AI at Scale: The $1T Chip Surge, Cybersecurity, and Mars | TSG Ep. 1015

Artificial intelligence is reshaping everything from the silicon, powering our systems to the defenses, protecting our digital world, and now to how we explore entirely new frontiers. This is the next phase of AI at scale. Hey everyone, it's Alan Shimel, and welcome to Monday Techron Gang Live, coming at you noon on, uh, east Coast time, nine o'clock for our friends out in California, a little later in the afternoon, maybe for our friends across the pond in Europe.

Um, I hope this new time works for you, but you know, quite frankly, like everything else we do here at Tech Trunk, it's a bit of an experiment. If we get, you know, feedback that this isn't the right thing to do, we won't do it right. We'll go back to the morning show.

I kinda like the morning show anyway, but let's see, let's see how it goes. We've got our, our lunchtime gang here assembled right to talk to us today. Um, actually this is a, this is a pretty cool gang.

This might be the first time we have half, half us and half Canadians, but you know, if you saw the big Bunny show last night, we're all Americans, right? Whether we live in North America or South America or Central America, we're Americans in Viva America. So let me introduce you to our gang, Chris Blask, Garima Bal The Garima is your first time on this year, right?

Yes. Welcome back, I hope. I know you've been busy and traveling and doing things, and of course we have the dean up in New York.

Mike Vizard. Hello. Dang.

Welcome. Welcome, and happy Monday. Um, Mike, you know, I, I don't think you need to be a rocket science to figure out that if every single wafer of silicon that gets produced has a home, the semiconductor market's probably gonna be pretty healthy.

But, you know, that's what we're leading off with today. A captain obvious said, well, But there are issues as always. There are always issues.

So the Security Industry Association thinks that they're gonna pass $1 trillion in sales for new processors, or mainly a lot of it is AI related, and a lot of it is also the memory that goes into those systems. And that's all good for the chip industry at least. 'cause at least in this goal rush there, the folks making the blue jeans and the picks and the shovels, and selling that and making a lot of money.

However, demand is far exceeding supply, and they are talking about new classes of processors. And, um, it's more of a marketing term than an actual name these days. But two nanometers are here and people are working on one nanometer processors.

And these things are really not so much smaller as much as they are just 3D designs where the things are stacked on top of each other, but they're more complex to build and more costly to build. You can assume as a result of that. So the question becomes, will we ever get to the point where we have enough processors, or is demand always gonna exceed supply?

And the cost of AI is always gonna be well, relatively high. Alan, what do you think? Well, no, no, and no short answers.

But, um, here's the thing. Obviously, look, we're in a market right now for semis, 175% of supply. We have a market for.

So prices are high. We're also at the beginning of the evolution of this market in terms of what are the, the semis we're going to need going forward? Are we going to need the big Nvidia GPUs as our computing workloads, especially around ai?

And, and these AI factories move more towards inference from, from training, right? Uh, but it, it's not, you know, even like memory chips, right? Even good old fashioned Dr.

Memory chips, you can't, you can't buy them. The entire world's production of them is basically sold already pre-sold, right? Um, so, you know, these are things that need to be figured out.

The other thing is, let's face it, for the highest level chips, whether we're talking the Nvidia GPUs or the Qualcomm, you know, there aren't that many fabs that make these chips at that level. In fact, there's one company, basically, right? TSMC, and then the folks in the, in the Netherlands who do the etching, have the etching technology.

When you have that sort of stranglehold, when you have that sort of, you know, supplier, uh, formula, what do you think they're going to do? Just say, oh, we're gonna make it more and cheaper and bring down our revenue. No, probably not.

Especially when one company uses that as a, as acru to make sure that someone will help them if their neighbors across the sea and, uh, invade them. I'm not talking about Greenland, right? So this is, you know, are we gonna hit a $1 trillion milestone this year?

Again, pretty obvious to me. I, I don't think, I don't know if I'd call it news. Well, I was talking to a couple ISVs about this particular issue, and they feel like they're already in a bind and they're, part of their issue is, well, they kind of, the end users have kind of been convinced that AI is free or cheap, and they're kind of looking at all these costs that they're absorbing to run the ai, and they're basically saying, Hey, you know, this doesn't look good for our business model.

'cause I can't increase the price of my subscriptions by a hundred percent, and yet I don't know how to not do ai. So all these companies are like feeling some significant margin pressure as a result of all this agreement. You're nodding your head, but I don't know if you're hearing the same things.

Yeah, I think, uh, you are spot on, Mike, because what I see from where I stand is that this is like a portfolio pivot, which we need to do because the market is flattish. You know, there is no return of investment on all these kind of AI centric, uh, you know, um, applications and the, the, the hype created around it, right? So what is happening in turn is that we have to ensure that we have 30 to 40% of, uh, CapEx flow into the infrastructure space, right?

With this chips and all that. So what interns is calling for is that we have to look at where we have to downsize, and there are repercussions, right? The repercussions is that from the portfolio pivot perspective, there's downsizing, there is layoffs, there is cost cutting, there are like margin pressures.

So I think there is a lot other things which are happening due to this kind of cap eccentricity for chips, and of course the supply chain and, uh, rewiring and everything else needs to be done. But right now, what I'm seeing is that there's, uh, like a substantial amount of pressure to, um, uh, have a lot of, uh, cap eccentricity and see where the cash flows from that perspective. Yeah.

Chris, you were playing around with AI a lot. Are you seeing these cost issues or what? Well, again, a lot of the work we do is, is focused on sovereign ai, on, on local systems and so forth, and, and in, in small canons and small data sets and so forth, you don't need a lot of GPUs.

Now, we're just, you know, hunky dory all the time. But on this topic, I can think of it, you know, I it's in a lot of ways, like Alan, you're saying, um, and Mimi IT oil, uh, uh, uh, chips are, chips are kinda the new oil. It is becoming commoditized, and yeah, we have fragile supply chains, and that has to be solved.

But I think about Gwen Shotwell at, at SpaceX, right? You know, the, the, if anybody's noticed this, this last week, SpaceX got, uh, approval to launch a million AI satellites in, in 24 hour, uh, uh, solar orbits. And, uh, and this is, you know, so, uh, Gwen had a, a company called Orbits Edge in 2020, you know, when yeah, during the early, before Starling sort of launched g when SpaceX is going really well.

Um, she was really excited about that, right? And that was sort of what fascinated me at the time. So now that now it's started, you know, and any any space geeks like me, you know, the Starship platform is coming along this year.

Next year, look out of the next decade, they will launch a million AI satellites, pack for chips. What does that mean? Well, to this topic, yeah, supply chain of chips is going where it's going, and it's, and we've talked about this a lot on this, on this show, week to week over the last year, you know, how they get used, where, what kinds of trips and so forth.

I mean, again, there's better experts than me on, on that, but the amount of processing we will do in the next 10 years, the next 20 years, you know, whether it is, you know, a a million, uh, uh, satellite constellation and space and or massive data centers and or, you know, inference and everything else being done on all sorts of small little devices in the ground. Um, there's a lot of chips there. Yeah.

But I agree what Chris, you are saying, but I also recognize, and I want to pick, uh, Ellen's brain on this, that, you know, I see that we are in a cycle of AI monetization lag, which means that we are debt funded for all the chips and the, all the hype which is created, uh, and the investment on the AI data centers. What do you think, uh, Ellen, from your side or your perspective, how, how things change this year and next year? So I, I do think we're in the middle of the training to inference change, which is gonna be big because it, it takes Nvidia out of the only game in town seat, right?

You'll have the Amazon chips and Google chips, which I, I think are made by Broadcom. You have Broadcom itself making a lot of chips. Microsoft's making chips, uh, arm, I think is doing something with.

So you are going to see more suppliers and more suppliers is a good thing, right? So unfortunately, we're already in this year. So if you're saying this year to next year, maybe 26 to 27, well, maybe by 27 that the, the mar, you know, it's not as tight a market.

I don't think the market goes down at Chris's chip. We're gonna need more chips than, uh, to Chris's point, we're gonna need more chips than ever, but I think we'll have more suppliers. And so it won't be as tight as it is now where you just gotta, you know, this is like, you know, trying to buy the hottest toy in Christmas for your kids, and you, you didn't go out early enough.

But to put Mike to your point, you know, this is, what do you expect from an industry that says, here, take the first one. It's free until they get you hooked, right? This is kinda outta big tobacco.

And, and now, and now, um, well, now you want to keep using it. Oh, you gotta pay for this, this, you know, I, I can't make it up on volume. I think, Garima, I want to ask you this.

I think all this puts a significant amount of pressure on the developers. 'cause I think that the metric that they're gonna be judged against is how many tokens are they processing to deliver this functionality? And that means they gotta write better and smarter software.

So I think this all comes back to the hit the developer's heart, and it's gonna be, you know, you, you need a lot of talent to write this correctly. Exactly. And this is cyclic pressure, right?

I mean, you have the chips, you have the infrastructure, but you don't have the developers because we are in that kind of a tight, uh, uh, curve of, uh, revenue and return of investment. So now we have to force to use our developers to go in the direction of seeking support from AI ss tools capabilities. And in turn, what it means is that, uh, you know, you also have to look at, you know, how you lower the cost pressure from, you know, not hiring or not, like stabilizing your hiring a little bit, because developers were scars from the beginning.

So it's a cyclic pressure which is being built up at the end of the day. I think it all boils down to how all this monetizes in the longer run, because we, it this cannot continue, uh, if we do not have enough, you know, significant, substantial evidences that these are the revenue drivers. So I am yet to see like where the balances, right?

And as you rightly pointed out, uh, you know, we will also see developer scarcity in the longer run because people are also using or even moving towards the AI system tools. This it true. So when you think about it, and you start with the chips hack, it can even work your way all the way down.

And the, the metals used to build the chips, and then the chips and the systems and the data centers and the developers, the entire AI supply chain is construed, and it appears it won't be such the case for at least a year, year and a half. That's true. Yes, Absolutely.

Mm-hmm. I I think it'll be more than that even, because you don't know what's next. And, and, and you know, it's one thing if you're just talking about these cutting edge AI tips, which we we're focusing on, but guys, we even have capacity problems with just Dr.

Regular memory chips that, you know, they're kinda running the mill. Um, you know, it makes, it makes me wonder why, shoot, if Intel can't thrive on that foundry business model in this environment, who can? I mean, I, you know, you are selling hamburgers to starving people in the desert here.

If you're a foundry, think about it, right? How I just put up a sign that says, I can make your chips. You want more chips?

Come see me, Mr. Chips. Yeah.

And to your point, Alan, I think not all segments are all sectors or all industries will benefit equally from this, right? So the winners will write the history. So who are the people who will extract revenue out of it would be the people who will drive innovation and who will drive the market strategy.

And again, don't, uh, also forget about geopolitics here, right? So things, you know, there are certain aspects and different kind of dimensions we have to look at it from. Got it, guy.

I don't disagree, but look, it is, but let, let's step back. The semi market's a trillion dollar market. That's huge.

It, it is some, it's a, a red letter, right? This Is high margin business. Yes, it's, and we shouldn't lose sight of that, but we gotta hop into our next SEC section here.

Um, cybersecurity in the age of ai, we've seen craziness over the last two, three weeks. Chris, I don't know about you, but it, it, it keeps my blood pump in reading about all this crazy stuff. Mike, what do we got?

I don't know if it keeps my blood pumping as much as maybe it just terrorizes me. So, um, the folks with the opus language model have come up with, uh, I guess they ran it against all the open source software looking, not All, not all, just some, just Some. Anyway, they found 600 vulnerabilities in a matter of couple of days.

And, you know, this is something that human researchers would've taken years to go do. And now, at the same time, we've also got reports now where somebody stole somebody's credentials for AWS and then used AI to reverse engineer their way into that system from there and basically hacked in in eight minutes. It seems like Chris, to me, that the bad guys are gonna be able to discover and exploit vulnerabilities in minutes, hours now.

And are we able to respond to that? Or is there just gonna be havoc? Yeah.

Is It Yes to both? You know, and again, one of the reasons I love this show, because we've been doing this week to week, and we've talked about this, you know, a month ago, three months ago, six months ago. And for me, this maps all the way back 30 so years, because we will get two points when, and we are demonstrably at this point when for a lot of reasons, and it's not just the speed of ai, you know, the working supply chain security over the last, you know, 6, 7, 10, 12 years.

You know, we get to the point where we now have this level of complexity. And look, look at what we just talked about in the last segment. So fly a million satellites packed with GPUs and tell me where the software's, you know, bill of materials is.

It's a matter of scale. And I, I think that, I don't know, and we keep in my entire career, we keep working through this, you know, because if we had, if we didn't have something with the capabilities of large language models and everything, you know, we call AI these days. Um, we would have other problems with security.

You know, the bad guys that wouldn't have 'em with the good guys wouldn't have 'em either. And you have to have them just to have coherence at the size and scale of cybersecurity writ large today at all. So, you know, the usual, you know, in security and conflict, you know, the, the, the, the underdog always has the first move, right?

You know, the people who are not constrained by rules have the first move. Yes. And, and if Alan you'd mentioned, you know, phishing, you know, you, you got, uh, passed by, uh, a, a phishing, uh, thing four or five, six months ago.

Phishing protections are dead. I get such awesome phishing emails right There, go Watching you on the text pro and gang, I really like your work. No, you don't.

You're a bot, but you just the research that fast. So, uh, you know, Mike, to your question, yes, you know, my advice to, uh, particularly small operators, small industrials and so forth in the last six months is go get a chat GBT account or something for 20 bucks a month. Say, tell her who you are.

Say look out on the internet and find what Chris Blask and Fred Cohen and Gene Sp and all these old fools who have said about cybersecurity, and help me compare. What do I have? Because I'll tell you, for all the products and services that getting experts to come in and do any assessment, even poorly of organizations to give them advice as opposed to selling them products, that's been the big gap.

AI with all its faults right now, does a damn fine job of that. So if you're, you know, long answer. But if you're a small organization, use some ai, ask it some intelligent questions.

Pretend it's a consultant, it'll probably give you really good pointers. If you're a bigger company, use it properly. You should be able to tighten up your networks to make them quite hard to, to, to mess with.

And it shouldn't take more time than you have before they take you down because they will otherwise. So I, I, here's the thing though, guys. 6 found 600 vulnerabilities and in, in some very well known open source projects in and of itself is a scary proposition.

But Chris, did you not, we all suspected that these things have lots of vulnerabilities. Do you poke if you poke 'em enough? We just didn't have enough pokers, Right?

And now we do. So if you're the Developer now, we do Use it yourself before your adversaries do. Absolutely.

Now. And, but here, 'cause that's the other point of this, you could say, well, don't look. Yeah, I think I forgot what actually, I know what president it was.

A president said, well, you know, don't, don't try so hard to find things or, you know, something along those lines. Uh, and, uh, you know, don't, don't test people for that's what it was. Don't test people for COVID.

So our COVID numbers aren't so bad. Don't run these things. So we don't find these vulnerabilities.

Why do you think the bad guys aren't running them? Do you think? They're not.

We're the only ones in town who get to run a ludicrous. So this is, you know, Gotti, Gotti ever, who I talked about in this article that I wrote and Heather from Google, they said six months ago that we are about to come to this cataclysmic thing where the amount of of vulnerabilities we're gonna find is gonna overwhelm our ability to patch them six months to the day Opus comes out. And we get this.

But here's the good news. I wrote another article that just came out this morning. It's called, uh, what the heck is it called?

Son of Not Claw Bot. What, what's the c Mal moba son of MBO versus Opus. Kinda like, remember the old Japanese Martha versus Godzilla and all that kind of stuff, son of Godzilla.

So I was riding my bicycle this mo this weekend. I do some of my best thinking when I'm riding my bike and I'm riding down a one a thinking about the world, me ai. 'cause this is what I, this is what my life is.

Uh, and I came up with the greatest idea. You got this mal bot, which has its own security issues, right? Terrible.

I mean, this thing had no security built in, but what is it? It it does, it's good at going viral. It's good at getting a lot of these things working.

The only way to beat or to keep up with AI found vulnerabilities is to use AI to fix 'em. So why can't we have m bot an army of mal bots, you know, autonomously take the Claude Opus output and go, go fix on these open source projects, put, you know, put patches in or make 'em available, or, you know, write them and have them have them out there so that people use them better than that. Why don't we build Opus and Mbot into our platform engineering platforms?

So that boom, now as part of my platform, I actually do, I used to do a SAS scan and a DAS scan and a vulnerability scan and a pen test. Well, now let's do an opus test too. Find as many of these vulnerabilities as I can as part of my platform, as, as new software gets added to the platform, it gets an Opus scan and then have a, you know, think of them as T cells in your immune system have, have an army of mo Botts at the standby ready to rapid deploy to go fix these vulnerabilities that your opus is finding your opus scan found.

Look, this is, I'm, I'm ready to go start a new security company and do this, right? Yeah. This, this is where we are.

Chris, go ahead. 8, uh, amount. And in short, yes, you know, you know, people could and should probably do that, particularly in the open source library world, and that's a good thing.

But more importantly, yeah. And it's not about mal boden and, and CLA op particularly. It's doing things a certain way where you take these and because these products will change and so forth.

Mal, uh, I love mal book. I think it's a train wreck, but it's a fascinating train wreck. And it's, it's a lot of the, the dynamics are accurate and a lot of the flaws are, are instructive.

But if you ground these things, you know, like I said, for the open source world, you know, have at it, everybody, you know, I think it'll work out fine as an organization. Take these tools, put 'em inside your own organization, point 'em at your code, at your infrastructure, whatever it is. And it, it's, it's, it won't be about security.

You'll save money, right? Yes. You'll find a bunch of flaws and so forth, and you'll fix those, and you'll find out that you're losing a lot of money because of the disconnects across the right.

And that's in the late nineties, firewalls stop being about security. And that's when it took off. I think we're, we're here on this as, as well.

It's not about cybersecurity. You'll get better security, but if you do this correctly with your own code base as a consumer or as a, you know, software, you know, producing organization, you'll just get way more efficient, Which means saving money. Yes.

So you add that one. Good, Mike. So Garima, will developers spend more time building patches and or do you think that it'll be the same amount of time, they'll just build them faster?

6 has done, and two things which are evidently different this time. And they are the strength areas of this whole, uh, you know, approach is the reasoning part. If you see how, uh, Opus four six has approached it is as a programmer's intent, right?

To see, uh, from where these, uh, vulnerabilities have originated from checking the Git histories, for example, code patterns, uh, generating precise input sequences for complex bots. And that's where, uh, the first win starts. Like the reasoning approach is quite different and powerful.

The second thing, which I would say is performance. Now this outperforms human teams, as Alan mentioned. So definitely this has been a winner, because of course, um, uh, if you detect the, these new vulnerabilities at scale at this, uh, speed, I think this is great.

But the trick is in the balancing act, right? I mean, we cannot just say that it's the best security researcher, which we have found, right? And of course, as Ellen also mentioned, that it is susceptible to how, uh, you know, misuse of these detection and, you know, techniques would be done in the longer run.

So how do we ensure that we do not fall trap off these, you know, uh, opportunities and of course, see this opportunity in the space where we can, um, also redesign our workflows, right? And to ensure that we have, uh, linear to agent kind of, uh, you know, roadmaps and how do we induce or introduce these kind of new capabilities into the developer ecosystem. So to answer your question, Mike, I think there's a lot of upskilling needed to be done.

Uh, nothing overrides human judgment, uh, for sure. At least not now, right? And then we also have to ensure that we understand the key strength areas of this powerful technique, and also expose our teams to use them in a proper way.

Fair enough? Fair enough. Guys, I'm going to jump to the next section if it's okay, though.

But yeah, I like the MBO versus the Opus kind of thing, you know, little Japanese movie kind of stomping on buildings, and at the last second, getting together to do what's right for the planet. But let's, let's, you know, speaking on sci-fi, let's go to space. You, you like horror movies too, I take it then, right?

I never liked the horror movies, but I did like the Japanese, Godzilla, Martha, Rodan, Gira, you know, we have a new three-headed monster today, Gemini, Claude, and I don't know, perplexity. But anyway, let's talk about space, the final frontier. Yeah.

Well, let's talk about something fun here. It's, uh, the folks at NASA have figured out that they can use AI to help better navigate the Mars rover because they can figure out where the bad terrain is and maybe avoid it, and maybe not wind up in a pothole or something. But gima, you know, what's your take on this?

'cause, you know, I'm excited about this for one reason. It's like a use case that I have not seen. And it's not just somebody kind of, you know, doing something task a little bit faster.

It's like something entirely new and different, at least it is to me. Exactly. So I mean, if you trace back, uh, this robot was launched in 2020, right?

And, uh, uh, in 2021, approximately touchdown. The core objective of this was, I think, uh, was this micro biological, you know, signatures, which they, uh, they were kind of trying to find. And of course, uh, returning with soil use, et cetera, et cetera.

So now what was the problem with, uh, uh, with the rover was that there was a 20 minute, uh, one way, uh, a lag between Earth and Mars radio, um, uh, communication delays, right? And what has happened, and this is again, a very interesting and powerful is, um, um, the AI implementation has, uh, ensured that they have a digital twin. And through that digital twin, they have been able to lay out the TRA trajectory.

Also, I think, uh, this is again, uh, uh, breakthrough from a communication and planning perspective, because there was a lot of limits associated with how, um, real time corrections on the trajectory could be done for the role, for example. And, uh, we could get rid of a lot of mobility constraints with this implementation. And how this whole implementation has been like processed is that, uh, of course, cloud A is kind of, uh, having the data sets which human use, but also it has enhanced and, um, uh, projected some of the features and generated some kind of continuous safe paths for the row, right?

And it remarkably improves the communication as well as the efficiency of the rover. Now, why I, I see this as an opportunities, uh, commercial AI in the space, uh, aerospace area, right? If we can fine tune, let's say, through the domain data, I think it will be great.

We can also do some kind of integration and edge cases for robustness, for space contract, for example. You will start to see a lot of these kind of, uh, you know, uh, RFPs, which we'll talk about this ML navigation tools, which can also boost, uh, you know, some kind of market opportunity for all firms, which can include AI developers. So I think it's, uh, uh, it's kind of exciting, but again, it's also challenging because, uh, it has bias and risk associated with it.

And it's not a proven technology yet. It's just the first breakthrough. So we'll have to watch out for how this whole thing's, you know, um, develops.

Agreed, you know, a couple of weeks ago, I, I did a, uh, shimmy says on what, what are all these AI jobs going to be anyway, all these new jobs that AI's going to create? And I mentioned that, you know, one of the most exciting areas was space, right? Where we could start putting, you wanna call 'em satellites, you wanna call 'em floating or orbiting data centers, permanent stations, lunar, maybe on Mars, ma mining asteroids, right?

These are things, these are in the industries of the, of this last half of this century, perhaps, right? And into the next. And, and AI is gonna enable it.

'cause you need AI to do these things, but there will be human jobs that co go along with it. I completely agree with you, Ellen, because I think, uh, in from the beginning, um, at least I see two potential industries getting benefited out of this AI movement space being one of them, the exploration into space, and, you know, a lot of these things which were not humanly possible. If we could utilize AI technologies for that, that would be great.

You know, and the second industry, which will be quite benefited, and I'm not like, you know, no guesses here. It's health science and, you know, drug industry, because that is the, the second industry segment I feel that will be highly benefited out of this AI movement. And having said that, I mean, uh, it, uh, brings to, uh, us to a point where, uh, responsible sourcing becomes very, very important.

You know, even if for space technologist, we need to ensure that we source, uh, you know, responsibility and ethical Well, yeah. As an Apollo era space geek kid, right? You know, I've got a, you gotta reflect on this topic, right?

You know, it's, I had a Curiosity Rover prototype wheel at my house for, for a number of years. It had jpl L on on the, there's a, there's a hack in there. The ones they actually shipped have JPL and Morse code.

'cause NASA wouldn't let 'em do it. So this topic, right? And, and since we're talking about rovers, you know, you guys remember the Opportunity Rover that its last message was, you know, batteries are low and it's getting dark, and the whole world, you know, cried because, oh, this little rover with before ai, but, uh, agree to your point, right?

You know, this, this is where this is going. Autonomy, the dollars involved. Look, you know, Elon Musk is totally right about this one, right?

These next 20 years, we're gonna colonize Mars the moon and so forth. What does that mean? We're going to have all sorts of autonomous stuff that's going to generate a lot of information locally.

The, the, the economics just demand it. And as we, we move further into that period, it's not just sourcing ethically here, it's sourcing, you know, manufacturing on, you know, non, uh, non tarn bodies will become a thing. So what we're putting together now is the logistics that needs to operate in that environment.

And yet we are, we're humans. We always make those, this mistake, we say the last 10 years of my life, hand in even folks like us who've lived through this rapid acceleration of change. But if anybody knows a way to do that in 10 years and 20 years, they know 30 years whenever it is from here, we're, we're building that out on Mars without massive au autonomy.

And I think we get the good parts of, of, of sci-fi. The risk of the bad parts are certainly there, but I think a lot of the practical realities and, and, and, and current systems are driving down this path where the systems need to be reliable enough. Who's controlling 'em mean?

Is it all centralized? Huge concerns about that. But the systems, they, themselves, I think they need to evolve and, and, you know, the environment is what it is.

But here, here's the problem, here's the problem. And this is something I think AI can help us with. You know, I said it last week, and one of the things, one of the segments, you know, when John Kennedy announced the Apollo mission, right?

We choose to go to the moon. Not because it's easy, because it's hard. We knew it was gonna be hard, but, and we knew we were gonna have to invent technologies, but we made a commitment.

And leadership made that commitment, and you made the commitment and you did it. Now, we did that to win a race against our rivals. We didn't do it to make a lot of money or to we commercialize things, though it did wind up throwing off a lot of commercial products.

This one's a little different. We've gotta make the commitment to industrialized space, to commerce size, if that's a word, commerce size space. Because we know that the money is there for it.

It is a viable, uh, mar a a viable source of wealth, right? And whether that's done via government or via private industry or combinations, or via world, well now, not now, it's not a great time for the world to do anything together, it seems. But maybe this is something that brings the world together, right?

And maybe with having, knowing that we have AI at our disposal to do this, we'll be more, uh, more likely to make that kind of commitment. But 'cause to do this, to do what we're talking about, it's a commitment. I think it's, it's bigger than any one Elon Musk can do.

You need more than SpaceX? I'm sorry, Joe, Chris El. Yeah.

You know, any, you know, an audience like this is probably some Isaac, Isaac Arthur fans as well. You know, look at, go back and forth o over the present to the predictable futures over the next decades, you know, century or two. What does it look like?

These capabilities will come along. The question is just when and how. And, and, but the, the ground-based, ground-based, uh, uh, telescopes, you know, come to mind because yeah, with, with the politics of the world, a lot of friends of mine are big into that and complaining about starlink and all these satellites out there.

And I have to ex help them understand that China's la launching thousands too. This, if you, if you're a sixties kid, you've been following, you know, sci-fi and science, this is the era when there's a lot of stuff starts to orbit the earth. It just is good, bad, otherwise, I, it is, this is where we're going on this issue with AI and forget about Mars so much, but the moon is right there.

We are gonna put a lot of gear on that, on that body soon. It's all being built now. It's being launched.

Now how does that play out? And again, if there's other ways then what we're calling AI and these sort of systems and so forth to do that efficiently, that, that, you know, drive the economics of it that I'm, I don't know what they are. So yeah.

I also see, uh, another side of this, uh, opportunity is that making it more participative and inclusive. Why I say so is that if you think about simulators or digital twins, right? Uh, tracing the trajectory paths, I think technology makes a lot of these things essential that it becomes for masses, right?

So developers, for example, can participate in this. They can also contribute to this because explain, explain explanation of, or explainability of this whole, you know, how this whole technology is evolving, uh, should be done. And this can also be done by developers, for example, right?

So it's not like ki kind of a premium opportunity for a certain segment of people or certain segment of society, right? So I feel that, you know, technology also brings a lot of contribution and participation, which is exciting. Well, you know, I'm involved and always have been, uh, with a lot of international standards organizations and, and so on and so forth.

And, and just two outta three meetings before this call today have been on that. And what I've said in, uh, in both those is that I think we're moving, looking at that way from standards and protocols that historically have been developed by a handful of people. You know, then those of us who spend our time on that come from various per, per per perspectives for whatever reasons.

And that doesn't scale. And, you know, a lot of, a lot of us knew it wouldn't scale. And, you know, just like having one Google and one and this and one that, and not enough DNS service, not enough PKI, it's good and good enough.

However, ultimately it doesn't scale. We have to get to the fully federated model that was, the internet was designed in the first place. And you know, this come up a number of times, I'll just say it right, the risk rate now is centralization, authoritarian control, which if you don't agree with it, you know, whatever it is, you know, whoever has it that has a problem.

But I would go beyond that and say, from an engineering perspective, that is a crash waiting to happen. It is brittle and cannot be done. And if we go down that path for setting ourselves up for problems that will play out over the next couple decades.

I think green, to your point, I think quite the opposite is already happening, is the low entropy, cheap, better, faster, cheaper path where you enable everybody, it, it, uh, put this in there. So what I've been trying in the standard space, I think that most protocols will be developed by two nodes, two machines, two companies. No one will also ever hear about them because they'll be developed all the time in real time based on the agreements and relationships you already have in place.

And we'll only hear about it, SPDX and Cyclone Dx and you know, you know, H-D-P-H-T-P-S when it becomes important enough globally that a bunch of mammals in a, in a working group can wrap it up. But most of this stuff goes to autonomy. And if that's not fundamentally better based on sovereign sovereignty at the bottom end, then my name's Mickey Mouse.

Well, that's all folks. Uh, anyway, hey, we're outta time, man. I gotta pull it.

Chris Garima, thank you for joining us. Mike is always great. Hope you've enjoyed this.

You know, drop us a line if you like it better here at noon versus at 10:00 AM Eastern. 'cause we'd love to hear what you have to say. Uh, you know, we had text Drunk TV on before this, and we'll be replaying our text Drunk TV broadcast after this.

So hopefully it works better for you. But I, I'd really love for you here to know. We'll be back tomorrow at noon again on the gang with some fresh new topics and some more, more of our great gang friends.

Until then, though, this is Alan Shimel. We're out.