and guess what? You're really not Oh, and guess what? You're really not going to be able to tell the difference going to be able to tell the difference going to be able to tell the difference in performance going forward. in performance going forward. in performance going forward. Great job, Great job, Great job, Rob. Um, you know, no, seriously, dude. Rob. Um, you know, no, seriously, dude. Rob. Um, you know, no, seriously, dude. remember. Okay, so before we go any remember. Okay, so before we go any remember. Okay, so before we go any further, hey, uh remember to take us further, hey, uh remember to take us further, hey, uh remember to take us seriously at your own risk. Kind of seriously at your own risk. Kind of seriously at your own risk. Kind of like, you know, being in the car with like, you know, being in the car with like, you know, being in the car with Rob while he's driving. That's very Rob while he's driving. That's very Rob while he's driving. That's very risky. You just don't want to do Huh? risky. You just don't want to do Huh? risky. You just don't want to do Huh? It's not happening. It's a video. Yeah. It's not happening. It's a video. Yeah. It's not happening. It's a video. Yeah. I mean Yeah. So that that's I mean Yeah. So that that's I mean Yeah. So that that's hypothetical. Hypothetically, if you hypothetical. Hypothetically, if you hypothetical. Hypothetically, if you were in the car with Rob driving in real were in the car with Rob driving in real were in the car with Rob driving in real life instead of the simulation he's life instead of the simulation he's life instead of the simulation he's driving in right now in the Nvidia driving in right now in the Nvidia driving in right now in the Nvidia Omniverse. Yes, that's where I am. It Omniverse. Yes, that's where I am. It Omniverse. Yes, that's where I am. It would Yeah, it it would just be one of would Yeah, it it would just be one of would Yeah, it it would just be one of those risky things. So, yeah. Um, and if those risky things. So, yeah. Um, and if those risky things. So, yeah. Um, and if you do any kind of weird stuff, um, that you do any kind of weird stuff, um, that you do any kind of weird stuff, um, that weirdness is entirely attributed to your weirdness is entirely attributed to your weirdness is entirely attributed to your own decision making and yeah, should not own decision making and yeah, should not own decision making and yeah, should not be attributed to anything that we say.

be attributed to anything that we say. be attributed to anything that we say. Uh, and uh, Uh, and uh, Uh, and uh, remember if you're not into charity, but remember if you're not into charity, but remember if you're not into charity, but you're into fashion, buy one of these you're into fashion, buy one of these you're into fashion, buy one of these shirts, as well as a elevate our kids shirts, as well as a elevate our kids shirts, as well as a elevate our kids um, t-shirt uh, to help bridge the um, t-shirt uh, to help bridge the um, t-shirt uh, to help bridge the digital divide for kids K through 12 in digital divide for kids K through 12 in digital divide for kids K through 12 in underserved and unserved communities. underserved and unserved communities. underserved and unserved communities. And so that is the charity Elevator Kids And so that is the charity Elevator Kids And so that is the charity Elevator Kids at www.elevatorids.org. at www.elevatorids.org. at www.elevatorids.org. So consider donating and uh making So consider donating and uh making So consider donating and uh making everyone feel good. So anyway, that everyone feel good. So anyway, that everyone feel good. So anyway, that change what? Take that look at the man change what? Take that look at the man change what? Take that look at the man in the mirror and make that change. in the mirror and make that change. in the mirror and make that change. Yeah, Yeah, Yeah, maybe that's what we'll call this one. maybe that's what we'll call this one. maybe that's what we'll call this one. The man man in the mirror in the omniverse was Jensen's keynote. omniverse was Jensen's keynote. Um, actually it was pretty Um, actually it was pretty Um, actually it was pretty interesting. You know, dude, I I don't interesting. You know, dude, I I don't interesting. You know, dude, I I don't know how many times I seen him. I mean, know how many times I seen him. I mean, know how many times I seen him. I mean, literally like in the same freaking literally like in the same freaking literally like in the same freaking room. Um, yeah, got to be at least every room. Um, yeah, got to be at least every room. Um, yeah, got to be at least every other week in the past four months, you other week in the past four months, you other week in the past four months, you know, and so for sure I am very know, and so for sure I am very know, and so for sure I am very wellversed in all of his talking points.

wellversed in all of his talking points. wellversed in all of his talking points. So, um yeah, there were some new things. So, um yeah, there were some new things. So, um yeah, there were some new things. Uh I I have to remember what they are Uh I I have to remember what they are Uh I I have to remember what they are because there was so many things because there was so many things because there was so many things um that were that I picked up on. There um that were that I picked up on. There um that were that I picked up on. There weren't actually, this is the weird weren't actually, this is the weird weren't actually, this is the weird thing, Rob, there weren't a lot of thing, Rob, there weren't a lot of thing, Rob, there weren't a lot of announcements. And so, going in there announcements. And so, going in there announcements. And so, going in there were a lot of people saying, "Well, I were a lot of people saying, "Well, I were a lot of people saying, "Well, I don't I'm not going to go to Compy don't I'm not going to go to Compy don't I'm not going to go to Compy checkex. Um there's nothing going on. No checkex. Um there's nothing going on. No checkex. Um there's nothing going on. No new announcements." But um I think more new announcements." But um I think more new announcements." But um I think more than the announcements was actually what than the announcements was actually what than the announcements was actually what they didn't they didn't they didn't announce what wasn't happening was just announce what wasn't happening was just announce what wasn't happening was just as interesting as what was happening. as interesting as what was happening. as interesting as what was happening. You know what I'm saying? Yeah. Yeah. You know what I'm saying? Yeah. Yeah. You know what I'm saying? Yeah. Yeah. you know, you know, you know, obviously the that little thing that you obviously the that little thing that you obviously the that little thing that you love so much, the DJX Spark, you know, love so much, the DJX Spark, you know, love so much, the DJX Spark, you know, little station little station little station u that's about the size of a nook. u that's about the size of a nook. u that's about the size of a nook. That kind that took center stage. I I That kind that took center stage. I I That kind that took center stage. I I think um Jensen has one basically in his think um Jensen has one basically in his think um Jensen has one basically in his pocket as he's wandering around. I love pocket as he's wandering around. I love pocket as he's wandering around. I love it. Yeah. And so yeah, oddly AI it. Yeah. And so yeah, oddly AI it. Yeah. And so yeah, oddly AI workstations really big thing this year.

workstations really big thing this year. workstations really big thing this year. That's cool. Yeah. Wow. Did you have a That's cool. Yeah. Wow. Did you have a That's cool. Yeah. Wow. Did you have a fun time in Taipei? fun time in Taipei? fun time in Taipei? Yeah. So, Yeah. So, Yeah. So, um because you know I went on my own um because you know I went on my own um because you know I went on my own dime, dime, dime, um you um you um you know, some of these companies. Um I know, some of these companies. Um I know, some of these companies. Um I ended up exploring the city a lot more ended up exploring the city a lot more ended up exploring the city a lot more and having to use public transportation. and having to use public transportation. and having to use public transportation. Well, you know, it was more of an Well, you know, it was more of an Well, you know, it was more of an optional thing, but you know, the hotel optional thing, but you know, the hotel optional thing, but you know, the hotel I was staying at was so close to the I was staying at was so close to the I was staying at was so close to the subway line. I I started taking the subway line. I I started taking the subway line. I I started taking the subway and dude, it is it is awesome, subway and dude, it is it is awesome, subway and dude, it is it is awesome, you know? It's so you know? It's so you know? It's so But get this, man. A ticket cost like 60 But get this, man. A ticket cost like 60 But get this, man. A ticket cost like 60 cents to a a dollar. It's like Oh, wow. cents to a a dollar. It's like Oh, wow. cents to a a dollar. It's like Oh, wow. Yeah. You know, Yeah. You know, Yeah. You know, Yeah. And then like an all day all day Yeah. And then like an all day all day Yeah. And then like an all day all day pass is something like two I think it's pass is something like two I think it's pass is something like two I think it's like like like 250 250 250 somethingly cheap. That's great. I mean somethingly cheap. That's great. I mean somethingly cheap. That's great. I mean you could actually live and thrive in a you could actually live and thrive in a you could actually live and thrive in a city like that with that kind of public city like that with that kind of public city like that with that kind of public transportation.

transportation. transportation. Exactly. Exactly. Without owning a car. Exactly. Exactly. Without owning a car. Exactly. Exactly. Without owning a car. Without owning a car. Perfectly cool. Without owning a car. Perfectly cool. Without owning a car. Perfectly cool. You get you know you get a little bit of You get you know you get a little bit of You get you know you get a little bit of exercise exercise exercise into your life. into your life. into your life. Go exercise. I mean, like, that's a bad thing, right? I mean, like, that's a bad thing, right? You probably need some of that. Yeah, I You probably need some of that. Yeah, I You probably need some of that. Yeah, I need some of that. I mean, I was like I need some of that. I mean, I was like I need some of that. I mean, I was like I landed like freaking bloated and not landed like freaking bloated and not landed like freaking bloated and not feeling very healthy. But after feeling very healthy. But after feeling very healthy. But after um you know, steaming in their really um you know, steaming in their really um you know, steaming in their really humid, it was really humid there, man. humid, it was really humid there, man. humid, it was really humid there, man. That's gross. But That's gross. But That's gross. But yeah, and and then, you know, walking yeah, and and then, you know, walking yeah, and and then, you know, walking around a crap ton. Yeah, I think I lost around a crap ton. Yeah, I think I lost around a crap ton. Yeah, I think I lost a little bit of weight. So, great, man. a little bit of weight. So, great, man. a little bit of weight. So, great, man. That's awesome. Exercise. So, what it's That's awesome. Exercise. So, what it's That's awesome. Exercise. So, what it's all about. Yeah, that's outstanding. So, all about. Yeah, that's outstanding. So, all about. Yeah, that's outstanding. So, since it was on your own dime, I take it since it was on your own dime, I take it since it was on your own dime, I take it you were not lying flat in business you were not lying flat in business you were not lying flat in business class.

class. class. No, No, No, no, no. no, no. no, no. Such a tragedy. I was in the middle. Such a tragedy. I was in the middle. Such a tragedy. I was in the middle. Oh, hey, you don't get it. Don't Don't Oh, hey, you don't get it. Don't Don't Oh, hey, you don't get it. Don't Don't get it twisted, man. It's not always the get it twisted, man. It's not always the get it twisted, man. It's not always the case. They fly you out business class. case. They fly you out business class. case. They fly you out business class. They sometimes put you in cattle class. They sometimes put you in cattle class. They sometimes put you in cattle class. So, yeah, that's true. That's a good So, yeah, that's true. That's a good So, yeah, that's true. That's a good point. Wow. Holy moly. I'm trying to point. Wow. Holy moly. I'm trying to point. Wow. Holy moly. I'm trying to think what was the big news this week. think what was the big news this week. think what was the big news this week. Well, we saw Well, we saw Well, we saw the Johnny the Johnny the Johnny IV video IV video IV video open Sam Alman. They're like sitting in open Sam Alman. They're like sitting in open Sam Alman. They're like sitting in a bar or something and it was so a bar or something and it was so a bar or something and it was so contrived and contrived and contrived and very well produced. very well produced. very well produced. [Laughter] [Laughter] [Laughter] So, is it possible that it's a total So, is it possible that it's a total So, is it possible that it's a total scam? I mean, it's like Open AI is even scam? I mean, it's like Open AI is even scam? I mean, it's like Open AI is even though they're technically valued at 300 though they're technically valued at 300 though they're technically valued at 300 billion, they don't actually have any billion, they don't actually have any billion, they don't actually have any money. They don't make money. They all money. They don't make money. They all money. They don't make money. They all they do is lose money. And so, they're they do is lose money. And so, they're they do is lose money. And so, they're doing doing doing They bought what? Wind surf or whatever They bought what? Wind surf or whatever They bought what? Wind surf or whatever the coding thing in all stock deal and the coding thing in all stock deal and the coding thing in all stock deal and then they buy Johnny I thing all stock then they buy Johnny I thing all stock then they buy Johnny I thing all stock and Johnny doesn't even have a product.

and Johnny doesn't even have a product. and Johnny doesn't even have a product. Exactly. And some smart people which is Exactly. And some smart people which is Exactly. And some smart people which is great hires um way back in the day but no one hires um way back in the day but no one would ever do an aqua hire for like $6 would ever do an aqua hire for like $6 would ever do an aqua hire for like $6 billion. billion. billion. I know. I mean, that's insane. Is is a I know. I mean, that's insane. Is is a I know. I mean, that's insane. Is is a genius, right? I mean, wow. Yes. I genius, right? I mean, wow. Yes. I genius, right? I mean, wow. Yes. I wonder I wonder if Sam's getting some wonder I wonder if Sam's getting some wonder I wonder if Sam's getting some kind of kickback on the I don't know. I kind of kickback on the I don't know. I kind of kickback on the I don't know. I don't It's I don't know. It's like It don't It's I don't know. It's like It don't It's I don't know. It's like It just seems crazy. Yeah, Johnny is just seems crazy. Yeah, Johnny is just seems crazy. Yeah, Johnny is awesome. You want to have him, you get awesome. You want to have him, you get awesome. You want to have him, you get him and whatever 20, 30, 40, 50 other him and whatever 20, 30, 40, 50 other him and whatever 20, 30, 40, 50 other people in his design studio and you paid people in his design studio and you paid people in his design studio and you paid insane money for it. insane money for it. insane money for it. Um, and it's not real money. It's all Um, and it's not real money. It's all Um, and it's not real money. It's all stock. And who knows what's going to stock. And who knows what's going to stock. And who knows what's going to happen, you know. Gosh. Yeah. I mean, happen, you know. Gosh. Yeah. I mean, happen, you know. Gosh. Yeah. I mean, you know, it's just remarkable that you know, it's just remarkable that you know, it's just remarkable that somehow companies that generate no somehow companies that generate no somehow companies that generate no profit, but just the Yeah. have a hockey profit, but just the Yeah. have a hockey profit, but just the Yeah. have a hockey stickish.

stickish. stickish. Yes. Depends on which hockey stick Yes. Depends on which hockey stick Yes. Depends on which hockey stick you're talking about. Um, what do you you're talking about. Um, what do you you're talking about. Um, what do you call it? A monthly active user growth call it? A monthly active user growth call it? A monthly active user growth number. Yeah. Or is it weekly or is it number. Yeah. Or is it weekly or is it number. Yeah. Or is it weekly or is it daily? You know, uh, and by the way, daily? You know, uh, and by the way, daily? You know, uh, and by the way, people are really playing that game. Um, people are really playing that game. Um, people are really playing that game. Um, they're playing pulling that card in all they're playing pulling that card in all they're playing pulling that card in all kinds of directions, man. Just up the kinds of directions, man. Just up the kinds of directions, man. Just up the numbers. And numbers. And numbers. And that was the thing. That was always that was the thing. That was always that was the thing. That was always remember the dot times. Yeah. And they remember the dot times. Yeah. And they remember the dot times. Yeah. And they just said, "Make sure that you don't just said, "Make sure that you don't just said, "Make sure that you don't make any revenue because then Wall make any revenue because then Wall make any revenue because then Wall Street will be able to price you and Street will be able to price you and Street will be able to price you and it's going to destroy your your it's going to destroy your your it's going to destroy your your valuation. So, as long as you're valuation. So, as long as you're valuation. So, as long as you're pre-revenue, pre-revenue, pre-revenue, you don't make me think it's mysterious you don't make me think it's mysterious you don't make me think it's mysterious and you're the hot new thing and you can and you're the hot new thing and you can and you're the hot new thing and you can have a high, you know, and they even have a high, you know, and they even have a high, you know, and they even made fun of it like one of the in made fun of it like one of the in made fun of it like one of the in probably the first episode of the probably the first episode of the probably the first episode of the Silicon Valley TV show. They're like, Silicon Valley TV show. They're like, Silicon Valley TV show. They're like, "Don't make money. Whatever you do, "Don't make money. Whatever you do, "Don't make money. Whatever you do, don't make any money." don't make any money." don't make any money." Yeah. Keep keep um But then, you know, Yeah. Keep keep um But then, you know, Yeah. Keep keep um But then, you know, there's revenue and then profit. Most of there's revenue and then profit. Most of there's revenue and then profit. Most of these guys don't make any revenue and these guys don't make any revenue and these guys don't make any revenue and they're nowhere near profitable. they're they're nowhere near profitable. they're they're nowhere near profitable. they're um racking up um racking up um racking up massive monthly opex and capex and so I massive monthly opex and capex and so I massive monthly opex and capex and so I don't know how all you know I mean dude don't know how all you know I mean dude don't know how all you know I mean dude it was like when I was at EY back in the it was like when I was at EY back in the it was like when I was at EY back in the dot days there was a partner that was dot days there was a partner that was dot days there was a partner that was going around doing a road show and his going around doing a road show and his going around doing a road show and his whole thing was um valuation is dead

whole thing was um valuation is dead whole thing was um valuation is dead there's only internet internet valuation there's only internet internet valuation there's only internet internet valuation I'm like what the hell are you talking I'm like what the hell are you talking I'm like what the hell are you talking about valuation is valuation. Right. about valuation is valuation. Right. about valuation is valuation. Right. Right. You're absolutely right, man. Right. You're absolutely right, man. Right. You're absolutely right, man. They come up with some kind of weird new They come up with some kind of weird new They come up with some kind of weird new science and Yes. science and Yes. science and Yes. about valuation, due diligence, what about valuation, due diligence, what about valuation, due diligence, what matters, the network effect, all this matters, the network effect, all this matters, the network effect, all this like nonsense. like nonsense. like nonsense. And a lot of it is to just create these And a lot of it is to just create these And a lot of it is to just create these massive numbers, you know, cuz that's massive numbers, you know, cuz that's massive numbers, you know, cuz that's I'm getting I'm getting out of my I'm getting I'm getting out of my I'm getting I'm getting out of my virtual car. Oh, okay. All right. And virtual car. Oh, okay. All right. And virtual car. Oh, okay. All right. And then you're in your virtual home. then you're in your virtual home. then you're in your virtual home. Something like that. Something like that. Something like that. Well, you sound a hell of a lot better. Well, you sound a hell of a lot better. Well, you sound a hell of a lot better. We made it virtually. We made it virtually. We made it virtually. So, you had like managers and leaders at So, you had like managers and leaders at So, you had like managers and leaders at EY back then. Just saying that kind of EY back then. Just saying that kind of EY back then. Just saying that kind of BS. Oh, yeah. Yeah. Yeah. They thought BS. Oh, yeah. Yeah. Yeah. They thought BS. Oh, yeah. Yeah. Yeah. They thought it made them sound smart maybe or Well, it made them sound smart maybe or Well, it made them sound smart maybe or Well, I mean it made them sound like I mean it made them sound like I mean it made them sound like um innovators or some revolutionary um innovators or some revolutionary um innovators or some revolutionary thought leader type. Oh yeah. All it thought leader type. Oh yeah. All it thought leader type. Oh yeah. All it ended up Yeah. I mean it's it's a ended up Yeah. I mean it's it's a ended up Yeah. I mean it's it's a formula for snake oil salesmanship and formula for snake oil salesmanship and formula for snake oil salesmanship and which has kind of which has kind of which has kind of dominated what dominated what dominated what we years now since then, right? Because we years now since then, right? Because we years now since then, right? Because Silicon Valley wasn't really like that Silicon Valley wasn't really like that Silicon Valley wasn't really like that because if you wanted to get an Ibanking because if you wanted to get an Ibanking because if you wanted to get an Ibanking job back then, you had to go to you had

job back then, you had to go to you had job back then, you had to go to you had to pretty much go to Manhattan, right? to pretty much go to Manhattan, right? to pretty much go to Manhattan, right? You had to or London or Japan or Tokyo, You had to or London or Japan or Tokyo, You had to or London or Japan or Tokyo, right? Right. Those are like the three right? Right. Those are like the three right? Right. Those are like the three places that you get, you know, if you places that you get, you know, if you places that you get, you know, if you want to be big in Japan. Yeah. That's want to be big in Japan. Yeah. That's want to be big in Japan. Yeah. That's where the capital markets were. And then where the capital markets were. And then where the capital markets were. And then San San San Francisco, nothing. You know, they had Francisco, nothing. You know, they had Francisco, nothing. You know, they had MC McGomery Securities. And then they MC McGomery Securities. And then they MC McGomery Securities. And then they had had had um uh what do you call it? Oh jeez. um uh what do you call it? Oh jeez. um uh what do you call it? Oh jeez. Sansini uh you know the law firm law Sansini uh you know the law firm law Sansini uh you know the law firm law firms and uh and then the VCs came in, firms and uh and then the VCs came in, firms and uh and then the VCs came in, right? Well, actually the VCs didn't right? Well, actually the VCs didn't right? Well, actually the VCs didn't come in. They were incubated in the come in. They were incubated in the come in. They were incubated in the valley. there were like these private valley. there were like these private valley. there were like these private placement, private equity, placement, private equity, placement, private equity, equityish kind of um firms that came in equityish kind of um firms that came in equityish kind of um firms that came in and started in my opinion polluting uh and started in my opinion polluting uh and started in my opinion polluting uh that really cool innovators vibe that that really cool innovators vibe that that really cool innovators vibe that was in the Silicon Valley at the time, was in the Silicon Valley at the time, was in the Silicon Valley at the time, you know.

you know. you know. Yeah. And Silicon Graphics. Yeah. That Yeah. And Silicon Graphics. Yeah. That Yeah. And Silicon Graphics. Yeah. That that but that all that stuff was like that but that all that stuff was like that but that all that stuff was like real, right? I mean, Silicon Graphics real, right? I mean, Silicon Graphics real, right? I mean, Silicon Graphics Sun Microsystems, man. They couldn't Sun Microsystems, man. They couldn't Sun Microsystems, man. They couldn't just make crap up and say, "Oh, hey, just make crap up and say, "Oh, hey, just make crap up and say, "Oh, hey, look, you know, here's a company that look, you know, here's a company that look, you know, here's a company that makes absolutely nothing." And isn't it makes absolutely nothing." And isn't it makes absolutely nothing." And isn't it funny? I mean, they made hardware. They funny? I mean, they made hardware. They funny? I mean, they made hardware. They made real things that changed the world. made real things that changed the world. made real things that changed the world. And um whereas two guys sitting at a bar And um whereas two guys sitting at a bar And um whereas two guys sitting at a bar at a totally highly produced video at a totally highly produced video at a totally highly produced video yesterday where I'm going to buy your yesterday where I'm going to buy your yesterday where I'm going to buy your consulting firm basically for $6.5 consulting firm basically for $6.5 consulting firm basically for $6.5 billion. Whatever. Yeah. Yeah. I don't billion. Whatever. Yeah. Yeah. I don't billion. Whatever. Yeah. Yeah. I don't get that. I apologize. I'm so jaded. get that. I apologize. I'm so jaded. get that. I apologize. I'm so jaded. That sounds perfectly legit. And I'm That sounds perfectly legit. And I'm That sounds perfectly legit. And I'm doing product placement for Dr. Pepper. doing product placement for Dr. Pepper. doing product placement for Dr. Pepper. Anyway, Anyway, Anyway, so you had a good week, huh? You saw so you had a good week, huh? You saw so you had a good week, huh? You saw that and got Well, you know, here we that and got Well, you know, here we that and got Well, you know, here we have to warn our audience. This is like have to warn our audience. This is like have to warn our audience. This is like we are recording at the end of the day we are recording at the end of the day we are recording at the end of the day on Friday. Exactly. So, we're even on Friday. Exactly. So, we're even on Friday. Exactly. So, we're even gruntled, right? Yes. Grumpy.

gruntled, right? Yes. Grumpy. gruntled, right? Yes. Grumpy. Absolutely. Absolutely. But we're Absolutely. Absolutely. But we're Absolutely. Absolutely. But we're looking forward to threeday weekend. looking forward to threeday weekend. looking forward to threeday weekend. Yeah. Yeah, that's right. So, that's Yeah. Yeah, that's right. So, that's Yeah. Yeah, that's right. So, that's good. Well, you are. I don't. There's no good. Well, you are. I don't. There's no good. Well, you are. I don't. There's no I was about to, you know, it's funny you I was about to, you know, it's funny you I was about to, you know, it's funny you said that. I was about to say, but do said that. I was about to say, but do said that. I was about to say, but do you have a three-day weekend? It's your you have a three-day weekend? It's your you have a three-day weekend? It's your deal. It's your gig. And if you don't deal. It's your gig. And if you don't deal. It's your gig. And if you don't work, work, work, Diero. No Diero. Yeah. Yeah. You got to Diero. No Diero. Yeah. Yeah. You got to Diero. No Diero. Yeah. Yeah. You got to keep cranking it out. You got to write keep cranking it out. You got to write keep cranking it out. You got to write another 5,000word document and get it another 5,000word document and get it another 5,000word document and get it out there. Pump it out, man. out there. Pump it out, man. out there. Pump it out, man. Yeah, exactly. So there's no no risk for Yeah, exactly. So there's no no risk for Yeah, exactly. So there's no no risk for um the entrepreneur, you know, you know, um the entrepreneur, you know, you know, um the entrepreneur, you know, you know, the two documents I wanted to write and the two documents I wanted to write and the two documents I wanted to write and I finally did I wrote they're not I finally did I wrote they're not I finally did I wrote they're not published yet, but one of them is um published yet, but one of them is um published yet, but one of them is um they're both about private cloud, but they're both about private cloud, but they're both about private cloud, but one of them is uh why industry 4.0 needs one of them is uh why industry 4.0 needs one of them is uh why industry 4.0 needs a private cloud. a private cloud. a private cloud. And the thinking is, you know, so much And the thinking is, you know, so much And the thinking is, you know, so much of that lives in factories and all that of that lives in factories and all that of that lives in factories and all that kind of stuff and it's really on prim kind of stuff and it's really on prim kind of stuff and it's really on prim and very edgy and so I kind of did that.

and very edgy and so I kind of did that. and very edgy and so I kind of did that. So I just wanted to get it out of my So I just wanted to get it out of my So I just wanted to get it out of my system. And the other one I wanted to system. And the other one I wanted to system. And the other one I wanted to get out of my system was I wrote one get out of my system was I wrote one get out of my system was I wrote one called the emergence of micro clouds. called the emergence of micro clouds. called the emergence of micro clouds. So that's the the little mini me clouds So that's the the little mini me clouds So that's the the little mini me clouds that kind of have cloud private cloud that kind of have cloud private cloud that kind of have cloud private cloud characteristics but run at the edge like characteristics but run at the edge like characteristics but run at the edge like not in a data center. But who knows? not in a data center. But who knows? not in a data center. But who knows? Maybe in your car. I don't know. That's Maybe in your car. I don't know. That's Maybe in your car. I don't know. That's why you need to go to Computex, bro, why you need to go to Computex, bro, why you need to go to Computex, bro, because they if you go to the because they if you go to the because they if you go to the Vertex, you went to the Vertex booth. Vertex, you went to the Vertex booth. Vertex, you went to the Vertex booth. Yeah. You saw these little micro data Yeah. You saw these little micro data Yeah. You saw these little micro data centers in sort of a car, you know, like centers in sort of a car, you know, like centers in sort of a car, you know, like cargo container size. Yeah. Yeah. Yeah. cargo container size. Yeah. Yeah. Yeah. cargo container size. Yeah. Yeah. Yeah. That's right. It's a it's a thing. So, That's right. It's a it's a thing. So, That's right. It's a it's a thing. So, you know, they are looking at more you know, they are looking at more you know, they are looking at more modular data center form factors other modular data center form factors other modular data center form factors other than these massive than these massive than these massive Yeah. massive um data centers that we Yeah. massive um data centers that we Yeah. massive um data centers that we see being built in Texas and Tennessee see being built in Texas and Tennessee see being built in Texas and Tennessee and all that other Right. That's exactly and all that other Right. That's exactly and all that other Right. That's exactly what I wrote. A large scale microcloud what I wrote. A large scale microcloud what I wrote. A large scale microcloud might be a shipping container. Yeah.

might be a shipping container. Yeah. might be a shipping container. Yeah. Yeah. Right. Absolutely. It's all fun, Yeah. Right. Absolutely. It's all fun, Yeah. Right. Absolutely. It's all fun, man. Yeah. So, man. Yeah. So, man. Yeah. So, and then you have your mini me micro and then you have your mini me micro and then you have your mini me micro cloud. Yeah. Three nooks or it could be cloud. Yeah. Three nooks or it could be cloud. Yeah. Three nooks or it could be just three independent Snapdragon just three independent Snapdragon just three independent Snapdragon processors. Yeah. Even smaller with processors. Yeah. Even smaller with processors. Yeah. Even smaller with microclustered storage and it's in your microclustered storage and it's in your microclustered storage and it's in your Tesla or something. I don't know. Tesla or something. I don't know. Tesla or something. I don't know. Whatever. Whatever. Whatever. Could Could Could be. Well, you know that that is kind of be. Well, you know that that is kind of be. Well, you know that that is kind of like that the whole idea of that DJX like that the whole idea of that DJX like that the whole idea of that DJX Spark, right? cuz you can you can daisy Spark, right? cuz you can you can daisy Spark, right? cuz you can you can daisy chain those things. I think they have chain those things. I think they have chain those things. I think they have like connect connect X7. like connect connect X7. like connect connect X7. Okay. For you know connecting. So it's Okay. For you know connecting. So it's Okay. For you know connecting. So it's Ethernet connecting Ethernet connecting Ethernet connecting um these you know those X number of um these you know those X number of um these you know those X number of units you can just daisy chain those units you can just daisy chain those units you can just daisy chain those things and then scale them out. So, things and then scale them out. So, things and then scale them out. So, okay, you know, they can get powerful okay, you know, they can get powerful okay, you know, they can get powerful and it's all, you know, envy linked and it's all, you know, envy linked and it's all, you know, envy linked together. So, you're golden, right? And together. So, you're golden, right? And together. So, you're golden, right? And then pretty soon they're going to have then pretty soon they're going to have then pretty soon they're going to have an interface between Envy Link and an interface between Envy Link and an interface between Envy Link and Neurolink and they'll be plugged in your Neurolink and they'll be plugged in your Neurolink and they'll be plugged in your brain. Yeah, brain. Yeah, brain. Yeah, you got you got that going for you is you got you got that going for you is you got you got that going for you is the plan. Oh, hey. Yeah, speaking about the plan. Oh, hey. Yeah, speaking about the plan. Oh, hey. Yeah, speaking about Ner uh what do you call Envy Link? So, Ner uh what do you call Envy Link? So, Ner uh what do you call Envy Link? So, one of the big announcements was uh Envy one of the big announcements was uh Envy one of the big announcements was uh Envy Link uh Fusion, okay? Which is funny Link uh Fusion, okay? Which is funny Link uh Fusion, okay? Which is funny because, you know, it's like Apple Ultra

because, you know, it's like Apple Ultra because, you know, it's like Apple Ultra Fusion and then now um you know, that Fusion and then now um you know, that Fusion and then now um you know, that was like uh you know, see that's the was like uh you know, see that's the was like uh you know, see that's the thing. Apple people talk crap about thing. Apple people talk crap about thing. Apple people talk crap about Apple. Apple was the first company to I Apple. Apple was the first company to I Apple. Apple was the first company to I I'm I'm pretty sure I'm I'm pretty sure I'm I'm pretty sure uh use co-as, right? um chip on uh use co-as, right? um chip on uh use co-as, right? um chip on substrate on wafer that that uh advanced substrate on wafer that that uh advanced substrate on wafer that that uh advanced packaging technology that TSMC to create packaging technology that TSMC to create packaging technology that TSMC to create the ultra. So they got two um what do the ultra. So they got two um what do the ultra. So they got two um what do you call it? Uh you call it? Uh you call it? Uh A1 no A1 that's a M1. Yeah, that's a two A1 no A1 that's a M1. Yeah, that's a two A1 no A1 that's a M1. Yeah, that's a two M1 Ultras. Yeah, M1 No, M1 Maxes. And M1 Ultras. Yeah, M1 No, M1 Maxes. And M1 Ultras. Yeah, M1 No, M1 Maxes. And they together, right? Right. Yeah. And they together, right? Right. Yeah. And they together, right? Right. Yeah. And so when you look at what so when you look at what so when you look at what um um you know like what uh basically um um you know like what uh basically um um you know like what uh basically Blackwell is is a two H 100 die stuck Blackwell is is a two H 100 die stuck Blackwell is is a two H 100 die stuck together, right? Okay. Together exactly together, right? Okay. Together exactly together, right? Okay. Together exactly the same way, but that's like 3 years the same way, but that's like 3 years the same way, but that's like 3 years after or four years after, right? So after or four years after, right? So after or four years after, right? So yeah, you know, Apple's in there doing yeah, you know, Apple's in there doing yeah, you know, Apple's in there doing advanced packaging before everyone else.

advanced packaging before everyone else. advanced packaging before everyone else. And yeah, um they're not getting any And yeah, um they're not getting any And yeah, um they're not getting any credit for it. No, they get zero credit credit for it. No, they get zero credit credit for it. No, they get zero credit for a lot of stuff. But the other crazy for a lot of stuff. But the other crazy for a lot of stuff. But the other crazy thing is when you look at the new three, thing is when you look at the new three, thing is when you look at the new three, the the the M3 Ultra. Yeah. That thing that thing M3 Ultra. Yeah. That thing that thing M3 Ultra. Yeah. That thing that thing maxes out at something like five like maxes out at something like five like maxes out at something like five like half a ter half a ter half a ter um Yeah. What is it? 500 I think it's um Yeah. What is it? 500 I think it's um Yeah. What is it? 500 I think it's like 500 gigs of unified memory. Okay. like 500 gigs of unified memory. Okay. like 500 gigs of unified memory. Okay. It's insane because one of the things It's insane because one of the things It's insane because one of the things that starting to realize as you start to that starting to realize as you start to that starting to realize as you start to move toward inference computing, it's move toward inference computing, it's move toward inference computing, it's XPU. XPU. XPU. It's big XPU. You know, you have all It's big XPU. You know, you have all It's big XPU. You know, you have all these guys going on, hey, you know, XPU, these guys going on, hey, you know, XPU, these guys going on, hey, you know, XPU, XPU, it's a the small X is actually um XPU, it's a the small X is actually um XPU, it's a the small X is actually um small XPU is accelerator. Big XPU is small XPU is accelerator. Big XPU is small XPU is accelerator. Big XPU is heterogeneous computing meaning CPU, heterogeneous computing meaning CPU, heterogeneous computing meaning CPU, GPU, maybe you have like your MPU and um GPU, maybe you have like your MPU and um GPU, maybe you have like your MPU and um the inference workloads and depending on the inference workloads and depending on the inference workloads and depending on the model that you have coming in uh the model that you have coming in uh the model that you have coming in uh will use CPU, GPU and uh MPU in will use CPU, GPU and uh MPU in will use CPU, GPU and uh MPU in different ways, right? And also you're different ways, right? And also you're different ways, right? And also you're now in what we're starting to realize is now in what we're starting to realize is now in what we're starting to realize is that there's a tiered memory structure that there's a tiered memory structure that there's a tiered memory structure that you're likely going to be using and that you're likely going to be using and that you're likely going to be using and having that unified memory. Think about having that unified memory. Think about having that unified memory. Think about it how quick that that's going to be.

it how quick that that's going to be. it how quick that that's going to be. Yeah, totally. Right. And you have Yeah, totally. Right. And you have Yeah, totally. Right. And you have massive amounts of uh shared memory that massive amounts of uh shared memory that massive amounts of uh shared memory that you can tap between different IP. you can tap between different IP. you can tap between different IP. some of these models for inference are some of these models for inference are some of these models for inference are gonna are are I think um run much more gonna are are I think um run much more gonna are are I think um run much more efficiently and much faster. So, we'll efficiently and much faster. So, we'll efficiently and much faster. So, we'll see. But I I've heard and seen some good see. But I I've heard and seen some good see. But I I've heard and seen some good numbers come out and benchmarks for um numbers come out and benchmarks for um numbers come out and benchmarks for um for the uh M3 for the uh M3 for the uh M3 Ultra some of these large models. So, I Ultra some of these large models. So, I Ultra some of these large models. So, I can't remember. Did you buy the latest can't remember. Did you buy the latest can't remember. Did you buy the latest uh iPad Pro? uh iPad Pro? uh iPad Pro? No. Okay. Cuz that that one has the M4 No. Okay. Cuz that that one has the M4 No. Okay. Cuz that that one has the M4 and you can like max it out and make it and you can like max it out and make it and you can like max it out and make it like cost two grand or something. Yeah. like cost two grand or something. Yeah. like cost two grand or something. Yeah. Or it's like more like three Or it's like more like three Or it's like more like three for a tablet. I kind of like went down for a tablet. I kind of like went down for a tablet. I kind of like went down your route, man. Um I just got like a an your route, man. Um I just got like a an your route, man. Um I just got like a an an Air, you know, with an M processor.

an Air, you know, with an M processor. an Air, you know, with an M processor. And guess what? It It works great. Yeah, And guess what? It It works great. Yeah, And guess what? It It works great. Yeah, it works fine. You know, did you get 11 it works fine. You know, did you get 11 it works fine. You know, did you get 11 or 13 in? 13 always. Yeah. Yeah. I mean, or 13 in? 13 always. Yeah. Yeah. I mean, or 13 in? 13 always. Yeah. Yeah. I mean, I I have the old I've got the 10th I I have the old I've got the 10th I I have the old I've got the 10th generation iPad, which was the first one generation iPad, which was the first one generation iPad, which was the first one they had where they put the camera on they had where they put the camera on they had where they put the camera on the landscape side. Yeah. Like what I'm the landscape side. Yeah. Like what I'm the landscape side. Yeah. Like what I'm doing now, but more like that. And but doing now, but more like that. And but doing now, but more like that. And but I'm on the iPhone here. Yeah. And so, I'm on the iPhone here. Yeah. And so, I'm on the iPhone here. Yeah. And so, uh, you know, it got weird. Yeah. Yeah. uh, you know, it got weird. Yeah. Yeah. uh, you know, it got weird. Yeah. Yeah. Yeah. Yeah. Um Um Yeah. So that that was Yeah. Yeah. Um Um Yeah. So that that was Yeah. Yeah. Um Um Yeah. So that that was it was a great purchase for me. Yeah. it was a great purchase for me. Yeah. it was a great purchase for me. Yeah. You know, it's slower CPU and obviously You know, it's slower CPU and obviously You know, it's slower CPU and obviously stuff like that, but and it was also stuff like that, but and it was also stuff like that, but and it was also their first one that had 5G in it. And their first one that had 5G in it. And their first one that had 5G in it. And so it was like it was a deal. But yeah, so it was like it was a deal. But yeah, so it was like it was a deal. But yeah, it's not nearly as fast and as capable it's not nearly as fast and as capable it's not nearly as fast and as capable as like an M3 Air. Um and yeah, it's as like an M3 Air. Um and yeah, it's as like an M3 Air. Um and yeah, it's like 10.9 in, I guess. Yeah, mine is M2, like 10.9 in, I guess. Yeah, mine is M2, like 10.9 in, I guess. Yeah, mine is M2, but Okay. Um, yeah, I I I don't really but Okay. Um, yeah, I I I don't really but Okay. Um, yeah, I I I don't really carry around my my Mac all that much, if carry around my my Mac all that much, if carry around my my Mac all that much, if at all. So, right, I I use iPad all the at all. So, right, I I use iPad all the at all. So, right, I I use iPad all the time. When I see you at the conferences, time. When I see you at the conferences, time. When I see you at the conferences, you got your little small you got your little small you got your little small kind of backpack, whatever you want to kind of backpack, whatever you want to kind of backpack, whatever you want to call that.

call that. call that. Um, which is great. Yeah. Keeping it Um, which is great. Yeah. Keeping it Um, which is great. Yeah. Keeping it fast and light, man. Exactly. Exactly. fast and light, man. Exactly. Exactly. fast and light, man. Exactly. Exactly. No, I'm totally down with that. You use No, I'm totally down with that. You use No, I'm totally down with that. You use You use what you use. You don't And you You use what you use. You don't And you You use what you use. You don't And you take what you use. You don't take what take what you use. You don't take what take what you use. You don't take what you don't use. you don't use. you don't use. Genius. Yeah. You know, you learn after Genius. Yeah. You know, you learn after Genius. Yeah. You know, you learn after years of traveling, right? I mean, Yeah. years of traveling, right? I mean, Yeah. years of traveling, right? I mean, Yeah. I mean, cuz you're there and what are I mean, cuz you're there and what are I mean, cuz you're there and what are you doing? You're taking notes. Yeah. you doing? You're taking notes. Yeah. you doing? You're taking notes. Yeah. And then you're taking notes, dude. And then you're taking notes, dude. And then you're taking notes, dude. Think about how many I mean literally Think about how many I mean literally Think about how many I mean literally how many miles we sometimes walk at some how many miles we sometimes walk at some how many miles we sometimes walk at some of the larger conferences, right? You of the larger conferences, right? You of the larger conferences, right? You know, it's insane. Are you going to know, it's insane. Are you going to know, it's insane. Are you going to anything in June? Yeah, a lot of stuff. anything in June? Yeah, a lot of stuff. anything in June? Yeah, a lot of stuff. Uh I'm going to DTW and Copenhagen. So, Uh I'm going to DTW and Copenhagen. So, Uh I'm going to DTW and Copenhagen. So, I'll be flying there for all the OSS, I'll be flying there for all the OSS, I'll be flying there for all the OSS, BSS, you know, telco BSS, you know, telco BSS, you know, telco CRM, you know, I think they're going to CRM, you know, I think they're going to CRM, you know, I think they're going to have like network API related stuff, but have like network API related stuff, but have like network API related stuff, but getting into that. Um, and then I'm getting into that. Um, and then I'm getting into that. Um, and then I'm going to Cisco live going to Cisco live going to Cisco live um which is going to be here in San San um which is going to be here in San San um which is going to be here in San San Diego. So, that's awesome. Yeah. fly Diego. So, that's awesome. Yeah. fly Diego. So, that's awesome. Yeah. fly anywhere. Sensors anywhere. Sensors anywhere. Sensors converge and then Oh, yeah. Um, is that converge and then Oh, yeah. Um, is that converge and then Oh, yeah. Um, is that Santa Clara or San Jose? Okay. Yeah.

Santa Clara or San Jose? Okay. Yeah. Santa Clara or San Jose? Okay. Yeah. What else? I mean, dude, look at look at What else? I mean, dude, look at look at What else? I mean, dude, look at look at this week, man. We had Microsoft Build, this week, man. We had Microsoft Build, this week, man. We had Microsoft Build, we had Google IO, I think. That's right. we had Google IO, I think. That's right. we had Google IO, I think. That's right. And you had Dell. Yeah. Dell Tech World. And you had Dell. Yeah. Dell Tech World. And you had Dell. Yeah. Dell Tech World. Then you had Tech and God knows what. Then you had Tech and God knows what. Then you had Tech and God knows what. Oh, no. Red Hat had their thing. Yep. Oh, no. Red Hat had their thing. Yep. Oh, no. Red Hat had their thing. Yep. Right. Right. Right. catch up on. But there's like so much catch up on. But there's like so much catch up on. But there's like so much crap in just one week. So much, you crap in just one week. So much, you crap in just one week. So much, you know, so much. I've got the beginning of know, so much. I've got the beginning of know, so much. I've got the beginning of June. I got a NetApp thing in San June. I got a NetApp thing in San June. I got a NetApp thing in San Jose. I've got Jose. I've got Jose. I've got Pure Storage in Pure Storage in Pure Storage in Vegas at Resorts World. And I've got and Vegas at Resorts World. And I've got and Vegas at Resorts World. And I've got and then HP's big event. HP's big event, then HP's big event. HP's big event, then HP's big event. HP's big event, which is the Venetian at at the Sphere which is the Venetian at at the Sphere which is the Venetian at at the Sphere as well. They they'll do the their as well. They they'll do the their as well. They they'll do the their keynote at the sphere again. Again, like keynote at the sphere again. Again, like keynote at the sphere again. Again, like they did last year. Yeah, exactly. they did last year. Yeah, exactly. they did last year. Yeah, exactly. So, a lot of a lot of backtoback Vegas So, a lot of a lot of backtoback Vegas So, a lot of a lot of backtoback Vegas weeks actually next month. So, it's weeks actually next month. So, it's weeks actually next month. So, it's going to be crazy. Yeah, it's pretty going to be crazy. Yeah, it's pretty going to be crazy. Yeah, it's pretty nuts. Um I know obviously they're all nuts. Um I know obviously they're all nuts. Um I know obviously they're all trying to show off how advanced they are trying to show off how advanced they are trying to show off how advanced they are with the Agentic AI stuff and Exactly.

with the Agentic AI stuff and Exactly. with the Agentic AI stuff and Exactly. Exactly. Yeah. We got to give some Exactly. Yeah. We got to give some Exactly. Yeah. We got to give some honest assessments of these tools cuz honest assessments of these tools cuz honest assessments of these tools cuz there, you know, I don't think people there, you know, I don't think people there, you know, I don't think people really really really understand how you put these agents understand how you put these agents understand how you put these agents together and this whole notion of together and this whole notion of together and this whole notion of agentic tuning or agent tuning or agentic tuning or agent tuning or agentic tuning or agent tuning or whatever this thing that Satia announced whatever this thing that Satia announced whatever this thing that Satia announced yesterday. Oh yeah. dangerous that is if yesterday. Oh yeah. dangerous that is if yesterday. Oh yeah. dangerous that is if you're not on the back end talking about you're not on the back end talking about you're not on the back end talking about how do you how do you incorporate safety how do you how do you incorporate safety how do you how do you incorporate safety reliability and then governance around reliability and then governance around reliability and then governance around the entire op and SD like call it the the entire op and SD like call it the the entire op and SD like call it the not software the not software the not software the AL C or A I LC the is that no A I AL C or A I LC the is that no A I AL C or A I LC the is that no A I DLC Right. I mean going like, "Hey, you can Right. I mean going like, "Hey, you can fine-tune. Just push a button and then fine-tune. Just push a button and then fine-tune. Just push a button and then we have a bunch of canned stuff and we have a bunch of canned stuff and we have a bunch of canned stuff and boom, that's all you have to do." It's boom, that's all you have to do." It's boom, that's all you have to do." It's like, "No, that is not all you have." Oh like, "No, that is not all you have." Oh like, "No, that is not all you have." Oh my god.

my god. my god. I I You know what I'm saying? Is how do I I You know what I'm saying? Is how do I I You know what I'm saying? Is how do these guys get up on stage and say stuff these guys get up on stage and say stuff these guys get up on stage and say stuff like that? It It just blows my mind, like that? It It just blows my mind, like that? It It just blows my mind, man. Maybe it just is the same stuff as man. Maybe it just is the same stuff as man. Maybe it just is the same stuff as the silliness you were talking about the silliness you were talking about the silliness you were talking about earlier back in the dot era when people earlier back in the dot era when people earlier back in the dot era when people said crazy stuff. Um, you know, and said crazy stuff. Um, you know, and said crazy stuff. Um, you know, and almost all of them went out of business. almost all of them went out of business. almost all of them went out of business. No, seriously, man. If I if I was a CIO, No, seriously, man. If I if I was a CIO, No, seriously, man. If I if I was a CIO, if any, you know, CIOS I I advise take if any, you know, CIOS I I advise take if any, you know, CIOS I I advise take heed, you know, ask really hard heed, you know, ask really hard heed, you know, ask really hard questions and a lot of the stuff is not questions and a lot of the stuff is not questions and a lot of the stuff is not going to make sense to them. They're going to make sense to them. They're going to make sense to them. They're going to go, "Okay, you're telling me going to go, "Okay, you're telling me going to go, "Okay, you're telling me that any employee can go out there and that any employee can go out there and that any employee can go out there and then fine-tune a model, right? So, then fine-tune a model, right? So, then fine-tune a model, right? So, they're asking chat GPT or co-pilot, how they're asking chat GPT or co-pilot, how they're asking chat GPT or co-pilot, how do I create a agent for do I create a agent for do I create a agent for XYZ?" And they're going to go and XYZ?" And they're going to go and XYZ?" And they're going to go and fine-tune their own fine-tune their own fine-tune their own model and then they're going to create model and then they're going to create model and then they're going to create their own agent, chain them together their own agent, chain them together their own agent, chain them together using MCP.

using MCP. using MCP. You're done with MCP is not. You know You're done with MCP is not. You know You're done with MCP is not. You know me. You tell me how that me. You tell me how that me. You tell me how that is just simply not okay. I know. I know. is just simply not okay. I know. I know. is just simply not okay. I know. I know. No. No. Seriously, you It's not okay. No. No. Seriously, you It's not okay. No. No. Seriously, you It's not okay. It's just crazy town. It's um It's just crazy town. It's um It's just crazy town. It's um it it it it's like um it's like um what it it it it's like um it's like um what it it it it's like um it's like um what do you call it? Shadow it on like do you call it? Shadow it on like do you call it? Shadow it on like freaking crack cocaine. It's right. It's freaking crack cocaine. It's right. It's freaking crack cocaine. It's right. It's like a cocaine bear. you know, with my with my hat on and you know, with my with my hat on and these glasses and this beard, I kind of these glasses and this beard, I kind of these glasses and this beard, I kind of look like a Hollywood movie director, look like a Hollywood movie director, look like a Hollywood movie director, don't I? You know, and we're we're don't I? You know, and we're we're don't I? You know, and we're we're pitching this new movie. We're going to pitching this new movie. We're going to pitching this new movie. We're going to call it The Singularity of call it The Singularity of call it The Singularity of Death. AI's coming for you, baby. Well, Death. AI's coming for you, baby. Well, Death. AI's coming for you, baby. Well, I mean, think about it, dude. It's like, I mean, think about it, dude. It's like, I mean, think about it, dude. It's like, oh, yeah, you can fine-tune on your own oh, yeah, you can fine-tune on your own oh, yeah, you can fine-tune on your own data. It's like, data. It's like, data. It's like, yeah, does an average employee even know yeah, does an average employee even know yeah, does an average employee even know what the quality of data is? And where's what the quality of data is? And where's what the quality of data is? And where's a testing cycle? What tools do you have a testing cycle? What tools do you have a testing cycle? What tools do you have to help test whether or not that entire to help test whether or not that entire to help test whether or not that entire agent is reliable and actually can uh, agent is reliable and actually can uh, agent is reliable and actually can uh, you know, consistently execute on a you know, consistently execute on a you know, consistently execute on a certain range of tasks. It's a great certain range of tasks. It's a great certain range of tasks. It's a great that's a great point because you know that's a great point because you know that's a great point because you know stuff we've always said or or guests stuff we've always said or or guests stuff we've always said or or guests we've had on who remind us before this we've had on who remind us before this we've had on who remind us before this AI agentic thing got crazy or generative AI agentic thing got crazy or generative AI agentic thing got crazy or generative you know machine learning stuff people you know machine learning stuff people you know machine learning stuff people said yeah but most of the people who said yeah but most of the people who said yeah but most of the people who work in machine learning are data work in machine learning are data work in machine learning are data engineers and they spend all their time

engineers and they spend all their time engineers and they spend all their time cleaning the data and to make it cleaning the data and to make it cleaning the data and to make it actually work because most data is just actually work because most data is just actually work because most data is just crap and guess what your agentic stuff crap and guess what your agentic stuff crap and guess what your agentic stuff and your generous stuff isn't it's going and your generous stuff isn't it's going and your generous stuff isn't it's going to choke on it just like old school ML to choke on it just like old school ML to choke on it just like old school ML choked on it. Ex, dude. I'm telling you choked on it. Ex, dude. I'm telling you choked on it. Ex, dude. I'm telling you that that's the thing. And so when I was that that's the thing. And so when I was that that's the thing. And so when I was at Computex, I I was talking to a bunch at Computex, I I was talking to a bunch at Computex, I I was talking to a bunch of the um you know, like Foccom, uh uh of the um you know, like Foccom, uh uh of the um you know, like Foccom, uh uh Avantec and a bunch of other companies Avantec and a bunch of other companies Avantec and a bunch of other companies that are sort of doing the embedded that are sort of doing the embedded that are sort of doing the embedded things, you know, automation, IoTish things, you know, automation, IoTish things, you know, automation, IoTish kind of stuff. And obviously all of them kind of stuff. And obviously all of them kind of stuff. And obviously all of them are getting into the are getting into the are getting into the AI stuff. I mean, it's like literally AI stuff. I mean, it's like literally AI stuff. I mean, it's like literally taking a ML the ML problem and then taking a ML the ML problem and then taking a ML the ML problem and then taking to the next level of holy crap. taking to the next level of holy crap. taking to the next level of holy crap. You know what I'm saying? And nothing You know what I'm saying? And nothing You know what I'm saying? And nothing has changed. Nothing changed. None of has changed. Nothing changed. None of has changed. Nothing changed. None of this stuff all of a sudden got so much this stuff all of a sudden got so much this stuff all of a sudden got so much easier. You know what I mean? Right. easier. You know what I mean? Right. easier. You know what I mean? Right. My son just awesome. Um, he forgot that My son just awesome. Um, he forgot that My son just awesome. Um, he forgot that he forgot that you existed. Oh, I he forgot that you existed. Oh, I he forgot that you existed. Oh, I know. Family doesn't know I exist. They know. Family doesn't know I exist. They know. Family doesn't know I exist. They go, "Who is this strange guy? Who is go, "Who is this strange guy? Who is go, "Who is this strange guy? Who is this guy who broke into our house?" All this guy who broke into our house?" All this guy who broke into our house?" All the clicks are not.

the clicks are not. the clicks are not. Absolutely. Oh my gosh. That's crazy Absolutely. Oh my gosh. That's crazy Absolutely. Oh my gosh. That's crazy town, dude. Um, no. Think about it, Rob. town, dude. Um, no. Think about it, Rob. town, dude. Um, no. Think about it, Rob. I mean, what was ML like I mean, what was ML like I mean, what was ML like before, you know, all the AI stuff before, you know, all the AI stuff before, you know, all the AI stuff before this version of AI and what was before this version of AI and what was before this version of AI and what was it like, man? You know what? It was very it like, man? You know what? It was very it like, man? You know what? It was very it's very narrow AI. It's just narrow it's very narrow AI. It's just narrow it's very narrow AI. It's just narrow narrow narrow narrow use narrow narrow narrow use narrow narrow narrow use cases. We are sending good and we're cases. We are sending good and we're cases. We are sending good and we're sending you all this data about every sending you all this data about every sending you all this data about every parameter of every little feature of a parameter of every little feature of a parameter of every little feature of a machine whatever it is a car machine in machine whatever it is a car machine in machine whatever it is a car machine in a factory and we're sending tons of data a factory and we're sending tons of data a factory and we're sending tons of data you know time series data about you know time series data about you know time series data about everything. And along with that is when everything. And along with that is when everything. And along with that is when it's running and when it's failed or it's running and when it's failed or it's running and when it's failed or when it's about to fail. Yeah. And then when it's about to fail. Yeah. And then when it's about to fail. Yeah. And then over a long period of time with enough over a long period of time with enough over a long period of time with enough data, machine learning could go, oh, data, machine learning could go, oh, data, machine learning could go, oh, it's just finding a needle of hay stack it's just finding a needle of hay stack it's just finding a needle of hay stack and it goes, oh, when I see these and it goes, oh, when I see these and it goes, oh, when I see these properties of the way this machine is properties of the way this machine is properties of the way this machine is behaving right behaving right behaving right now, there's more than likely the now, there's more than likely the now, there's more than likely the machine's going to fail. you know, and machine's going to fail. you know, and machine's going to fail. you know, and that that was what everybody's focused that that was what everybody's focused that that was what everybody's focused on, you know, is uptime is using machine on, you know, is uptime is using machine on, you know, is uptime is using machine learning for predictive maintenance. Um, learning for predictive maintenance. Um, learning for predictive maintenance. Um, and it still is, you know, they want to and it still is, you know, they want to and it still is, you know, they want to they want to forecast when things are they want to forecast when things are they want to forecast when things are bad, things are going to happen. You bad, things are going to happen. You bad, things are going to happen. You know, the whole deal was, hey, we always know, the whole deal was, hey, we always know, the whole deal was, hey, we always depend on the old guy at the factory and depend on the old guy at the factory and depend on the old guy at the factory and when he puts his hand on the machine, he when he puts his hand on the machine, he when he puts his hand on the machine, he feels a weird vibration. He knows, oh,

feels a weird vibration. He knows, oh, feels a weird vibration. He knows, oh, this means that it's going to fail next this means that it's going to fail next this means that it's going to fail next Tuesday, you know. But what do we do Tuesday, you know. But what do we do Tuesday, you know. But what do we do when that guy retires, you know, and so when that guy retires, you know, and so when that guy retires, you know, and so how do we replace all the what like how do we replace all the what like how do we replace all the what like tribal knowledge as we call it with tribal knowledge as we call it with tribal knowledge as we call it with machine learning, you know, and that's machine learning, you know, and that's machine learning, you know, and that's been a lot of work and now it's getting been a lot of work and now it's getting been a lot of work and now it's getting pretty good that that you know, I've pretty good that that you know, I've pretty good that that you know, I've talked about that using that autoML to talked about that using that autoML to talked about that using that autoML to make that a lot easier. U but that's make that a lot easier. U but that's make that a lot easier. U but that's getting overshadowed right now by getting overshadowed right now by getting overshadowed right now by generative. Yeah. Yeah. So you actually generative. Yeah. Yeah. So you actually generative. Yeah. Yeah. So you actually have some tech that actually is finally have some tech that actually is finally have some tech that actually is finally working and actually can deliver real working and actually can deliver real working and actually can deliver real value for companies, but no one's even value for companies, but no one's even value for companies, but no one's even looking at it now because they're like, looking at it now because they're like, looking at it now because they're like, "Oh, look at this shiny object over here "Oh, look at this shiny object over here "Oh, look at this shiny object over here now." You know? So yeah, I mean, so one now." You know? So yeah, I mean, so one now." You know? So yeah, I mean, so one of the things that was interesting that of the things that was interesting that of the things that was interesting that I noticed in the in Satia's keynote was I noticed in the in Satia's keynote was I noticed in the in Satia's keynote was he mentioned you can intermingle with it he mentioned you can intermingle with it he mentioned you can intermingle with it with deterministic workflows, right? And with deterministic workflows, right? And with deterministic workflows, right? And so I also assume he means other so I also assume he means other so I also assume he means other deterministic functions. But but that so deterministic functions. But but that so deterministic functions. But but that so when you start talking that about that when you start talking that about that when you start talking that about that you're either relegating you're either relegating you're either relegating u generative AI to a high level. So it's u generative AI to a high level. So it's u generative AI to a high level. So it's not it's not doing everything. It's not not it's not doing everything. It's not not it's not doing everything. It's not doing the critical um bits and pieces doing the critical um bits and pieces doing the critical um bits and pieces that are actually going to make that are actually going to make that are actually going to make something happen, your business happen something happen, your business happen something happen, your business happen with reliability, right? Consistency and with reliability, right? Consistency and with reliability, right? Consistency and quality. It it's up here and then you're quality. It it's up here and then you're quality. It it's up here and then you're squishing it, you know? You're like squishing it, you know? You're like squishing it, you know? You're like creating like little isolated

creating like little isolated creating like little isolated um applets, not even apps. I mean, it's um applets, not even apps. I mean, it's um applets, not even apps. I mean, it's like you do this, you just check this like you do this, you just check this like you do this, you just check this stuff and hopefully we can tune it and stuff and hopefully we can tune it and stuff and hopefully we can tune it and do all this expensive, you know, um, do all this expensive, you know, um, do all this expensive, you know, um, fine-tuning so that you can at least do fine-tuning so that you can at least do fine-tuning so that you can at least do this narrow thing. Going back to what this narrow thing. Going back to what this narrow thing. Going back to what you said, narrow little you said, narrow little you said, narrow little function and that we know actually function and that we know actually function and that we know actually works. Yeah. And then all of these works. Yeah. And then all of these works. Yeah. And then all of these things are strung together in a logical things are strung together in a logical things are strung together in a logical flow. Isn't that something that we flow. Isn't that something that we flow. Isn't that something that we heard? like there was a lot of noise and heard? like there was a lot of noise and heard? like there was a lot of noise and FUD with DeepSeek R2. But I thought in FUD with DeepSeek R2. But I thought in FUD with DeepSeek R2. But I thought in the midst of that, one of the if you the midst of that, one of the if you the midst of that, one of the if you want to call it innovation, I think it want to call it innovation, I think it want to call it innovation, I think it was them just being pragmatic was they was them just being pragmatic was they was them just being pragmatic was they actually used a deterministic decision actually used a deterministic decision actually used a deterministic decision tree kind of thing upfront to vector tree kind of thing upfront to vector tree kind of thing upfront to vector questions to parts of the LLM or questions to parts of the LLM or questions to parts of the LLM or whatever that was smart in certain whatever that was smart in certain whatever that was smart in certain areas. I'm probably not saying this the areas. I'm probably not saying this the areas. I'm probably not saying this the right way, but they they used a right way, but they they used a right way, but they they used a combination of deterministic and combination of deterministic and combination of deterministic and probabilistic to get better answers more probabilistic to get better answers more probabilistic to get better answers more quickly from deepseek instead of like quickly from deepseek instead of like quickly from deepseek instead of like here's this giant mega LLM that's here's this giant mega LLM that's here's this giant mega LLM that's supposed to be all knowing. Um, it's supposed to be all knowing. Um, it's supposed to be all knowing. Um, it's better to use some of the just typical better to use some of the just typical better to use some of the just typical analytics that we've used to to help analytics that we've used to to help analytics that we've used to to help fine-tune what is this question about?

fine-tune what is this question about? fine-tune what is this question about? I'm now gonna send it to Yeah. this, you I'm now gonna send it to Yeah. this, you I'm now gonna send it to Yeah. this, you know, small language model or this LLM know, small language model or this LLM know, small language model or this LLM that's really more tuned to it, you that's really more tuned to it, you that's really more tuned to it, you know. Um I mean, it's possible that's know. Um I mean, it's possible that's know. Um I mean, it's possible that's the world we're going to be going to, I the world we're going to be going to, I the world we're going to be going to, I think, cuz we're a lot of people are think, cuz we're a lot of people are think, cuz we're a lot of people are saying, "Hey, we're getting some good saying, "Hey, we're getting some good saying, "Hey, we're getting some good results from these smaller models." results from these smaller models." results from these smaller models." Yeah. Well, here's Okay. So, what you're Yeah. Well, here's Okay. So, what you're Yeah. Well, here's Okay. So, what you're describing is a model experts, right? Or describing is a model experts, right? Or describing is a model experts, right? Or a mixture of experts. So, that whole a mixture of experts. So, that whole a mixture of experts. So, that whole thing Yeah. And you're you're sort of thing Yeah. And you're you're sort of thing Yeah. And you're you're sort of right. Um the idea is right is that you right. Um the idea is right is that you right. Um the idea is right is that you have a number of smaller models or LLMs have a number of smaller models or LLMs have a number of smaller models or LLMs that are trained on a specific you know that are trained on a specific you know that are trained on a specific you know set of information. One's good at math, set of information. One's good at math, set of information. One's good at math, one's good at history, one's good at, one's good at history, one's good at, one's good at history, one's good at, let's say, you know, science. Yeah. have let's say, you know, science. Yeah. have let's say, you know, science. Yeah. have a judge a judge a judge orchestrator type LLM and that that's orchestrator type LLM and that that's orchestrator type LLM and that that's sort of that um you know filter and sort of that um you know filter and sort of that um you know filter and orchestrator right it filters the orchestrator right it filters the orchestrator right it filters the request and figures out oh okay what the request and figures out oh okay what the request and figures out oh okay what the hell is this guy asking then it goes hell is this guy asking then it goes hell is this guy asking then it goes down to its orchestration function and down to its orchestration function and down to its orchestration function and it determines which of these experts it it determines which of these experts it it determines which of these experts it should tap gets answers brings it back should tap gets answers brings it back should tap gets answers brings it back up and then it uses like sort of a up and then it uses like sort of a up and then it uses like sort of a ranking to figure out okay what which of ranking to figure out okay what which of ranking to figure out okay what which of these re responses is uh most kosher these re responses is uh most kosher these re responses is uh most kosher right and so the theory here is that now right and so the theory here is that now right and so the theory here is that now instead of scaling out a single like

instead of scaling out a single like instead of scaling out a single like what you're saying a single model to be what you're saying a single model to be what you're saying a single model to be a gajillion parameters you and having a gajillion parameters you and having a gajillion parameters you and having that subject to all kinds of crazy that subject to all kinds of crazy that subject to all kinds of crazy issues let's have these smaller models issues let's have these smaller models issues let's have these smaller models that are orchestrated and managed by a that are orchestrated and managed by a that are orchestrated and managed by a uh an Uber model so that uh an Uber model so that uh an Uber model so that the collective mixture of expert model the collective mixture of expert model the collective mixture of expert model can be a little bit more domain can be a little bit more domain can be a little bit more domain specific. Do you know what I'm saying? specific. Do you know what I'm saying? specific. Do you know what I'm saying? Instead confused and hallucinating like Instead confused and hallucinating like Instead confused and hallucinating like crazy. Yes. Yeah. So like from you know crazy. Yes. Yeah. So like from you know crazy. Yes. Yeah. So like from you know jat GPT to GPT4 which is like that step jat GPT to GPT4 which is like that step jat GPT to GPT4 which is like that step that was that um open AI made to a that was that um open AI made to a that was that um open AI made to a mixture of um mixture of um mixture of um experts model we saw those improvements experts model we saw those improvements experts model we saw those improvements right but since then with the long right but since then with the long right but since then with the long thinking and the reasoning thinking and the reasoning thinking and the reasoning models there's some improvement but you models there's some improvement but you models there's some improvement but you know it's flattening right these things know it's flattening right these things know it's flattening right these things are uh materially getting smarter there are uh materially getting smarter there are uh materially getting smarter there there's definitely a a statistical wall there's definitely a a statistical wall there's definitely a a statistical wall that they're hitting. Yeah. Um, right.

that they're hitting. Yeah. Um, right. that they're hitting. Yeah. Um, right. See, so a lot of confusion about what's See, so a lot of confusion about what's See, so a lot of confusion about what's the difference between AI agent and the difference between AI agent and the difference between AI agent and agentic AI, right? But the agent that agentic AI, right? But the agent that agentic AI, right? But the agent that that mixture of that mixture of that mixture of experts is that first agentic AI because experts is that first agentic AI because experts is that first agentic AI because you can think of all these individual you can think of all these individual you can think of all these individual um experts as being sort of agents. um experts as being sort of agents. um experts as being sort of agents. They're Yeah. Did I hear you say earlier They're Yeah. Did I hear you say earlier They're Yeah. Did I hear you say earlier on in this just thread right here, did on in this just thread right here, did on in this just thread right here, did you say the term expert system or expert you say the term expert system or expert you say the term expert system or expert systems? Do you remember that phrase, systems? Do you remember that phrase, systems? Do you remember that phrase, that terminology? Yeah, of course. Yeah, that terminology? Yeah, of course. Yeah, that terminology? Yeah, of course. Yeah, cuz earlier versions of AI we were cuz earlier versions of AI we were cuz earlier versions of AI we were called expert systems, but what we call called expert systems, but what we call called expert systems, but what we call those today is rules engines. Yeah. And those today is rules engines. Yeah. And those today is rules engines. Yeah. And they worked. They're very deterministic they worked. They're very deterministic they worked. They're very deterministic though, and they worked. But pretty soon though, and they worked. But pretty soon though, and they worked. But pretty soon you you know it's like you know early you you know it's like you know early you you know it's like you know early days of Tesla self-driving is I'm doing days of Tesla self-driving is I'm doing days of Tesla self-driving is I'm doing rules engines and I'm trying to come up rules engines and I'm trying to come up rules engines and I'm trying to come up a rule with every possible edge case. a rule with every possible edge case. a rule with every possible edge case. Yeah. And it's really hard versus I'm Yeah. And it's really hard versus I'm Yeah. And it's really hard versus I'm going to use AI and it's just going to going to use AI and it's just going to going to use AI and it's just going to learn and it's going to magically figure learn and it's going to magically figure learn and it's going to magically figure stuff out. It doesn't learn. We have stuff out. It doesn't learn. We have stuff out. It doesn't learn. We have Oh, painful. You know, this is a the a Oh, painful. You know, this is a the a Oh, painful. You know, this is a the a conversation I had with um a a a an AI conversation I had with um a a a an AI conversation I had with um a a a an AI expert is that these LMS don't learn, expert is that these LMS don't learn, expert is that these LMS don't learn, you have to train them, right?

you have to train them, right? you have to train them, right? We learn, you know, like we can observe We learn, you know, like we can observe We learn, you know, like we can observe something and then we can adapt very something and then we can adapt very something and then we can adapt very quickly and we can absorb new knowledge quickly and we can absorb new knowledge quickly and we can absorb new knowledge and then we can we can change the and then we can we can change the and then we can we can change the weights in our brain, right? Like real weights in our brain, right? Like real weights in our brain, right? Like real time. Yeah. Today's AI can't do that. So time. Yeah. Today's AI can't do that. So time. Yeah. Today's AI can't do that. So these aren't learning systems. You can these aren't learning systems. You can these aren't learning systems. You can train them and then they can infer and train them and then they can infer and train them and then they can infer and provide that whole illusion of thinking. provide that whole illusion of thinking. provide that whole illusion of thinking. But that thinking is different from But that thinking is different from But that thinking is different from learning, you know. Yeah. Also when you learning, you know. Yeah. Also when you learning, you know. Yeah. Also when you think about the experts and panning it think about the experts and panning it think about the experts and panning it out to the right, you know, history, out to the right, you know, history, out to the right, you know, history, math, science, math, science, math, science, whatever. I just for the audience here, whatever. I just for the audience here, whatever. I just for the audience here, I just want them to know that we just I just want them to know that we just I just want them to know that we just coined a new term here today on IoT coined a new term here today on IoT coined a new term here today on IoT Coffee Talk. It's called model sharding. Coffee Talk. It's called model sharding. Coffee Talk. It's called model sharding. Ah yeah yeah yeah. It it came from IoT Ah yeah yeah yeah. It it came from IoT Ah yeah yeah yeah. It it came from IoT coffee talk model sharding. I guarantee coffee talk model sharding. I guarantee coffee talk model sharding. I guarantee you in about a few weeks you're going to you in about a few weeks you're going to you in about a few weeks you're going to start seeing that in the media start seeing that in the media start seeing that in the media and it's like yeah I sharted my model and it's like yeah I sharted my model and it's like yeah I sharted my model across different domains and it's across different domains and it's across different domains and it's working a lot better. I tried to do one working a lot better. I tried to do one working a lot better. I tried to do one giant model for one billion people and giant model for one billion people and giant model for one billion people and my my system crashed. Yeah. Now I'm my my system crashed. Yeah. Now I'm my my system crashed. Yeah. Now I'm sharting models.

sharting models. sharting models. Yeah. my expert system and sending it to Yeah. my expert system and sending it to Yeah. my expert system and sending it to the right model. Um, just like sharding the right model. Um, just like sharding the right model. Um, just like sharding a database, you know, uh, and you and a database, you know, uh, and you and a database, you know, uh, and you and you're going to hear a lot of people you're going to hear a lot of people you're going to hear a lot of people say, "Oh my gosh, I got such great say, "Oh my gosh, I got such great say, "Oh my gosh, I got such great results from my my model. I sharted my results from my my model. I sharted my results from my my model. I sharted my pants." Hey, Will. Good one. It will. Yeah. I Hey, Will. Good one. It will. Yeah. I love it. Love it. Oh my god. The level love it. Love it. Oh my god. The level love it. Love it. Oh my god. The level of genius here is just unparalleled, you of genius here is just unparalleled, you of genius here is just unparalleled, you know. 72. Well, one is it 72 and 140. know. 72. Well, one is it 72 and 140. know. 72. Well, one is it 72 and 140. 144 is like um Yeah. 144 is like um Yeah. 144 is like um Yeah. Oh my god. Good times. Oh my god. Good times. Oh my god. Good times. Yeah. Yeah. Yeah. Um it's it that that was a really good Um it's it that that was a really good Um it's it that that was a really good that was a really good conference. I was that was a really good conference. I was that was a really good conference. I was surprised. I I thought, you know, surprised. I I thought, you know, surprised. I I thought, you know, because so many people were saying because so many people were saying because so many people were saying there's nothing going on there and there's nothing going on there and there's nothing going on there and everyone's at Dell Tech World and I everyone's at Dell Tech World and I everyone's at Dell Tech World and I thought I mean it's like, okay, you're thought I mean it's like, okay, you're thought I mean it's like, okay, you're just trying to make stuff that you just trying to make stuff that you just trying to make stuff that you largely presented last year largely presented last year largely presented last year um look like it's progressed some, um look like it's progressed some, um look like it's progressed some, right? Yeah. Yeah.

right? Yeah. Yeah. right? Yeah. Yeah. But I like last year I was already But I like last year I was already But I like last year I was already writing as like hey man this stuff looks writing as like hey man this stuff looks writing as like hey man this stuff looks like VPN plus RPA plus no code stuff like VPN plus RPA plus no code stuff like VPN plus RPA plus no code stuff that we that's already out you guys are that we that's already out you guys are that we that's already out you guys are just you know so air on top of this just you know so air on top of this just you know so air on top of this thing and uh and then Deltech world I thing and uh and then Deltech world I thing and uh and then Deltech world I didn't I didn't really see anything for didn't I didn't really see anything for didn't I didn't really see anything for shattering as I already know that shattering as I already know that shattering as I already know that they're doing all the stuff with Nvidia they're doing all the stuff with Nvidia they're doing all the stuff with Nvidia and you know sh boxes for that's what and you know sh boxes for that's what and you know sh boxes for that's what actually that's what every company has actually that's what every company has actually that's what every company has in common. They're doing something with in common. They're doing something with in common. They're doing something with Nvidia. Yeah. All they want to do is Nvidia. Yeah. All they want to do is Nvidia. Yeah. All they want to do is have Jensen on stage with their CEO and have Jensen on stage with their CEO and have Jensen on stage with their CEO and say they're doing something with Nvidia say they're doing something with Nvidia say they're doing something with Nvidia in order to pop their stock price and in order to pop their stock price and in order to pop their stock price and get some love. Yeah. Well, I have to get some love. Yeah. Well, I have to get some love. Yeah. Well, I have to give Jensen credit, man. I was at the give Jensen credit, man. I was at the give Jensen credit, man. I was at the Q&A, Q&A, Q&A, too. He didn't talk about quantum, so too. He didn't talk about quantum, so too. He didn't talk about quantum, so good for you. Yay. Good for you. Hey, I good for you. Yay. Good for you. Hey, I good for you. Yay. Good for you. Hey, I I'm sure he really wanted to, but um he I'm sure he really wanted to, but um he I'm sure he really wanted to, but um he didn't go off on quantum and try to say didn't go off on quantum and try to say didn't go off on quantum and try to say that we're doing, you know, exponential that we're doing, you know, exponential that we're doing, you know, exponential quantum XYZ, you know, one slide that quantum XYZ, you know, one slide that quantum XYZ, you know, one slide that talked about how uh GPUs are enabling, talked about how uh GPUs are enabling, talked about how uh GPUs are enabling, you know, like a hybrid model, which is you know, like a hybrid model, which is you know, like a hybrid model, which is okay, that's fair. I actually I think okay, that's fair. I actually I think okay, that's fair. I actually I think that's a really good story because that's a really good story because that's a really good story because obviously with accelerated computing in obviously with accelerated computing in obviously with accelerated computing in his stack. Yeah. Perfect for supporting his stack. Yeah. Perfect for supporting his stack. Yeah. Perfect for supporting the science of you know the computing the science of you know the computing the science of you know the computing science, right? Absolutely. Yeah. But I

science, right? Absolutely. Yeah. But I science, right? Absolutely. Yeah. But I mean that's research that's like stuff mean that's research that's like stuff mean that's research that's like stuff that is probably years away. It is. that is probably years away. It is. that is probably years away. It is. So I'm glad that he didn't go around So I'm glad that he didn't go around So I'm glad that he didn't go around just um talking about Oh yeah, our next just um talking about Oh yeah, our next just um talking about Oh yeah, our next big thing is quantum computing. Yeah, I big thing is quantum computing. Yeah, I big thing is quantum computing. Yeah, I mean, you know, that's kind of like, mean, you know, that's kind of like, mean, you know, that's kind of like, yeah, I think his GPUs plus yeah, I think his GPUs plus yeah, I think his GPUs plus state-of-the-art AI is going to help state-of-the-art AI is going to help state-of-the-art AI is going to help accelerate our ability to get quantum to accelerate our ability to get quantum to accelerate our ability to get quantum to actually work and be viable and and make actually work and be viable and and make actually work and be viable and and make it mainstream rather than being a it mainstream rather than being a it mainstream rather than being a science project. science project. science project. Yeah. Because it it'll it'll you know, Yeah. Because it it'll it'll you know, Yeah. Because it it'll it'll you know, gosh, you know, I I listened to some gosh, you know, I I listened to some gosh, you know, I I listened to some podcast, I think I sent it to you. It podcast, I think I sent it to you. It podcast, I think I sent it to you. It was just an hour long. It was some guy was just an hour long. It was some guy was just an hour long. It was some guy who had left OpenAI. He was disgruntled who had left OpenAI. He was disgruntled who had left OpenAI. He was disgruntled and he he was like he thinks AGI is like and he he was like he thinks AGI is like and he he was like he thinks AGI is like two years away. Um he was pretty freaked two years away. Um he was pretty freaked two years away. Um he was pretty freaked out actually. It was pretty scary you out actually. It was pretty scary you out actually. It was pretty scary you know. Um in fact he was funny. Who was know. Um in fact he was funny. Who was know. Um in fact he was funny. Who was it? Uh Chaitton Chararma was like what it? Uh Chaitton Chararma was like what it? Uh Chaitton Chararma was like what is the deal? Why is it all the smartest is the deal? Why is it all the smartest is the deal? Why is it all the smartest people on this subject are the ones people on this subject are the ones people on this subject are the ones telling the most dystopian scary telling the most dystopian scary telling the most dystopian scary stories? And then you have these other stories? And then you have these other stories? And then you have these other guys who say, "We're about to emerge guys who say, "We're about to emerge guys who say, "We're about to emerge into a world of abundance and it's going into a world of abundance and it's going into a world of abundance and it's going to be magical for everybody." And it's to be magical for everybody." And it's to be magical for everybody." And it's like, which is it? Is it going to be like, which is it? Is it going to be like, which is it? Is it going to be abundance or is everybody going to be abundance or is everybody going to be abundance or is everybody going to be out on the street?

out on the street? out on the street? Well, we talked about this last week, Well, we talked about this last week, Well, we talked about this last week, right? It's it it you know both the right? It's it it you know both the right? It's it it you know both the dystopian and the utopian dystopian and the utopian dystopian and the utopian uh uh uh pictures both look dystopian. There really is no such thing dystopian. There really is no such thing as abundance, right? as abundance, right? as abundance, right? There's how can there be unless you There's how can there be unless you There's how can there be unless you combine that with depopulation and then combine that with depopulation and then combine that with depopulation and then there's less people, less Thanos. Yeah. there's less people, less Thanos. Yeah. there's less people, less Thanos. Yeah. you you flipped the script on economic you you flipped the script on economic you you flipped the script on economic uh on you know economic principles of uh on you know economic principles of uh on you know economic principles of humanity. You know you now less is less humanity. You know you now less is less humanity. You know you now less is less is more less is better. You know that's is more less is better. You know that's is more less is better. You know that's not how we live right now. It's like not how we live right now. It's like not how we live right now. It's like more more more more right. We're either more more more more right. We're either more more more more right. We're either going to have to take care of the going to have to take care of the going to have to take care of the population to pacify them. We're either population to pacify them. We're either population to pacify them. We're either going to have to build our matrix or going to have to build our matrix or going to have to build our matrix or some version of Logan's run. some version of Logan's run. some version of Logan's run. Yeah. It's going to be something like Yeah. It's going to be something like Yeah. It's going to be something like that. Well, yeah, that that was the that. Well, yeah, that that was the that. Well, yeah, that that was the other creepy thing is um not I mean I other creepy thing is um not I mean I other creepy thing is um not I mean I can understand they always use a can understand they always use a can understand they always use a feel-good feel-good feel-good or possibly feelgood story to to sell or possibly feelgood story to to sell or possibly feelgood story to to sell the merits of these types of the merits of these types of the merits of these types of technologies. But um they're using technologies. But um they're using technologies. But um they're using crisper now to treat genetic diseases, crisper now to treat genetic diseases, crisper now to treat genetic diseases, right? They were sh how oh yeah you know right? They were sh how oh yeah you know right? They were sh how oh yeah you know the the the young lady had a rare kidney disease and young lady had a rare kidney disease and young lady had a rare kidney disease and you know transplant didn't work so

you know transplant didn't work so you know transplant didn't work so they're now administering crisper to see they're now administering crisper to see they're now administering crisper to see if they can change at a code and it if they can change at a code and it if they can change at a code and it looks like I'm going to have to leave looks like I'm going to have to leave looks like I'm going to have to leave here. Okay. here. Okay. here. Okay. Yeah, on that note, um, you're right. Yeah, on that note, um, you're right. Yeah, on that note, um, you're right. And then also Khan, And then also Khan, And then also Khan, right? right? right? Conra, what what's his name? Sing Jen Conra, what what's his name? Sing Jen Conra, what what's his name? Sing Jen Khan. What's his Oh, I don't know. First Khan. What's his Oh, I don't know. First Khan. What's his Oh, I don't know. First name. His name isn't just Khan. He name. His name isn't just Khan. He name. His name isn't just Khan. He actually has a pretty long name. But actually has a pretty long name. But actually has a pretty long name. But anyways, maybe thing we have to be anyways, maybe thing we have to be anyways, maybe thing we have to be concerned about is people genetic concerned about is people genetic concerned about is people genetic engineering themselves into super engineering themselves into super engineering themselves into super soldiers and superhumans. Exactly. soldiers and superhumans. Exactly. soldiers and superhumans. Exactly. That's where we're going to have That's where we're going to have That's where we're going to have xenomorphs on the planet Earth. Yeah. xenomorphs on the planet Earth. Yeah. xenomorphs on the planet Earth. Yeah. Yeah. Exactly. Which, by the way, I saw Yeah. Exactly. Which, by the way, I saw Yeah. Exactly. Which, by the way, I saw that movie Romulus. It was pretty good. that movie Romulus. It was pretty good. that movie Romulus. It was pretty good. It was good, wasn't it? I almost um It was good, wasn't it? I almost um It was good, wasn't it? I almost um shorted my pants. Pretty freaky ending, shorted my pants. Pretty freaky ending, shorted my pants. Pretty freaky ending, huh? huh? huh? Didn't see that coming. Yeah. Yeah.

Didn't see that coming. Yeah. Yeah. Didn't see that coming. Yeah. Yeah. Okay. Well, everybody have a wonderful Okay. Well, everybody have a wonderful Okay. Well, everybody have a wonderful Memorial Day weekend everyone. Three Memorial Day weekend everyone. Three Memorial Day weekend everyone. Three days. Yes. Take care everyone and days. Yes. Take care everyone and days. Yes. Take care everyone and remember um to subscribe, like, and remember um to subscribe, like, and remember um to subscribe, like, and share. www.iotalk.com and donate to Elevate www.iotalk.com and donate to Elevate Kids at Kids at Kids at www.elevatorids.com. And we'll see you www.elevatorids.com. And we'll see you www.elevatorids.com. And we'll see you next week. I Happy Memorial Day. Bye next week. I Happy Memorial Day. Bye next week. I Happy Memorial Day. Bye y'all. Be much love. Love you. Heat. y'all. Be much love. Love you. Heat. y'all. Be much love. Love you. Heat. Heat. Heat. Heat. [Music]