SPEAKER_01 0:53

You know, that's a like it's up. Yeah. Yeah, there's a part that that it's like a squeaky thing that goes. I think that's how it goes. Yeah. Yeah, there's a lot of hard parts.

SPEAKER_02 1:19

Wow. Tough stuff, man.

SPEAKER_01 1:23

Yeah. Anyway. Welcome to IoT Coffee Talking, Bad Karaoke.

SPEAKER_02 1:30

Yes. Yes. You know, but remember it's guitar karaoke. You don't hear any of us singing, do you?

unknown 1:37

No, I know.

SPEAKER_02 1:37

Every once in a while. Every once in a while. And it's not air guitar, it's real guitar. This guy could go on the road. This guy is actually he's on the road. He's pretending to be an analyst and just carves out a little bit of his time for us once a week. You know, sometimes he's at home, sometimes it's in a hotel, because you know, life on the road. Um, dun dun dun dun dun dun dun dun dun dun I'm on my way. Anyway, yeah, there you go.

SPEAKER_01 2:09

Yeah. Um, but so yeah, it's pretty cool. Yeah, you gotta do something. You know what I'm saying? I mean the years and years being uh like 20 years in consulting, traveling around the world, and you know, literally the entire week is in a different city, gotta do something. Yeah, yeah. And so I used to take my guitar around and play all over the place. So that's awesome. And you know what's a great way to meet people.

SPEAKER_02 2:43

I was about to say the same thing. I think having your guitar and uh it's great to meet people, it's great to create relationships, like especially when you're consulting and you've got customers, and all of a sudden, because you know what, you need to have a relationship on a different level than the business thing. You really do.

SPEAKER_01 2:58

Yeah, yeah. You know, although uh a lot of my clients never came out to see me except for a couple, and yeah, and um, yeah, and then just recently when I was at RSAC, you know, I did that thing on stage, you know, the band. And um it's pretty cool. Uh no, the we uh what was that one tune that we played that was really cool? Oh yeah, we played ACDC Hell's Belt. Awesome. And um that was really good. I I don't know if anyone recorded it. I should have recorded it, but yeah, we we we we killed it. The the band was awesome, you know, the house band. Um and that's so cool. Yeah, now now I got like Vinke, like he called me up and said, Hey, can you play at Things Conference? I'm like, okay, dude, that would be awesome. Dude, I'm yes, no, so I I'm I I'm considering it. Just bring your guitar, man. But I but I told him I was like, dude, you know this is not my day job, right?

SPEAKER_02 4:07

But maybe uh yeah in the long tale of life. And I mean I I think that I think it makes a lot of sense to go to the things conference there with your guitar. Are they gonna do you have to bring your own amp?

SPEAKER_01 4:28

I'll I'll have something. I'll probably take just two little thing. Tiny amp.

SPEAKER_02 4:34

You guys go to them? Like, so David, you know, Vinke is saying that Leonard, you know, for their things conference, he should bring his guitar and rock out at the conference in Amsterdam.

SPEAKER_00 4:45

I I personally think that's a great idea. Yeah, I do too.

SPEAKER_01 4:50

I think uh huh. Are you coming? I'm thinking about it, yeah. Uh I just have to make sure that it doesn't confli uh conflict with a bunch of client obligations and and uh events. Yeah, I mean those I can't weasel my way.

SPEAKER_02 5:06

So that's like that event's always like is it in September?

SPEAKER_01 5:09

September, but it used to be um earlier. It used to be in the summer, and I think that was a better time.

SPEAKER_02 5:15

Oh, right, because we get we we hit it conference season, you know, kids back in here in September.

SPEAKER_01 5:23

Oh my god, dude. I have uh I have companies in uh that are booking my time, trying to book my time in November. Wow. What month is this? April. I don't know anymore. It's April April. Which is smart, which is smart because this is a collision. And um yeah, it it it and thank god for these virtual uh um these virtual events, right? Like uh Google Cloud. I just you can watch the live stream. And they usually have great more content than you would ever be able to absorb, even if you were live. And you know, what I do is I just go and I take notes, I have questions, I just shoot it off to the AR people and then see whether or not they'll even answer it because sometimes I ask shit that they that scares the crap out of them or whatever, and they don't want to answer, and they you know, yeah, but I'm gonna text Pete.

SPEAKER_00 6:27

I'm I might go to the uh edge AI thing in London because that's close to me. Yeah, yeah. You should. You should. That's a good idea. You can just compete plus one. I'm not I'm not paying 300 pounds.

SPEAKER_02 6:39

No, no, no, no. He'll give you the special code.

SPEAKER_00 6:43

Yeah.

SPEAKER_02 6:43

Ah, okay.

unknown 6:44

Yeah.

SPEAKER_00 6:45

Otherwise, no go.

SPEAKER_02 6:46

Special code, invite to the good dinners, you know. And luckily, you're right, you're so close you can hop on a like a $49 Southwest Airlines flight to London, right?

SPEAKER_00 6:57

Uh EC Jet, London Luton.

SPEAKER_02 7:01

I love when they always say Clucent London Luton, and you get on the stupid oh my god, the ride the bus or get the trah, right? Ride the train.

SPEAKER_00 7:13

The dart is the shuttle from the airport to the train, which is the TAMS link, and then you take the Tams link into the city.

SPEAKER_02 7:22

Right, right. Yeah, I've done that a couple times. I was just like, okay, close in looting.

SPEAKER_01 7:28

I freaking love London, man. I freaking love London. I was tempted, you know, I just can't go to all these things, you know. It's just just too much. Um yeah, all these all these people are going, oh, are you gonna hang out in Las Vegas and you know, stay for Google? Um, you know, I was at Nab, I was at the NAB show, which was uh really good. Um lighter attendance than last year and previous years, but there's always great content, they have great sessions.

SPEAKER_02 8:02

And um you know, for our audience that has no idea what NAB is, broadcast. Yeah, broadcast.

SPEAKER_00 8:11

You guys are not even recording.

SPEAKER_02 8:13

Yeah, we are. You are? Yeah, yeah, he is. He started. I know.

SPEAKER_00 8:17

How come I have an option to record?

SPEAKER_02 8:20

That's weird. Oh, dude. Never never never mind recording. Because it did pop up, because you're right. I'm usually the recording person who goes, hey Leonard, push record.

SPEAKER_00 8:30

Oh man, I am glad I noticed and I didn't reveal the already revealed um um real life uh Satoshi Bitcoin guy, Adam Beck. Thank you, New York Times.

SPEAKER_02 8:44

I saw something about that on the news. I I didn't watch this what what there's some headlines about that.

SPEAKER_00 8:50

So the recent uh so two recent things. Uh there was uh an HBO documentary uh looking into the real identity of uh Satoshi. I cannot pronounce the last name, but you know, Nakamoto, yeah, that guy, the person credited with creating uh the Bitcoin. Uh he's the one that published original 9 or 10 uh, you know, uh technical uh 9010 page technical paper on it, etc. And on the back of that documentary, a reporter from the New York Times did like a whole in-depth research, and the way he thinks he has the right person is he was matching the writing style of the initial 9 to 10 page uh technical white paper, which coincided when Bitcoin came out, the person that wrote it, Satoshi, disappeared, and then this other guy started writing. So it's kind of like a Batman Bruce Wayne sort of thing, you know, at the same time. But they have the same writing style. Uh, I mean, old school investigative journalists, uh, old school uh, you know, um forensic reading, investigation, but forensic writing, super interesting. And they did a whole podcast, and of course the guy denies that he is asatoshi. I think he's got two big reasons. Number one, you don't want people to know that you potentially have 70 billion dollars worth of uh Bitcoin. And number two, the guy has a regular job and he's trying to take this company, which is Bitcoin related, public. And the SEC, when the SEC works, I'll say nothing more about that, uh, they would want to know that an executive that's going uh public with a company this closest if they have any anything else tied to this company. So anyway, very very interesting, um, good list. That is.

SPEAKER_02 10:50

That is. Yeah, I mean, I saw I saw some interview briefly the two guys that I think were the reporters on this. Hey um and but also someone said they thought whoever is Satoshi is dead. So maybe it's different people. Because they're like, why hasn't this person done anything with the billions that would be sitting in his wallet?

SPEAKER_00 11:15

I um I don't want to well, I mean, you know, I know that IoT Coffee Talk really has the majority of podcasts listening globally, and more people listen to us than the New York Times. So I'm gonna give you the answer, Rob. Uh apparently uh the Satoshi character said that um it will be found out through uh inheritance. If and when he dies, right, things will pass on. And then the community, the the you know, the folks that hold 20% of Bitcoin, like uh Bill, who just decided to join us, by the way, uh they are waiting. The evidence that they think will verify whether this is the guy or not, is one, has any um monetary value move from the original wallet? And two, did the uh in well, and in order for the money to move from the wallet, the cryptographic keys that match the wallet needs to be used. So that is how the community thinks if it's A Toshi, these two things have to happen. And if those two things don't happen, is somebody else, is Bill, Bill only has 10 billion in Bitcoin, he's broke. Right.

SPEAKER_02 12:28

And so like one day some kid just gets billions of dollars showing up. It's like, woohoo! Yeah, inheritance. That's how we find out.

SPEAKER_01 12:39

Well, I think the problem is people think it's one person, and it wasn't a completely engineered thing. Um, you know, it it's origins, its origins are timed very closely with uh some of the sanctions that were happening. And think about its purpose. How it's being used right now. I mean, you know, I've already made a note of this that uh when you look at the volumes that are being transacted to through this thing, it outstrips the actual value, the market value of any of these cryptos. So it's not being used as a store of value, it's being used for money laundering. And then so when we have like the situations that we have around the world, the tensions, the sanctions, it's the perfect vehicle to undermine any of those regimes. And why do you think it's one person? You know, this is a you know incredibly well-engineered mechanism for undermining the the dollar regime. It is, and uh, and you know, probably the reason why BRICS doesn't have their own uh crypto, or maybe even China doesn't exert themselves in being a reverse currency is hey, they have this thing, and so they'll just use that to transfer funds. And if you look at what's happening in some of these conflicts, they're either paying in crypto, they're asking to be paid in crypto, right? For tolls and whatever. And you know, how do you think all these weapons that are being deployed everywhere, you know, whether it's uh drones, are being funded? I mean, it's like we have sanctions everywhere, but how are they holding? And it it seems that the funding for a lot of these um um arsenals out there uh that shouldn't technically shouldn't be there because you're economically constricting certain parties. How come they don't work? You know, where where are these getting all these?

SPEAKER_00 14:58

Because we are uh because we're humans, and if we I haven't done this, I'm gonna do this, I'm gonna give myself some homework. I want to look at what is widely known as the first global scale sanction against a country, just whatever that is. And immediately the next thing I'm going to look at is how people went around it. Because when you're talking about, you know, it's supply and demand, right? All of a sudden, it's a problem to get things from here. I mean, I think, you know, Rob and I have talked about uh moving medicines, not counterfeit, moving legit drugs from country A to country B, just because of the tax situation, you will make a killing. Why it if we're able to move something, whether it's money through crypto, uh it's it's going to be done. Um I just think it's funny that we were so worried about the impact on the grid from Bitcoin mining, and that seems like so 2008. Yeah. Yeah, yeah. And now it's like, no, that's not who ended up being the energy hog.

SPEAKER_01 16:08

Um but the problem there was really the question of um the transaction cost. Is do you need to burn that many uh do you need that much energy in order to burn that many tokens? Which is ironic because with the AI we're talking about tokens as well. And I have to give uh credit for that one dude in uh in Hong Kong, or is it Singapore? Uh, I forgot his last name. This was a you know, Rob, this is back when you were at IoT uh solutions world congress, and they had the big coin. Remember when they had the big the blockchain?

SPEAKER_02 16:50

It was the most it was the most popular thing there. We're there for this other event, and somebody had this whole conf mini conference on how you learn how to do an initial coin offering. Yeah, and no, it was it shows human nature. Yeah, people are like, uh screw that other stuff.

SPEAKER_01 17:10

I'm gonna learn how to do an ICO. But in 2017, that thing was packed, and then there was a gentleman there who had that that robot, you know. He made news all over the world because I forgot Ava, I think it was the name of the the robot. It's like the sexy-looking, you know, chatbot robot, whatever animatronic thingy. So everybody was tripping out, and that guy actually uh made the comment that AI and blockchain and this whole idea of token minting are tied together. And you know what? He's actually ironically right. And um, and it's weird how GPUs and NVIDIA kind of have this lineage um through this whole uh series of um you know uh hype cycles, right? Because you know, um NVIDIA was big on crypto mining, and then about three years, two years ago, two years ago, they completely dropped crypto mining because everyone wants to everybody went to A6, they already figured that out, kind of like what's happening now with AI. People are realizing uh GPUs aren't all that great. Um, and we're seeing that. I mean, look at Intel. Now everybody's talking about CPUs, right? Wow, surprise, surprise.

SPEAKER_00 18:30

It's cool again, man. It's cool again.

SPEAKER_02 18:33

Well, can someone tell me? Can someone speaking of Intel? Can someone tell me that how just a few months ago, Intel, you know, they dumped uh Pat Gelsinger because for some reason he wasn't getting it done, and then they bring in Lip Bhutan, and he just adopts the same plan as Gelsinger, basically. Yeah, and their stock was $19, and they were just like and then Lip Butan has to go have a meeting in the White House, and then all of a sudden, okay, I love you now. And then Intel, no change to the economics, no change to anything. Their stock is like $80, whatever dollars now. What the hell? Nothing has changed.

SPEAKER_00 19:16

I don't want to hear anything about the market. Yeah, anything that we thought we knew that has been happening for the last 40 years, all of it, it it's irrelevant. I mean, all birds. For those of you that don't know, all birds was a brand. They were making, they were making shoes, they were making apparel, this really light shoe, a shoe company, okay? They make shoes. And uh they were burning through money. I think their valuation at some point was like three billion. They sold less than a month ago for like 39 million with an M, right? So and within 10 days, they make an announcement that Alberts, the company I just said, was going to make shoes, they're going into AI. They didn't explain what that meant. Yeah, and all of a sudden the stock went from like a buck twelve to like exactly.

SPEAKER_01 20:11

And yeah, yeah. I mean, but see, that's why I, you know, um I I like my role as a technology analyst and an industry analyst because I don't have to deal with that. And you know, the ground truth, the markets are even more detached now from ground truth than before. A lot of these guys that you listen to on CNBC, a lot of the Wall Street analysts have no idea what they're talking about. There's a small number, there are few, you know, like Stacey Raskin, he he actually gets things, right? I mean, because he has a semi-background, you know, he's worked in fabs and stuff like that. So he he gets it. But the problem is this if you were a semi-analyst, you really knew semiconductors, and the focus was on the chip. You when you're covering companies that are in the semi-industry, you didn't really go upstream too much. You didn't really have to understand the dynamics of the systems and and markets as much as. And that was always really a question for a lot of the Wall Street analysts, right? Even for industry analysts, a lot of industry analysts. I mean, look at like with telco, everyone's trying to make a shift to AI because all of a sudden AI is a big thing, but these analysts have no background in AI, right? And so they're trying to find that thread that can where they can attach uh uh you know themselves and their legacy to something newer, and it's usually the networking, right? Um, but that the you know, these folks don't have a grounding on this stuff, and now these so called experts they're on a learning curve, most of Right.

SPEAKER_02 22:01

You know, I was listening to CNBC this morning and they said that true north was pivoting to AI today.

SPEAKER_01 22:08

And I was like, yeah, yeah. And then here's the other problem. You can't just think about the chip anymore. Now you have to think about software, you have to think about systems, you have to think about applications, you have to. These are things that, you know, a lot of like this the guys that are in the lower levels of the stack, they don't they don't cover this stuff.

SPEAKER_04 22:34

But I mean, but wasn't that wasn't that the case when we started talking digital transformation? Weren't you supposed to be thinking all those things at that point?

SPEAKER_01 22:42

That was like, you know, academic fluffy nonsense. Except for those that are. And then all the technical guys were going, what the hell are you talking about? Digital what?

SPEAKER_04 22:53

Except for those that actually do it.

SPEAKER_01 22:55

Yeah. The gap of the ugly. People who actually implemented stuff.

SPEAKER_04 23:00

Like uh I'm just a businessman giving the people what they want.

SPEAKER_01 23:04

Yeah.

SPEAKER_02 23:05

Yeah. Which is tough. Right. Right.

SPEAKER_00 23:09

It's you know, the I I'll leave with this. Uh and um why are you leaving? You can't. No, I have to take uh one of the kids to activity. So it's it's either that or uh or I get beat up by the kids. But you know, with a whole uh AI doomsayers, and you know, AI is doing this and AI is eliminating jobs, which it's not, it's just people they just want better portfolios, but that that's for another episode. But uh Nikesh from uh Palo Alto, he um I I I caught him the other day on a post where he is saying that they are there are software asset services that have been and are going to be crushed by AI because you just don't you can just do it yourself. And that is out there. And but that on the other no, hold on. I I no no we didn't say anything.

SPEAKER_01 24:05

Keep going.

SPEAKER_00 24:06

We're just uh in in the middle, uh then he put uh you know a lot of uh analytics that still requires uh human input and human analysis that is helped by AI, so not replaced. And then of course it's a little biased, you know, conveniently. The last bit, he he put all of the uh SaaS for cybersecurity uh on the last box where that one is going to be really accelerated by AI. Whether he's right or wrong, that that's not that's not my my beef. I just thought, I just hadn't seen it that way. Because it's kind of like the SaaS umbrella, and in his view, he's like, you guys are host, you guys are gonna be okay, and then this bit is accelerating. And uh for me, it was a refreshing take, because usually what I get is all the job eliminations, you know, like the um, what was it this week? You've told between Meta and you know, Microsoft. It's uh different take.

SPEAKER_01 25:11

They're not they're not uh, yeah.

SPEAKER_02 25:13

Um so we've heard about the SaaS apocalypse or whatever, and I remember a lot of the SaaS guys stock went down a month ago or whatever it was. You know, but I have heard the take, and I understand the take of AI is gonna make it so easy for you to make any software that you could ever go buy off the shelf that you'll just do it yourself. And it may be true-ish that you could do that. Who will actually follow through and do that will be a tiny subset of it's just like us with ideas or startups or anything. 10,000 people have the same cool idea that I have, and only one or two people are gonna follow through and act on it.

SPEAKER_00 25:56

Well, and so yeah. I just want to I I just want to be in the same room when Mr. Founder with Vibe coded and shipped something. I just want to be there when the first patch has to go. That's it, that's it. And and let's see what actually happens.

SPEAKER_04 26:12

Yeah. We don't care about the patch, I just want to be there when it scales beyond the damn virtual environment. How about that? Because the minute that it the minute that it breaks, I mean, this is the weird part about it, and this is what people don't get about the whole bob coding or I'm using AI to co-develop with me and everything like that, is that when you ask for a test sequence, okay, people challenge, challenge yourselves on this. When you look at the test harness that you're using to test A, B, C, and D, and you go, it passes. Why? Because AI says, Oh, this is A, and it's so perfect. Here's the test for it. Oh, it passes, see, but you're not you're not running it against random edge cases, you're not running it against something that is decoupled from the actual code.

SPEAKER_01 27:02

That and that's the big problem. It's the the You can't leave.

SPEAKER_00 27:06

You can't leave. I'm sorry, I miss you guys. Peace out.

SPEAKER_01 27:10

Yeah. All right. There hey, good seeing you. There's so many edge cases, and many of the edge cases can be catastrophic. That the volume of what you should test is impossible to test, and what you can test, the known knowns, if you will, right? Those those aren't the things that you have to worry about. And so with like all the Segentic stuff, nobody knows how to freaking test this stuff. And if you listen in on um Google, cloud, uh what the freaking hell is there? Conference. Next. Next when you look at all the security stuff, and even with Wiz, it's all like reactive kind of things. It it's like like you mentioned audit, right? The YAML audits and stuff like that. This is all like, you know, non-proactive things in the runtime environment, right? This is all like reactive. Um and it's also just security after you just push stuff out. So debugging production problems. What about debugging things before you put things out as part of your, let's say, AI development lifecycle? AI LDC, right? Um, or uh DLC. Um there are some approaches, like you know, um, I had a chat with this uh Cisco guys, Rajopra. He mentioned to me, and I I wasn't aware of this, that they have a thing called uh code uh code defense, right? And then you know, Wiz uh Wiz, the company that uh Google acquired, they also have tooling for doing um some code testing, but it's for traditional code, it's not for testing agents and you know, AI agents and AI itself. You know what I'm saying? And so people are getting really, really misguided on a lot of this stuff. So a lot of the test harnesses and tooling is really for traditional code, but not for the agency stuff, which which is really, really tough. And you know, uh I I made that post, right? Um uh uh Bill, and you commented on it about uh Nemo Claw and um Defense Claw, and what you're starting off with, with all the sandboxing for these, uh for open open claw. You're you're basically starting with uh here, a completely agency deprived agent. It's like now go and figure out what you're gonna provision. And then at that point, how is that different from a fucking script, man? You know what I'm saying?

SPEAKER_04 30:02

Because you now you have to do all this, like it's it's the reverse of coding, you have to decode the agent so it doesn't go off and and this is the whole thing where when I hear people say, Well, AI can now write code better than X, and I'm like, Yeah, but you know what happens is people fail to go through those thousand or fifteen hundred lines of code that were just written, yeah, and take ownership of that. Where are you committing? And how are you how are you governing this? Who's signing it? Who is saying that I own this so that it's defensible? Yeah, and that's the whole reason that I'm like, I'm like, do we use and remember? I said this, we use AI, we use Claude to do things. You're not committing any of that code until you've gone through it and you said it and clearly delineate it. Was AI assisted, and here's I own it. Because when you get to ask as to why did you use this over that, you better be able to fucking answer.

SPEAKER_01 31:08

Yeah, yeah, and then um um, yeah, you're so right. And if you're vibe coding, you're just like a casual, uh, you're not even a a software developer, you're not a systems integrator, you don't know the environment, you don't know the APIs. Forget about, you know, and forget about A to A and MCP and all this other shit. It's that stuff is not magic the way that people think it is, right? In fact, it's worse because it adds a semantic layer between um a requesting entity and a consuming entity, right? Which creates a lot of um reliability issues, right? Um so the the problem is is being a uh how do you have the understanding of all the systems that you uh might have to provide permissions to to account for them? Right? When you go and you create your policies, let's say that you have like something like uh um I mean shit, some of the stuff that Google announced, enterprise agentic platform or whatever the friggin' hell it's called. How do you go and you uh and responsibly and safely uh configure the policies? And that was one of the things that Raj and I talked about, and he brought up in the little chat that we had. It's a really good one where he he mentions how um yeah, it no one's going to be configuring millions and billions of um policies for each and every value. And then you know what? Yeah, but how are you gonna eat but you know he suggested intent, but intent is really, really hard to implement in any kind of management and orchestration uh layer. I mean, you know, Leonard. Yeah, yeah.

SPEAKER_02 33:06

I just remember when I was washing dishes at Denny's a long time ago, and it this thing popped in my head through the bubbles I saw in the sink, and it was it said to me, You need to turn electrons into tokens. And you need to convince the entire planet that that's the thing. Oh, yeah, yeah. You guys need to watch the video or listen to this podcast. You may have been hearing about it that happened with yeah, DoorCash Patel with Jensen. It kind of got hard to watch, a little uncomfortable, because man, he like he was like putting it to Jensen.

SPEAKER_01 33:45

Jensen started to I thought it was the other way around.

SPEAKER_02 33:48

I well, it can I mean he's it's because he was kind of basically going, how do we know this whole thing you've got going in NVIDIA is not just a big house of cards?

SPEAKER_03 33:57

Yeah.

SPEAKER_02 33:58

It's like, you know, the only moat you really have. You're trying to say your moat is this or it's CUDA or it's whatever. He goes, No, your moat is you've locked up the entire supply chain for all the pieces and components to make this real. That's your moat. Um, there were just times where all of a sudden Jensen just kind of kind of would hesitate or whatever, you know, because he's so salesman all the time. He's so, you know, he's a smart guy.

SPEAKER_01 34:22

Yeah. I mean, he um he typically doesn't, in those scenarios, doesn't answer your question.

SPEAKER_02 34:29

Um it's like, why can't anyone else turn electrons into tokens? Or you know, what's so special?

SPEAKER_01 34:40

And uh uh uh along the lines of dark cash, and I didn't get an answer. But I don't want to create drama like he did. And uh but anyway. Yes, you do. Yes, you do. Uh no, I just people want you to stir the pot. I tried to trade him this thing for a jacket, and we we have a we do kind of have a verbal contract of sort, but uh I haven't really pressed him on that.

SPEAKER_02 35:09

But um but I think it's not beyond the realm of I think it's not a stretch to say that it may not be the total moat, but a moat that NVIDIA has is they have locked up the entire supply chain for all the pieces needed to make all the shit. Yeah, because they have the money to lock it up, just like Apple locks up the supply chain.

SPEAKER_04 35:33

You know, stock share, your shares up, that's a legit mode. I mean, whether it's whether it's one that you like is different, but it is legit. If I'm the one that controls the supply chain, he who has the board puts the pieces where you want them.

SPEAKER_02 35:51

Yeah, yeah, true. You're right. You're right. It's just like when Apple bought up all those Toshiba one gig hard drives that are this big to create the iPod, you know? Yeah, when jobs like, I want to buy everyone that you're gonna make, all of them. I'm buying them all now.

SPEAKER_04 36:10

Yeah, no, I mean now now you're you're now you're creating that difference, that separation between you and the next. Yeah, yeah. You know, but I mean, but here's the here's the thing, and I'm gonna give this one to for free to all of the startups out there that are like, you know what, I'm lean, I'm gonna use AI to be this and to be that and to be that. Have a plan to get off of it because AI is a horrible drug.

SPEAKER_01 36:40

Okay, Chappelle. Okay, Rick Games.

SPEAKER_04 36:47

AI is a horrible drug. It is a hell of a drug.

SPEAKER_02 36:51

You know what? Actually, in the future, people are gonna go back to people are gonna go back to this episode when he said that because people are like when in the future, when it turns out you got all these AI junkies that's like their um heroin, yeah, and they can't live, they can't speak, they can't think for themselves. They're hooked so deeply.

SPEAKER_03 37:10

Yeah, I think that's right.

SPEAKER_04 37:15

Hey, hey, fuck your couch. AI is AI is a hell of a drug.

SPEAKER_01 37:26

That's the title of that's gonna be the title of this episode. Yeah, it's a hell of a drug.

SPEAKER_03 37:32

It's true.

SPEAKER_01 37:33

It is no no no, you're you're right in so many ways. I mean, I have a family member. She um yeah, she she she calls her uh chat GPT, um whatever the fuck. Um she gave it a name? Yeah, hubby. Oh my what the no no no people it's a it's like a freaking drug. And for people who are either you know, it's like it the whole question of AGI ASI, it's relative. If you think it's smarter than you are, you think that it's great, guess what? It who's gonna argue, right?

SPEAKER_02 38:11

You know what? For I'd say a giant percentage of the global population, it's way smarter than they are.

SPEAKER_01 38:17

Yeah. Way smarter, right? Right. And and so, I mean, you know, hey, so kudos to Sam Altman and Dario and all these other guys. You delivered AI and ASI and AGI, and I think they know it. Uh and GSI and GSI, whatever. All those but also at an emotional level as well, right? So there's intelligence and emotion. And so that drug part, the dopamine hit, right? It's for people who who need it, right? And there are a lot of lonely people out there.

SPEAKER_04 38:56

Lots of lonely people. The first one is free. Yeah. That's right.

SPEAKER_02 39:03

Absolutely. Oh, I just created a cool image. I just made a video. I just did wow, this is awesome. Oh, wait, I gotta pay 20 bucks a month. Oh, well, yeah, okay, maybe I will. Oh, actually, now I'm gonna give you all my money and I'm gonna turn to a life of crime so that I can feed my habit of AI.

SPEAKER_01 39:22

Yeah, or you know, I I get laid off by all these companies who think that uh AI, well, they tell everyone that AI is taking my job, right? And has replaced me when they're doing that so that they can they can um you know fund their AI age engines or whatever, you know.

SPEAKER_02 39:42

Yeah, yeah.

SPEAKER_01 39:43

That's that's scary stuff, man.

SPEAKER_02 39:46

It keeps happening. It's more it's more acute for us because we have so many friends who all work at these companies. And so while the while analysts are going, no, actually, we're not really seeing layoffs happen because of AI and the you know unemployment rates holding steady and we're good. But we just happen to know all the people who are all being laid off, and the new Microsoft thing where it's like, well, we're gonna do this cool retirement thing for you. And if you've got to why you out.

SPEAKER_04 40:15

If you if you close your eyes, did you actually see the accident? No, I mean you heard that it happened, but did you actually see it? So you weren't in the farm. You're not wrong when they're saying we're not seeing these things, it's just one of those things, don't tell me about it and just go do it.

SPEAKER_01 40:31

Yeah, I mean, you know, just if you've ever had a you know that's funny when you said the CD accident.

SPEAKER_02 40:38

It reminded me of like uh something important I learned from a Stephen Segal movie when they said, Yeah, so and so he's dead, and they go, Did you see the body? Well, no, but I did you see the body? Yeah, he ain't dead. He ain't dead, he's coming for you. He's gonna bring some a keto in, and it's gonna be bad news. Oh my god. Such good times. Wow, yeah, yeah. On my little, you know, different groups with Microsoft. I'm already seeing people talking about this weird calculation that they have to, like how long they've worked there, how old they are. It's like some bonkers bizarro thing if they qualify, you know, for to be one of the 8,000 and something that could do this retirement thing. Um, and if you're in sales, you don't qualify, you know, if you're an engineer or whatever, and you know, and it's really I'm seeing all these weird numbers. Oh, my number is 80. My number came out to 64. I don't even know what they're talking about. I know there's some equation that Satya has given to the employees. Maybe it's the new, they have a new HR head. They have a new HR head, and maybe she said, hey, instead of just doing mass layoffs, what if we do it this way where it feels like they have, oh my gosh, this is like the matrix, the architect in the matrix. The whole problem was the problem of choice, freedom of choice. And the only way they would accept the program is that they believed they were making a free choice, and that's what this is at Microsoft. All right, we just figured it out. They're all living in the matrix.

SPEAKER_01 42:17

Yeah. We're we're living in the matrix right now in really weird metaphorical form. Um yeah, that's really friggin' bad. Um how come we can't be uh optimistic and bright and cheery and I'm I'm optimistic as hell. It's just that these people keep funneling into the you know, like um I'm all for safe AI. It's just I know that it's hard to get there. And for vendors, you know, start focusing on that because you're not gonna, you're not gonna the reason why a lot of people are not selling this stuff is because it's not safe, you know, and the mentality is so twisted. With agentic AI, the last thing you want to do is take the MVP Silicon Valley mentality and use that to guide how you go to market, how you position your products. You don't want to do that. You know, because you will, you know, going back to your comment about being a startup, Bill, if that's how you approach this stuff, you will never make it because at some point you you have to be trustworthy, right? Especially where you're a small company um and you're trying to actually build a business, right? God forbid, that company is not going to trust you unless you can come to the table with some proof points that you can deliver safely and a reliable product. And you're not just experimenting on them, right? And this is why you get into POCs because you get a lot of these people going to companies saying, hey, try. Trust us, we have a solution. So that they'll they'll uh the company or this uh prospect that you're approaching will say, Well, okay, why don't you do a POC for us? Oh, by the way, that's your invest. We're not gonna pay for it. Prove to us, yeah. And then you get that POC hell, which is happening really right now with um a you know, AI for the enterprise versus AI, uh the consumer, this is a hell of a drug.

SPEAKER_04 44:32

But but I mean, but check check this out, right? I mean, we we've heard we've heard the term before, and everybody kind of throws it around. FUD, right?

SPEAKER_03 44:39

Yeah.

SPEAKER_04 44:40

If you think that we're not living in that space right now for the companies that are looking at things, for the companies that you're like, oh, you're not using AI, then you're missing the boat. Fear. If we don't adopt some of this AI, then we're gonna be in trouble. Uncertainty. If we don't have it embedded in our solutions, then we will not be able to deliver. Doubt. I don't think that we can achieve our financial goals if we don't do all of those other things. It's there right now, but the the makers of this, yes, why say it's a hell of a drug. The makers of this are doing it in such a fashion, people are like, it's changing. I don't know how many times I saw Claude get updated three times in one day. The pace at which these updates are happening are driven to outpace the policies and all of these other standards.

SPEAKER_02 45:36

Yeah, really, right?

SPEAKER_04 45:38

Because if I stay ahead of it, you can't catch me.

SPEAKER_02 45:41

Yeah, I see what you're saying. You're like, if I built these guardrails based on this version, and then the new version breaks the guardrails or whatever. Yeah. Okay.

SPEAKER_01 45:52

Yeah, and you know, it it's funny because the we're not dealing, uh, you know, Omnia came out with uh a report uh that the the models are not getting bigger. Remember like early on, it was like the race to big, big, big. Yeah, yeah. AGI, that you need bigger, bigger, bigger models, right? Well, that the the size of the models themselves, the LLM, actually probably peaked at around five, you know, actually pretty small. Um, probably around 70 billion. Um I probably I'm just guessing. Okay. 70 billion. And then what they did is they went to MoE, which is the mixture of experts, which are smaller, uh domain-specific um models that are then now being orchestrated by sort of this master um you know uh model, right? Right. That's not really a singular um model. It's kind of like uh you're applying a harness around this thing to extend or create the illusion that um a number of expert uh models can seem smarter, right? Now they're just extending this model, but the thing is you're hitting limits. And so nobody is actually getting past like 1.5 trillion, which begs the question why do we need all these massive GPU, quote unquote GPU systems now, right? Uh for training. Uh it doesn't make things faster. There's bottlenecks everywhere. Uh there's bottlenecks everywhere, and there's a limit to how much scaling compute um uh you know this way, scale out and scale across, really gets you in terms of benefit.

SPEAKER_02 47:49

So this is like some of the stuff that um you know, Rob I don't want to totally I don't want to totally bash GPUs, and I don't want to bash NVIDIA. It's so much faster than a CPU. It's not even in the same purpose. I've seen it work.

SPEAKER_04 48:04

I saw the technology work, but but I'm gonna call I'm gonna call something out.

SPEAKER_01 48:08

CPU is hot for a different reason.

SPEAKER_04 48:11

First and foremost, the the uh Leonard, to your point, the models getting large and everything like that. All of a sudden, somebody switched over and said, Hey, let's quantize this down.

SPEAKER_01 48:21

Yeah, that's where the innovation's happening. Yeah, innovation, let's reduce the floating point, yeah. So it's not as accurate.

SPEAKER_04 48:29

Now, the next thing is Rob to your point about the GPU. All right, GPU is great. I'm sorry, but on my end, as far as the twins go, the GPU for us only helps if we're using computer vision. If we're using computer vision, then it helps us, particularly from an edge perspective, because we'll build smaller models that sit at the edge, typically on something like an Oran, uh, which is a Jetson product, right? And and we will use the GPU then to help um us, you know, they got these other tools from Cosmos and all these other things for inference that then I can tie to the video clips that are being put into the twin. Now, there's an issue going on here that when you when you look at that. I want you guys to look up NVIDIA PVA. NVIDIA PVA is a DSP and it sits in the right, yes. It it does it does the same thing.

SPEAKER_03 49:37

Yeah, yeah, yeah.

SPEAKER_02 49:38

Oh, so wait, you mean Texas Instruments could be the leader instead? No, because they make that shit.

SPEAKER_01 49:45

No, no, hey, hey, no, no, no, it's not a joke. See that that's what I wrote about at CES, right? CES this year and last year. If you look at Cadence, you look at um, you know, uh, you look at um Ambarella, you look a lot at a lot of these companies that make um edge IP, right? Um, and what I mean by that is core IP. So whether it's DSP or MP or whatever, yeah, DSPs are everywhere uh for highly efficient um ML workloads. And increasingly you see these novel architectures to bring some of the generative stuff, whether it's like vision models, some of the multimodal uh capabilities for let's say scene detection or scene understanding. That that's happening there, but it's not happening on a GPU. And yeah, I I read about that, Bill, and I was kind of like, what the friggin' hell? Nvidia is using edge applicant.

SPEAKER_04 50:47

Yeah, it's it's 100% because we we're not giving, I don't care who we work with, we're not giving up the fact that we're edge-based. Now, I'm gonna hit you guys with something that I think is just totally awesome. I um I stumbled stumbled on on this young lady by the name of Twila Jackson, and she is uh the founder of Helixis Technology. I want you guys to go look look this up. She dropped a what 120-page document that's called the deployability thesis. Most, I mean, I read it literally, I read it, and I was like, oh my god. I I mean, we're trying to figure out a way to work with her because the stuff that she identifies, the stuff that she highlights and everything, I'm like, that's a that's literally, it's not a it's not just a document that you look at. It's literally a set, it could become its own thing. It's a set of deployable metrics, things that you should incorporate into your design by nature, by the simple fact that when you look at the uh autonomy, right? When you start talking about autonomous vehicles and all of these other things, all this stuff that we're talking about is pushing out is some level of autonomy. Yeah, and the thesis that she has develops an argument uh in clear progression. I promise that. And and I'm gonna come back and tell you guys more because we're we're gonna be working with her. Uh, and I just think that's cool. The larger world should hear about this young lady.

SPEAKER_02 52:26

That's awesome. Uh I love it.

SPEAKER_01 52:29

Anyways, wow, got hot at the end. Um, new hot new stuff dropping in your lap, you know.

SPEAKER_04 52:35

Hey, hey, it's hell of a drug. Like I said, hey, it's this is Fafo Friday. Yeah, and I'm just a businessman trying to give the people what they want. Exactly.

SPEAKER_01 52:46

In the gap of the ugly, yeah.

SPEAKER_02 52:49

In the gap of the ugly, man. That's where it happens. Did you guys watch the draft last night? Hell yeah. Oh, yeah. Oh, yeah. I I I was sitting there thinking, you got a good guy. Um, yeah. Two of them. Two of them. Yep. Yep. Oh man. And Mr. Raiders up here. Obviously, you got draft pick number one, Mr. Mendoza. That was not a surprise.

SPEAKER_03 53:12

No.

SPEAKER_02 53:13

Um, I feel I feel bad for him though, man, because there's nothing I hate worse than what a great player young player goes to the awful team like the Raiders are so bad right now.

SPEAKER_04 53:23

And has to land on the bigger, I mean, it's like it's it's like it's like uh it's like he he got, you know, his hands were tied, his feet were tied, and they were just throwing him into the deep end. It's like, now let's see you swim.

SPEAKER_02 53:35

Yeah. Yeah. It's just because I've watched the Raiders last few years. I think they get a new coach every year, and they, you know, they keep trying. And it's like putting it if you put in a good quarterback, it's not enough. They just they need more of a surrounding.

SPEAKER_04 53:51

He doesn't have somebody to throw it to.

SPEAKER_02 53:52

Yeah. Yeah, exactly. Exactly.

SPEAKER_04 53:56

And he's got to be upright, he's got to be upright long enough to throw it. There's that.

SPEAKER_02 54:02

He's probably gonna be looking up at the sky a lot. Exactly. Yeah, you know, but don't get me started. Don't get all hyped or pumped about the cowboys because when I see their GM, who is also their owner in there, it's like, man, they gotta get that guy out. This is old news, obviously. This is nothing new. They need to get that guy.

SPEAKER_04 54:23

Jerry, Jerry Jones is Al Davis Jr. What are you talking about?

SPEAKER_02 54:26

He is Al Davis Jr., you know, but at least Al Davis, yeah, whatever. Shit, you know.

SPEAKER_04 54:30

I mean, at least Al Davis had the courtesy to just die.

SPEAKER_02 54:33

Yes, yes. Jerry needs to have that courtesy because that team is never gonna he's the crypt keeper. He won't go anywhere. They're not gonna have any success until that guy's gone, and he can't be the GM like he is. He's he makes too many decisions.

SPEAKER_04 54:48

It it's gonna be the same thing. Like AI. Al Davis passed away, his son took over. Jerry's gonna pass away, his son's gonna take over. We're gonna get more of the same. Ouch. Yeah.

SPEAKER_02 55:00

I don't mind if his son takes over as long as he just stays back up in the owner's box and doesn't talk to anyone.

SPEAKER_04 55:07

Here's here's the I mean, I the the the one thing about it is I'm I'm always gonna be a Cowboys fan. I love it when we reach this part of the year because I get to go, all right, it's zero zero for everybody. Then I get to come to then I get to come to Christmas and go, all right, we're fucked now.

SPEAKER_02 55:27

Uh every year it's the same thing for Cowboys. This is gonna be the year. This is it. We're gonna do it. We got the always always the same.

SPEAKER_04 55:38

I get, you know, this is our year. Look, we got two last time we had two draft picks in the first round was uh DeMarcus Ware and Marcus Spears, and and and we went to the Super Bowl. But do I think we're gonna do that this year? Uh no. I think I think we are right now the way that we're set up. I do believe that we're the best in the NFC East.

SPEAKER_02 56:01

You think you can win 10 games?

SPEAKER_04 56:04

Yeah. Okay. As long as as long as George Pickens actually plays, he signed his uh he signed uh the uh the the franchise tag. So if he pay, if he plays, the offense was lights out last year. Dak was slinging the ball all over the they kept us in the game despite uh despite uh paraplegic defense.

SPEAKER_02 56:28

Exactly. Right, right. So hopefully you're getting that fixed up with these draft picks.

SPEAKER_04 56:32

Well, we got we got a new D coordinator. We got uh you know we got rid of some guys, we picked up some guys. Yeah I'm I'm thinking, I'm thinking right now, I'm thinking 12 to 13 games.

SPEAKER_02 56:45

That's a lot of wins.

SPEAKER_04 56:47

Wow, yeah, wow, yeah. Now, the trick is to get out of the first round. They struggle with that. Every time. They really struggle, you know, but we won't talk about we won't talk about you know January and whatnot. Let's let's let's live in the now.

SPEAKER_02 57:06

Yeah. All right. It's important, it's important to talk about football and IoT copy talk because it's the most important thing in the world.

SPEAKER_01 57:15

Well, we are in sports center, so this is uh it is sports center.

SPEAKER_04 57:20

And the fact that we have a since we're since we're there, my Lakers. Oh LeBron. LeBron is looking like he got some re-invigorated spirit.

SPEAKER_02 57:33

Uh he looks invigorated, and KD is not nope for the Rockets. So you guys have too much time on your hands.

SPEAKER_01 57:42

That's all I gotta say.

SPEAKER_02 57:45

Yeah, yeah. Right by the way, yeah, yeah. Um you needed you needed LeBron to step up. Yeah, yeah, for sure. You're right. Absolutely.

SPEAKER_04 57:55

Yeah, he's looking good. All right, guys. I mean, I know when I was 41, I couldn't do what he's doing. No, no way.

SPEAKER_02 58:02

Yeah, totally. Yeah, and that was just like a couple of years ago for me. Yeah, exactly. Exactly. All right, thanks everyone for joining us on IoT Coffee Talk. We've talked about a lot of stuff. I've already forgotten what it was. Whatever Mr. Vasquez said, just go with that.

SPEAKER_01 58:21

Yeah, it is a hell of a drug, and crypto is uh, you know, the guy you're gonna find out who started it uh through inheritance disclosure.

SPEAKER_02 58:32

Exactly. Exactly. I'm glad we're getting all that figured out. This is really important stuff. We will see you next week. I know you learned a lot of great stuff here. You're empowered, you're empowered by people instead of empowered by some agent. So that's a positive. That's good. We have accountability. If we said something that's not true, you can come reach out to us personally, as opposed to just some autonomous agent who has no liability, no accountability, no nothing. All right, all right.

SPEAKER_01 59:00

Seriously, at your own risk.

SPEAKER_02 59:03

And don't listen to us about financial disclosures or things like that or stock. This is for entertainment purposes only. And just say no. And just say no. Just say no to drugs. Yeah. Wow. Peace out. Adios. Peace out.