Techstrong Gang – April 22, 2025

Hey, everyone. Happy Earth Day. What's AI's role in it?

You're watching Techstrong Gun. Hi everyone. Happy Tuesday.

Our week is rolling along. You know, I just kind of view this week as the week before RSA conference and, uh, besides doing Textron Gang, the rest of my week is pretty much taken up with preparations for all of the various things that Textron will be doing at RSA, of course, our DevSecOps event on Monday. We're on broadcast allo week, we're sponsoring a few, uh, parties at night, speakers dinners, and all kinds of good stuff where we're gonna be out there in strength, along with our fu and brethren.

I think we'll probably be 10 to 12 people strong at, uh, RSA this week, so very, very excited. Um, but we still have a whole week before RSA. So let us, let's focus in on what we've got on this beautiful Tuesday.

Let me introduce you to our gang. First of all, joining us down Austin Way, the quintessential texting. He gave up his sweaters.

It must be warm there. Our friend Robert Reeves. Hey, Robert, what's happening?

Oh, not much. It's, I'm very excited, uh, because, uh, yes, it is getting warmer, but, uh, in, in preparation for Earth Day, I converted my yard to all native plants and, uh, got rid of the zoia, which was not working. And we went with buffalo grass and wildflowers, and they are popping right now.

It's beautiful. Very cool. Good for you.

Good work, man. All right. Moving from Texas to Hudson, Ohio.

I don't know if he's got any native plants there. I don't know what kind of plants he's growing. I'm not gonna ask, but it's, he's the founder of, uh, of Tech Field Dane, my friend Steven Foskett.

Hey, Steven. Hello everybody. Yes.

Um, and actually you said moving. Well, that's what I was doing all weekend. We moved to the village, and so we have a lovely little house, uh, in the village now.

And, uh, and it has now comes the fun of trying to figure out what's that native plant and is that a weed or is that a flower? Which is the, the most fun when you move into a new house? Absolutely.

Well, sometimes a weed could be a flower, be it smell. Well, whatever. I'm, I lost my Shakespeare quote.

Let's move over to Harrison or our chief content officer, Mike Huard. Hey, Mike, how are you? I'm doing well.

Earth Day, of course, Is an opportunity to celebrate baseball because, well, let's light on grass and dirt Play ball. That works. Build it.

They will come. All right. And then here in our audience, speaking of Earth Day and all things green and sustainable, is our Echo Insights advisor and editor Bonnie Schneider.

Hey, Bonnie, how are you? I'm great. Good to be here.

Happy Earth Day. Thank You. Okay.

Um, so let's jump into it, Mike. It, this Earth day's a little different in that like everything else. Today we're gonna ai we're Gonna do something different.

We're going right to Bonnie, who's gonna kind of introduce her segment and go from there. Fantastic. Go ahead, Ron.

Yeah, Well, you know, it's interesting. We talk about Earth Day every year, and typically, we'll, when it comes to big tech, we'll talk about their goals. How far along are they to the 20, 30 goals?

What have they done? And, and I'm still gonna do that in this upcoming report, but now things have shifted where when we talk about big tech, we also have to talk about AI and energy demand, because these pro the, the, what we're expecting in terms of energy use for data centers, every time I check the projections, it's even higher than it was, you know, a few months ago. So let's take a little bit a look at big Tech, some of the innovations they're doing on this Earth Day and how just looking at sustainability might be redefined, starting with this Earth Day as we go forward and look towards the energy demand of ai.

Big tech's green reputation is being redefined this Earth Day for years. Tech giants like Google, Amazon, and Meta have been celebrated for driving record investments in renewable energy and pushing sustainability to the top of the corporate agenda. Last year, wind and solar surpassed coal in US electricity generation.

A milestone made possible in part by these companies, massive clean energy contracts. But the situation is changing. The explosive growth of AI is fueling a new wave of data center construction, and with it soaring demand for electricity and water data centers now account for up to 4% of US electricity use.

And experts say this could double by 2030. So how are tech companies balancing their climate commitments with ai, increasing appetite for energy? The approach is multifaceted.

Renewables remain the backbone of their strategy with continued build outs of solar, wind, and storage projects. Efficiency gains like AI driven cooling and smarter hardware are helping to curb some of the impact. Meanwhile, nuclear is positioned as a long-term bet rather than a short-term fix.

There's no denying the achievements. Big tech's purchasing power has accelerated the global shift towards clean energy and has made cloud computing more efficient for businesses everywhere. The focus going forward is likely to be on flexibility as AI models evolve and efficiency approves, the actual energy impact could shift in unexpected ways, potentially easing some of the pressures or even revealing new ones.

For now, the most resilient tech leaders will be those willing to recalibrate quickly, matching their ambitions to what the grid and the broader energy system can support Earth Day 2025. Does Mark changes in, uh, this goal to be more sustainable? We've already seen some companies change and adjust their goals and their timelines because of the demands of ai.

So it's an ongoing report, but I think it's, this is the first year where this Earth Day, it's, it's really in prime focus. Absolutely. And you know, you're right Bonnie, it seems every time we turn around, there's another a hundred billion here, 200 billion there being pledged for these AI data centers.

And, and the, you know, and you know, Microsoft's made a deal with Three Mile Island, you know, to buy nuclear power from there. That doesn't sound like it'll end well. Um, but, you know, in light of Earth Day and, and also look, not to be political, but the fact of the matter is we're hearing all kinds of rumors about many of our environmental protection laws being rescinded or suspended.

Though I don't know how you could do that without an act of Congress and for the, for the sake of drill, baby drill, in spite of the fact that solar and wind seemed to be really catching, you know, tailwinds, no pun intended. Um, I don't know, panel. What do you think about?

Well, let me start with where these data centers are being built is kind of surprising to me. And I'm even gonna concede a point here to Steven. We on a show a few weeks back talking about data centers in Ireland, and he was suggesting the west coast of Ireland.

And I kind of said, that's unlikely given the fact that everybody is in Dublin and wants to go to Europe. And sure enough, over the weekend, I get a news bulletin outta Mayo in the west of Ireland talking about how they're building a data center in the west of Ireland, and here we go. And at the same time, they're in the Midwest too, in large ways, farm country, and even down in Texas, you're seeing these data centers showing up in odd locations.

Um, Steven, why are these data centers not where we think they might have been in the past? Well, I still think that, uh, data center location is primarily driven by the location of, uh, end users and also by the nexus of communication. So one of the things that Western Ireland has is that's the, where a lot of cables land.

And so it makes sense to have data centers there, um, you know, some of the other places as well. Uh, you know, it may not be obvious, but there's actually quite a lot of telecom in Columbus, Ohio and in central Texas. And, you know, and I think that that's one reason that these things are being built there.

But I do feel that, uh, increasingly we're gonna start seeing people locating data centers where the power is and where the environment is. Um, you know, they need water. Uh, you, you don't think that data centers need water, but they do.

Chip fabs also need incredible amounts of water. Um, so it makes sense to locate things there. Um, and, and if you're putting them in the Midwest, if you're putting them in Texas, uh, I was flying into Austin last week and, uh, and I got a, or actually Houston, I got a beautiful view of a massive solar farm north of Houston.

I think that this goes to the point that I've been trying to make here, which is no amount of, um, politics is going to derail the green energy future, simply because solar makes a lot of sense, especially in places like Central Texas where there's a lot of sun and a lot of land. And if you can put up solar panels and get free electricity out of that, well, that's gonna make it a lot of sense. Same with Iowa, where there's a lot of wind, and so there's a ton of wind energy there.

No amount of anti wind agitating by politicians is gonna derail the fact that if you built this tower and the thingy turns and you get the electricity out of it, then it's, then you're gonna use that. Um, you know, and frankly, if, uh, if you build a power plant where there's a whole bunch of coal or a whole bunch of natural gas, I imagine people probably use that too. But overall, I think a lot of these things are driven not by politics or even explicit, uh, corporate strategy, but, but by external externalities.

Like where the customers, where's the telecom? Where's the power? Where's the water?

Where's the land? And all of these things, I think are coming together to make, uh, green energy more effective and more widespread. I, you know, Robert, we were talking off off camera a little bit before we started today about the situation specifically in Texas.

You know, my, my youngest son is a sportscaster for K-T-A-B-K-T-A-B in Abilene, Texas, deep in big country, as they call it. And when I go out there to visit him, driving from, you know, Dallas Fort Worth to Abilene, once you get a little bit outta Fort Worth, you know, into big country, the hills are alive with the sound of wind turbines. Well, they don't really make a lot of sound, but there are wind turbines as far as the eye can see.

And, um, they're actually building a huge, huge new data center in Abilene that they think is gonna turn the entire economy. Right now, Abilene the biggest, uh, employer is this big Air Force base right there. So a lot of, well, Yeah.

And, and, and, and the law, the, uh, the one university there, Abilene, Christian Godly covers it. Well, if you're looking for a, you know, a CU coverage turn into K Ttab. But that being said, I saw an interesting report that, you know, for all, for all we think of in Texas with guys Drive, you know, from the Flintstones driving cars with the big Longhorns, you know, on the front and everything, Steven remembers that, um, the fact of the matter is that most new energy coming online outta Texas is solar and wind much that's more than oil.

Mm-hmm. Or gas. That's right.

And the fracking and all of that. Now, I'm sure I, I don't think Greg Abbott's jumping out of his wheelchair wa you know, waving flags to promote solar and wind energy. He doesn't have to 'cause it's happening Whether he wants it or not.

Well, geographically, if you look at that part of Texas, that's right before we get to the High Plains Mesa, that stretches all through, uh, New Mexico and the southern part of Colorado and those sorts of things, and where it goes from the plains Midwestern planes up to that, that is a perfect place to put turbines. 'cause that's where it captures the wind. Now, if you are living in that part of Texas, uh, from Abilene up to the panhandle, um, and further west, what you're gonna notice is a homeowner is that after 9:00 PM you get free energy.

They have deals where the energy providers, because they have an excess of energy, that's where you can run your dryer and those sorts of things. And so folks that are building data centers, and especially for ai, which is huge con consumer of power, well, you need water, you need power, and you need cheap land. And so this is just, all these things coming together has led to two areas where you're seeing growth in Texas, uh, for data centers.

One is that part of Texas from, you know, Abilene around I 20 and up to the panhandle, but also in South Texas where there is an excess of wind, right on the, uh, right on the coast and solar, but also LNG natural gas. Uh, they're pulling that out and there's a reason why they've built those LNG terminals into the, um, into the gulf where you don't have to come into a port to fill up ships. And so, um, and, and, and also, uh, you know, close to the Gulf, and if you have an excess of power, desalinization makes sense.

And so I, I do agree that in the past we have seen, and I'm by no means an expert on data centers or telecom, but we have certainly seen that we put data centers around where our big peering points are. But now with ai, we can, you know, have models built, have it work in the data center, and then we can take that model out of the data center once we get the data in there. It's just gonna work.

And so it's mainly about power and water and cheap land. And we've seen the beginnings of this going back 10 years, where cattle ranchers have leased out their land to put in turbines. So not only are they making money off the cattle, but also the leases for power.

And you're starting to see more and more. And in the past we saw this with oil and gas, you would have, uh, grazing land with a, uh, oil rig or a natural gas rig, and now it is, um, wind turbines and solar lifted up so that the cattle can graze around it. It's, it's very interesting.

Um, you know, if you just look at the press, you wouldn't think and conventional wisdom, but it is happening anecdotally. Um, I know of several landowners in South Texas that are getting crazy offers on their property for data centers. And I, I'll just point out that it's not just anecdotally either.

Uh, you're absolutely right, Robert. So, uh, wind power, Texas is the number one producer of wind power in the United States, and Texas is the number two, uh, producer of solar power in the United States as a percentage of consumption, uh, only slightly behind California. And, uh, Texas actually doubled their solar production in 2024 alone, which is wild.

They're number one, one in the country deploying solar. 'cause it makes sense. That's right.

Money. Well, So, so, so Bonnie, what's your assessment here? I mean, 'cause you will read a lot of coverage out there about how AI in these data centers may not be good for the earth, but can we sustain this?

You know, that's remains a a, a ongoing question a few weeks ago, you know, we talked about the possibility of nuclear that, that a lot of companies were investing in that, but Steven pointed out that's not a short-term solution. It's more of a long-term one. So, um, I think that it's gonna be a balancing game.

They need as much energy as they can get solar and wind work and make sense. Now, they, they have intermittent properties where they're not constant, so that that isn't an issue. But I think just the fact, what you were talking about in Texas, you don't see everyone outside of Texas talking about that or knowing about that.

But I think it's good that we're explaining that another area for, uh, power that's being explored when we were talking about being geographically aware is here in Florida and then up in New England, they're doing a lot more research with hydropower and current power, the power of the ocean currents to create energy. There's actually some, um, experiments and things that are going on right here in South Florida with that. Yeah.

So it's waves wave energy, right? So I think wherever they can get it, that it, the need will be there. And then the flip side, of course, course is that we're also talking about often is how efficient can this AI be?

Is it going to improve on its own and, and teach itself to be more efficient? So we're watching both of those, but it's still a big unknown factor because as I said, that energy demand just keeps growing. Even just in this conversation as you were talking about all these new data centers.

I was thinking about that. Wow, I hadn't heard about that one and that one. So there's probably so many globally that this is happening.

Now, there is one way that politics could get in the way here, and that is that most solar panels are produced in a place called China that you may have heard of and would be subject to punitive tariffs under, uh, both administrations in the US and in China. And that could cause problems. So I will say, despite all my enthusiasm for this as a business decision, if those solar panels cost 125% more than they ought, then they ought to, then that's gonna really eat into the, uh, market for them.

They would never do anything. Like, how would they, Steven? I hope not.

But there, but there might be an interesting windmill smuggling business coming up soon. They're real hard to smuggle real hard. They're pretty Big.

Um, there's actually a big wind turbine factory in a, in a, a, a magical land called Cleveland, Ohio. Oh. And, uh, so you will see those, uh, giant, uh, a blades rolling down the highway here, uh, in Ohio.

So, uh, perhaps that can be homegrown. I wonder what it would take for us to develop solar panels, solar film, you know, solar generating materials here in the us. Well, there was an attempt to do that under, uh, president Obama, and it became a major scandal after the funds were mismanaged.

And, uh, I don't know if you remember the whole Vaguely, I vaguely remember something. Yeah, Yeah, yeah. So we don't have that in that, uh, we don't, we don't make those Shame.

Anyway, an excellent opening for this lovely Earth Day. You know, guys, we only got one home on this planet and we all share it. So keep that in mind.

Let's take a break here on Textron Gang. We're gonna come back and talk about a big tech field day going on this week. Steven will lead us off on that.

You're watching Textron Gang, Discover Techron Group, the epicenter of tech innovation. We are your go-to for reaching IT leaders and practitioners worldwide. Our secret impactful content that sparks awareness, engagement, and top quality leads with us.

You'll access editorial websites, streaming videos, virtual events, custom content analyst research, and more. Join our satisfied clients. Let's revolutionize your tech journey.

Contact us today and tell your story to the world in the most powerful way with Textron Group. So, as we were just discussing, uh, the growth of AI has really driven the growth of, uh, data center and, uh, IT infrastructure spend for the last year or two. And that doesn't seem to be slowing down.

Uh, we, uh, at the tech on the Tech Field Day side here have, uh, created an AI infrastructure Field day event because we figured that there would be interest in sharing some of the news about what's going on on the, uh, IT infrastructure side with, uh, supporting AI applications. And to say that that was successful is quite an understatement. Uh, many tech field day events are two days long, this one's four days long.

That's the kind of response that we've gotten. And in fact, we already have about a half a dozen companies signed up for the next one that we haven't even announced yet, because frankly, there's that much excitement and interest in building infrastructure to support ai. So we are gonna be talking over the next few days here, uh, starting today when this episode airs, uh, you'll be able to see these presentations on Textron tv.

Uh, we're talking to a wide variety of companies, not just, you know, the traditional cloud companies. Yes, Google Cloud is live all day today, but also more of the infrastructure companies that you may say, wait a second, what's their AI angle? So we've got Juniper solid, IM Fon, uh, Nutanix, you know, companies like that that are gonna be talking about basically how they're making their corner of infrastructure better support ai.

I've been talking about this quite a lot from a storage perspective on our utilizing tech podcast series, which focuses on basically how does, how does storage support ai? It's all about data. And, and frankly, a lot of this spend is about data, whether it's data storage, data retrieval, data management, manipulation, processing, et cetera.

Uh, I can't wait to to hear what these companies have to say because frankly, we all are. It's in everybody's interest to make AI supporting infrastructure more efficient, uh, to move the technology forward. Because right now, we're actually not making great use of infrastructure.

One of the things that came up last time at our AI Field day event was the fact that on average, uh, production AI hardware, the, the GPUs, those incredibly expensive GPUs, only 23% utilized, uh, even when they're in production. So you hear about these companies spending billions of dollars for GPUs, and yet they still have three quarters of the time those things are idle. Why are they idle?

Is that, you know, because they don't have a a use case? Is it because they don't have data feeding them? Is it because they, uh, don't have optimal data pipelines or optimal software?

Well, it's all of those things. And so all of these companies are all working together to basically make better use of the resources that we have while they're also at the same time deploying ever more resources. So it's gonna be really interesting to learn about this.

I I'm curious what you all think, um, had you thought much about the infrastructure that supports AI and supports these GPUs? Well, I'm thinking that most of the data center is gonna need to be re-architected, whether it's on premise or in the cloud, because, um, if every application is making some use of ai, the infrastructure needs to be optimized. And most of what's out there today wasn't built with AI in mind.

And to your point about, uh, utilization rates, they were abysmal even before ai, and now they're just getting worse and more expensive. I can't help but wonder though, um, I feel like as I look at it, we're doing all this stuff in a piecemeal fashion, and I'm wondering if there is a need for a more holistic conversation and approach, because otherwise we wind up with a couple of GPUs over here, but oh yeah, now we gotta fix the network. Oh, wait, now we gotta fix the storage systems.

It seems like we're kind of doing this ho podge thing that never works out. Well, You know, to me, this is a case of, it's always the guys who sell the picks and shovels that make the money in the gold rush, Right? Yep.

And, and look who's selling these picks and shovels? You got the server companies, you know, you, you got HPE and Dell and Lenovo, you know, you got the storage companies, you got Juniper and Cisco and Broadcom and these companies. Yeah, they're, they're certainly making a lot of money right now.

And of course, big daddy and green, Nvidia is making a lot of money selling fix and shovels. Well, gosh, if only there was some way that we could maybe pull our resources together and maybe a, you know, one company would have a bunch of GPUs and people could use them as they need them if only, yeah. Like something like that where you could spin it up, spin it down, but I don't know if that concept would ever work.

I don't know. That's an awful Cloudy to me. I'm not sure I'm on board With that.

You're have a tough time selling it to the, to the mainstream. Yeah. It, it's just, you know, look, I love technology and I luckily I've, I've been around long enough to be able to see patterns evolve over and over again.

And I love how we have found solutions for problems that we've had in the past. Uh, for analogs. It's good.

Yes, companies are not utilizing them, um, but they are going to increase then get more efficient. Uh, it's, it's, you know, something that this is an opportunity for companies to build startups around this. And, and as a, you know, former two time startup founder, I would encourage anybody looking at this space to look at previous analogs and apply it to this new space because they've already run the traps and figured out what solution, you know, how to market, how to sell.

You just need to build your solution and followed a, uh, well tread path. And well, to Mike's point too, you know, they're the, the big cloud providers are absolutely looking at this from a holistic perspective. I mean, Google and, and companies like them, are they, they've got their own switch hardware, they've got their own, uh, AI accelerator hardware.

Uh, they're deploying a complete soup to nuts, top to bottom infrastructure stack. I mean, you look at Azure that's, uh, powering, uh, open ai, uh, among others, there are a lot of companies that are building hardware and software to serve ai. And, um, and it is interesting to say, wait a second, uh, are they, you know, what, what is the result of that?

Are they able to deliver this at a consistently lower price point than cobbling something together in your, oh, that doesn't sound good, but, you know, then building a best of, best of breed solution of your own. Or are there other reasons that you would want to build it on, on your own in, in-house instead of running it in the cloud? And, and I think that there's a lot of sensitivity as well on that point.

So, I I, you bring up a good point, Steven, and I think that's going to be a, a back and forth kind of battle the same way to Robert's point about, you know, hey, maybe there's this thing called the cloud, even though we're, we've been in the cloud era now for going on 20 years, right? 2005, 2006. We still, I mean, we're just a point approaching a tipping point where there's probably more infrastructure in cloud than in data centers.

I don't know if we're there yet, though. I think there's probably still more in private data centers than in public cloud. So, you know, in terms of that, I think it's, it's the same thing here.

Yes, the hyperscalers do enough scale where it pays for them to develop their own ecosystem, their own switches, their own hardware, their own, you know, much like Google kinda reinvented the server market when they started using their own servers, right? Yeah. Now look at OCP, right?

Yeah. Michael Dell still re reuse that day. But, um, you know, so I think that's gonna be a battle that goes back and forth, like a tug of war.

You'll see it, uh, certainly people who don't have that kind of scale, it doesn't pay for them to develop their own complete ecosystem. They'll buy cuts hardware. You're, you've already, you've already seen AWS and Google and Microsoft build stacks of their cloud infrastructure that you can deploy in an on-premise environment.

So logically they would do the same thing with their AI additions of that platform. So Yeah. But then there's this thing called locked it Lockin and I, and I think a lot of the smaller people want to avoid lockin, but here's another, If I get locked into, if I get locked into Dell, HP or Amazon or Google, what difference does it Make?

It's all No, because, you know, what, if I wanna switch from Delta HPE to Lenovo, it, it used to be in the world, right? The, the underneath the nameplate, there was a lot of similarities, and it wasn't as a big deal to switch versus I'm buying into the Google stack. Buying into the Google stack is, is a quite another story.

You know, you can't just migrate that. Well, And a lot of that lock-in is software driven too. Yeah.

The reason that people choose, for example, Google versus, uh, Amazon versus, uh, Azure for, for building AI applications or other applications, is because they're buying into that software stack. I mean, Google has spent tons of money trying to woo developers to come into the Google Cloud. And I would say that, you know, at least from my perspective, Google's primary pitch to the world is our stuff is developer friendly.

Amazon as well. I mean, if you go to AWS reinvent, you know, just look around there. It's, it's, it's, it's all about developers.

And, and then that of course is Microsoft's history too. So I, I feel like a lot of 'em are using software rather than hardware to try to drive people to their platforms. But one of the crazy, So was Nvidia Steven, right?

Absolute was Cuda. Absolutely. Well, it, it, it wasn't for Cuda Nvidia.

Where would they be if, if, if that hadn't been the development target for all these AI models? And that's one reason, for example, that Intel hasn't had more success. It's not necessarily a hardware problem as much as it is a an API problem, because people would prefer to use the Nvidia stack to deploy their models.

And there you go. And, uh, you know, a, a attempts to, to, to, to woo developers in a different direction have not always worked out very well. Uh, and Nvidia, here they are.

But one of the things I wanted to point out is that, um, if you look at some of these traditional AI infrastructure companies, one of the surprising things that we learn at Tech Field Day when we're talking to them is that they are actually powering the clouds as well, the hyperscalers. And so we were at HPE last week, they can't talk about it, but I would guess that a significant version of HPE E'S server revenue is driven building racks of, of cloud, yeah. For HPC and for, uh, cloud service providers, I mean, h hp, HPE rules, the HPC market, they, they build most of the supercomputers.

And, and the same thing with some of these other companies. You look at Juniper, and actually, here's another dark, dark horse. There's a company Keysight that makes testing equipment, and, and they have specific AI testing improving products.

And it's basically a product for the hyperscalers, to be honest. It is the product that the hyperscalers are, are, are, are leaning on to make sure that their, uh, products are efficient. You know, there's basically a market of 20, but it's like a huge market because these companies are buying so much.

So even these traditional companies, I mean, Keysight is, is derived from the original garage, the h the Hewlett and Packard garage. That's how far the back, they trace their roots, and yet here they are developing products, targeted specifically at clients like AWS and Azure and Alibaba and so on. It's, it's kind of amazing to see Robert, I thought, I just heard a plaintiff cry for an open source alternative to Cuda managed by a consortium.

So we don't get locked into all this stuff, but I don't know, is that possible? I, I Thought isn't Cuda open? Yeah, but it's, it's still Nvidia, Right?

Well, we certainly have OPA, which is, uh, Intel donated that to the Linux Foundation. And that's seeing, uh, certainly, uh, enterprise model for delivering AI solutions. Um, there are things out there.

I have always cautioned people that are concerned about vendor lock and to use open source, because it is a, it is a standard. Um, and you have to be a bit militant about, uh, not buying into other parts of a vendor's ecosystem that does lock you in. Uh, so for example, Kubernetes, well, then you're starting to use proprietary services at that cloud provider.

Great. You can go to another Kubernetes service, but you're still locked in on these proprietary services, so you have to be a bit militant. Um, and, uh, well, maybe not militant, perhaps, um, aware, we'll, we'll say aware, be very aware, regimented of what you're, mm-hmm.

Yeah. Um, but it is, look, we are, I'm just very happy to see that customers are concerned about vendor locking, because there was a time where, uh, not too long ago, where you use solutions based on who your CIO or CTO played golf with most recently. And I'm very happy to see that developers are driving this.

And my recommendation for developers who are picking solutions is not just pick on ease of use. Certainly that should be a concern, uh, and features and functionality, but also what it means to the business. Be aware of what these choices, the impact that these choices have on your company and, and be a good steward and leader at your enterprise.

Man, Robert, you sound like a tech field day delegate. I, we gotta talk, Would love to it. It's just, look, I really want developers to live their best life and enjoy what they do, but at the same time, I don't want to see them or the business paint themselves into a corner that's very difficult to get out of because I've been there and it's not a happy place to be.

And I, I want people to be in happy places. Here's, and end this up in, in this way. I'm waiting to see an AI infrastructure glut.

'cause we'll over build as we always do. And then, you know, that'll, that'll even itself out too, when it'll be, I don't know. But anyway, Steven, I do know when AI Infrastructure Field Day is, it's on immediately after the gang today.

I don't know if it's immediately after, but after the gang today. So if we've wet your appetite or what your whistle for any of this, stay tuned on Textron TV today. You can catch the AI Infrastructure Day live from Silicon Valley on Textron tv or on YouTube on the Tech Field Day YouTube channel.

It, it'll, well, no, there it goes. That's recorded next week. Yes.

Yeah. But it's gonna be live streaming on Techstrong tv, uh, all, all the rest of this week. Um, this, and, and basically there's something for everybody, you know, whether you're a storage nerd like me, or, uh, whether you're one of those weird network people, uh, there's all sorts of stuff.

Even, even even those antisocial security folk. Um, anyway, we're gonna take a break here on Textron Gang. We'll be back and we're gonna talk about slop squatting.

Okay. This sounds like something for Robert Reeves. You're watching.

com is the leading resource for news analysis and education on challenges facing the cybersecurity industry. com covers all aspects of cybersecurity, including data security, DevSecOps, cloud security, application security, network security, security threats, and more. com has the largest selection of security content featuring breaking news, blog posts, podcasts, and more.

com to learn more. com. Home of security Bloggers Network.

Hey folks, we're back. And we're talking about this slop o squat term that has been kicked around. It seems to be a cousin of typo squatting, but the idea here is that, uh, bad guys are convincing l LMS that a package may exist, that the LLM then inserts into people's code or uses as a baseline to create some additional code and surprise, there's malicious code injected into these applications that might be invoked later by somebody who has a command and control system for that thing.

Robert, this feels like, you know, a, a nightmare out of DevSecOps gone wrong with ai, but how serious an issue is this? Um, not very. Uh, I think, look, we have had injection, uh, you know, security injection challenges around libraries for decades.

Uh, we've certainly seen people use social engineering, uh, as recent as year and a half, two years ago into the kernel they caught it. Uh, we've certainly seen, uh, people hijacking, uh, JavaScript, uh, uh, packages and those sorts of things. I don't see very much difference here.

Um, as far as if a ai, if LLM is hallucinating that a library exists, somebody creates that library, has base functionality for it, and then has some sort of way of injecting that into the system, phoning home collecting data, that's the idea behind it, which is very similar to any other hijacking of another library. Um, I think it's novel and unique, uh, but I do think it points out a bigger problem, which is AI hallucinating that these things exist. And I would say that it is really about how you use AI for code generation.

If you are using it to say, Hey, I need to change, uh, my, I want to send, uh, all of my data in this service to o an hotel collector, open telemetry collector. Great. Hey, copilot, go ahead and add all the calls that we need, uh, to create this, and it'll do it.

And it's wonderful at that, uh, if you are using, uh, AI for vibe coding, uh, where you're just like kind of interacting with, uh, the prompt, you've got the microphone on, you're, you're, of course, you've got a dark screen and you're listening to the Tron soundtrack. Great. Yeah, Huie.

Yeah. You really, yeah, you're hacking. It's great.

Um, if you're doing these things, um, for anything other than prototyping and wire framing, you're doing it wrong. Um, I think it really elevates the conversation of what is AI's role today could change later, but what is the role of AI in supporting coding and programming practices? Uh, is it for this, is it for that?

But I do like the term I like slop squatting. Mm-hmm. It's catchy.

And I do think it's yes, it, and it gets crass and it's wonderful. And I do like the fact that it is getting elevated and people are aware of it. But the bigger question, I think, and I'm curious what the panel here thinks is like, what is the role of AI in coding?

Is it something that should be putting out production ready systems as some executives might believe? Or is it a role? Is it something for product managers to try things out and figure it out?

Maybe it's, it's whiteboarding, but, uh, via vibe coating? Well, I think the answer is today versus tomorrow, right? Today, maybe it is vibe coding today.

Maybe it is, uh, prototyping, experimenting, playing, maybe not production ready. I think the other big thing is no one in their right mind is just gonna let AI do some code and, and hit the deploy button on it, right? Not without, well, Mike says maybe mm-hmm.

But, but you know, you, you need humans. I'm questioning how many of 'em are in their right mind, right? So there you have, but you know, it certainly need, needs to be reviewed, tested, what have you, by humans and maybe maybe the today anyway, again, the role of AI is better off testing code than humans generate.

Well, to that point, think about what the two things you just said, Alan, it needs to be tested, and AI is testing. It's true. A lot of people are using AI for QA and validation of code.

Well, if we're using AI to generate code and then you're using AI to QA and validate code, uh, maybe there's a problem. Yeah. That, that's definitely a problem.

I mean, it, it, I think it's gonna be one or the other, other, but, but let's be clear, the promise that all these tech execs, you speak to Robert, who want to get rid of their junior and intern developers and turn AI loose on it, the promise is there that it could do that sometime, some way, some, right? Um, somewhere I think Dakota, Barbara Streisand song, but, um, that ain't today. That, that, that's the point I'm trying to make is today.

I don't think it's real. Um, but it's interesting, you know, it's why we can't have nice things on the internet. It's not even real yet.

And already the bad guys are trying to de not derail, but inject themselves into the, into the equation. Well, sure, Sure. There's always going to be crooks and criminals.

And I'm, I'm happy that this has been, that this has come out. I'm happy that we have people in this industry that are looking for these things. And I, I, I've seen, you know, AI hallucination myself, I never thought of it to, to use it in terms of like, oh, okay, let's, let's start taking advantage here.

Um, I have used it for, I actually leveraged that hallucination for prototyping. Um, at my last company I was asking it, show me how liquid-based supports couchbased. We didn't have couchbased support, and I was curious and it spit out documentation on how to do it.

And I was like, well, this is a lie. 'cause I know for a fact it doesn't. But then I was like, hmm, copy and paste put into Jira.

And that just became, well, I gotta tell you though, the J ticket, but I've seen college student hallucination too, right? Mm-hmm. Uh, tricky Mickey lady, blue fingers and all of that good stuff.

So, you know, why, why should AI be different hallucinations? Uh, you know, Oh, have a good time. You have a good time.

Chat, GPT, enjoy yourself. Right? But I mean, but all kidding aside, we've gotten better, we've gotten better with the hallucination stuff.

I think, well, Being aware of it. Remember those lawyers that were citing fake cases in court and they were chat GT Crazy. Yeah.

But To that point, Bonnie, that's actually really smart. So take what, what's happening here with slop squatting, right? Okay, so somebody creates a library.

So AI hallucinates a library that doesn't exist. Somebody creates that library as a malicious library, and then they use that to do, basically to bring vulnerabilities into software. Slap squatting, doesn't have to be software development.

Think about what could happen to your point, Bonnie, about somebody creating basically a, uh, a fake academic paper, uh, based on a frequent AI hallucination, a fake court case. Imagine somebody creating, you know, a fake GitHub or a fake Lexi Nexus or something filled with fake court cases or fake, you know, academic citations or something to basically just derail the process of humanity and culture as, as a whole. You know, if AI is gonna say, you know, Bonnie wrote an academic article for when she was back at Princeton about such and such, and, um, and then somebody actually creates that paper probably using AI in 15 seconds and, and uploads that to something, then suddenly somebody's gonna start crediting you with that idea.

Slap squatting is way bigger than code. Yeah. Interesting.

Absolutely. It can be, you're right, Steven. So, so Robert, though, I kind of feel we maybe need to update our best DevSecOps practices for the age of ai.

I think people are all over the place on this stuff. And is there something that should be done here in terms of having a, a moment where we, you know, and this document would be a living, breathing thing. 'cause AI changes all the time, but I feel like we don't really have something that kind of tells people what's real, and that's what's not.

Well, we, we, that's true. We don't, uh, but we do have lots of options to validate the libraries that we're using and that we're pulling in to, and that our code depends on because we have decades and decades of experience with that. I will say this though, actually, it might not be a bad thing for somebody to engage in slop squatting, because they'll have to build the base functionality of that library to actually work.

And yes, there will be some security, you know, some issues in there where they're stealing data or doing something bad. But at the very, very least, somebody could take that library and, uh, remove the bad stuff. And here you have, uh, the library working.

Um, maybe I'm, I'm looking at this glass half full, but I do think that there is lots of great security practices in place today that we can take those approaches to this as well. Uh, we do have protection against naughty libraries that are, uh, do you know, that we don't wanna bring in, uh, I'm, I'm re reminded, uh, the SolarWinds hack and, uh, you know, secure software supply chains. So those practices would, um, hopefully protect us from this, and we would just have to apply it, uh, to this, this new vector of attack.

My issue with that is, so I can see a world where we're gonna use AI to fix the code generated by the ai, but the thing that we're relying on to fix the AI is also flawed. So now I need another thing to fix the AI that fixes the ai. Where does it end?

Well, it doesn't, and that is what, um, I call the circle of suck and where it just feeds on itself and it just gets worse and worse. Uh, and, and it is, um, yes, it's going to be a challenge, but haven't we seen this pattern before? Uh, where things, uh, we, we just start parroting things over and over again, uh, with urban legends, uh, and, and those sorts of things.

Now it's just accelerating and gotten worse. Um, and I, I do believe that we, we should be aware of it, but I am confident that my security colleagues are on this and that they can apply practices that they've learned from other issues and apply it here. Very cool.

Hey guys, I gotta pull the plug. We've got a lot going on today on Text Trunk tv. This has been a great gang, Bonnie, always great seeing you.

We thank you'll see you next week here on the Gang. Steven, I hope, I hope this AI infrastructure day four days of it is gonna be as great as it sounds. We're looking forward to seeing that.

Rob, Robert, we haven't got an update from you in a few weeks. What's going on in your world? Well, I am just having a blast, uh, working with all of my friends, uh, in startups, uh, around Austin, around the country, and, uh, really enjoying, um, spending time bringing that 1984 GMC Army blazer back to life.

It is just been For people who say, I'd like Rob, I'd like to pick Robert Reeve's brain. How, what's the best way to reach on LinkedIn? Or what do you think?

Well, Absolutely, I think LinkedIn is a great place. Uh, you know, that's, uh, always respond there. Um, and, you know, certainly, uh, you know, I very much enjoy having a nice chat with folks and seeing how we can leverage open source to improve their business outcomes.

I do think that there is time and time again, we've proven that open source is the model for software development that works. Um, and companies, larger enterprises, uh, need to take more and more advantage of that, you know, to avoid vendor lock-in, to have more people that they can recruit and retain. And certainly, it's just cool, man.

I think, I think Alan inadvertently gave the best advice in the entire show when he said, when it comes to ai, I gotta pull the plug. Well, that's one man's opinion. Stay tuned for a full day of, uh, tech Drug tv, including Tech Field Day.

Infrastructure Field Day today, tomorrow, Thursday, and I guess Friday and Friday. All right. Until then, this is Allen Shimel.

Thanks for watching. We're out.