Linux Foundation’s AI Agentic Push, Security Risks, and AI Energy Efficiency | TSG Ep. 985

Hey everyone. Did you hear about the latest foundation? No, not another sci-fi book You're watching.

Textron Gang. Hi everyone. Happy Friday.

Hey, man, I am, I gotta be honest. Fridays, as we get closer to Christmas and New Year's, they're kind of special 'cause we've got stuff doing all weekend, getting ready for the holidays and enjoying that festive spirit. I hope you're all festive out there.

I hope your plans for the holidays are going well. Um, I, I know mine are, I'm, I'm happy to be home, to tell you the truth. Everything feels like months of traveling.

We've got a great gang, a full house here for us today. Let me introduce you to our amazing gang. We've got my friend Fred Wilmar out in Seattle.

We've got Gina Rosenthal, Wiki Wang, Mitch Ashley, John Schwartz, John Schwartz, the Bard, Mike ard, and yous Truly Alan Shimel. We've got a great gang for us. Mike, I, I said something about a new foundation.

I'm not talking about a new Asimov book or an Apple TV spinoff, but no, a Linux Foundation Foundation. It, it is Yet, yet another foundation from the folks at the Linux Foundation. I've lost track of all the different foundations and I've also lost track of all the different AI projects.

But this one seems to be about, while the AI Agentic Foundation is gonna be home to the MCP protocol, which is being used widely in servers for different use cases, and also something called GOOSE at Block Built, which is an integration framework for AI agents. md that you can use to provide a standard interface for accessing code repositories. So, Mitch, what's your take on what's going on here?

Because these guys also have the A to A protocol and there's a couple of gateways and it's starting to look a little much like maybe, I don't know, shake and bake. You throw all this stuff in a bag and something good happens. What's going on?

I, I think that's the idea, is not put 'em all into one place and either have them get lost inside the Linux Foundation or get coupled into something else, but give its, its its own space to develop. And does the world need another foundation? Yeah, I actually think it does another foundation book too, in Siri, but also another tech foundation.

So, thi this, what what's impressive about A A IF is that it, one is that MCP was donated to it. That is like the most popular, of course, open Standard, if you've heard of one, you've heard of that. And, but there are a lot of other really important contributions.

Um, you mentioned there's something called Agents md. If you're not a developer, you're probably not that familiar with it. What Agents MD is, is actually a markdown document that provides a structured way of how agents are built.

So it gives the structure of the agent the required components, what the inputs and outputs are, um, interfaces for tooling constraints and guardrails, et cetera. So it's kind of like you heard about structured prompts. This is like structured how to build your agent.

Uh, so it's almost like an instruction manual for it. I think that's super helpful. OpenAI gave, gave another number of other things to the foundation.

So you, it, we are in this age of, okay, if we're gonna start doing more agentic type workflows, agentic work, whatever it is that agents are doing, you know, enterprises, all of us are asking, well, what are these things doing? How do I trust them? How do I control 'em?

You know, anybody creates one. What do these things look like? They're gonna be well developed ones, they're gonna be poor, poorly developed.

So there's a lot of really good reasons why I think a IF has been set up and kudos to the, uh, links foundation. So I I, I have some views on this. There's one might Expect, Lemme give you shimmy steak here.

I'm, I'm of a mixed, mixed mind on this. Part of me says, look, they were donated this MCP, uh, protocol, right? And they had to find a home for it.

They probably didn't wanna put it in CNCF. It really didn't belong there. And do we really need more players in CNCF with over 200 projects there.

Now at the same time you, I, I wasn't the a to a, uh, moved over to Linux Foundation too. I thought that this would be in here. Yeah, I I thought it might be too.

Maybe those things will move. I don't know. I I think it has to be in there.

I I felt like, you know, this was the first time where they had a, they had an open source protocol or an open source project, and they, they literally, you know, they created a whole foundation around it and threw everything else in. My, my issue is because autonomous AI is going to be so intrinsic in every single thing that we do, how do you separate out the autonomous AI work being done in CNCF projects, the autonomous AI being done in mainframe projects, the autonomous AI being done on the Linux kernel itself and everything else that the Linux Foundation does to try to say that we're going to keep all of this autonomous AI stuff in this foundation, I think we're kidding ourselves. It's gonna leak into everything and then it's gonna get messy.

Well, does it belong in in this one? Why is it in that one? And, you know, and there'll be, there'll be political fights over at Turf Wars.

Um, and then if, if the MCP is really the centerpiece of this whole thing, I, I got news here. Look, it went out there. It's, it's a year old and it became the defacto standard.

Doesn't mean it's gonna be the standard next year. We're, we're liable to have something that is replaces it or updates it or what have you. Or maybe it is a to a or something else.

Um, I, you know, I, I do think, look, we are in the foundation era, right? Open source can't be ruled by one big brother per project. You, it's good to have it in a, a not-for-profit foundation where you can have cooperation among competing companies.

But, you know, I I, I think this one's gonna have to be handled delicately. So let me, let me jump in. 'cause I don't want, I don't wanna take the microphone from everybody else, but I will for a little bit.

This is like CNCF forming. The CNCF, with the exception of it has the characteristic you're talking about Cloud native isn't go, doesn't go across everything like agentic AI does. So I think that's inevitable that whatever you do, Alan, is, if you didn't set this up, it's gonna be that way anyway.

What are the things that you want to have common and done more consistently across the industry? And given the list of vendors that have signed up for this, it's pretty massive. Uh, very extensive.

You know, it could either be a big political fight or it could be really helpful, um, to the industry. So I think, I think the goal is to make this successful, and if it is, it will deliver about a lot of value. So I'll be quiet now.

I'll let everybody else talk. I think my concern is though, to Alan's point, is that if you look at the CNCF, there's all these projects, right? And the CNCN landscape is unreadable.

No one knows what all, all these things do. And all these projects have a life of their own, and they never get folded into any other project. And then the whole thing becomes unwieldy.

So, is anybody concerned that this is gonna happen again in the age of ai? I don't know. Fred, what do you think?

Well, I think there's a couple of the, it's a good, you know, how do we manage projects that have been officially open source is definitely one problem. But I think there's a second problem. There isn't a standard today, and this sort of codifies a place where the Frontier Labs can actually do the work and be accountable for that work in an open source way, right?

So you have the ability to understand, Hey, I have an MCP server standard that's required, uh, for protocol behaviors. You, you have some other things that go with that instructions that are required in order for me to build an agent. And the agent framework, right?

So this is, if anyone wants to walk into how do I do this as a business and now have a set of, uh, of, of conscripts in this project that allows 'em to do that. And I think that how that, how that materializes into other parts of the business of that I, that's to be seen. And I agree, a a two a probably, you know, belongs in here, but I would say it's the first consorted effort to, to get the technology into one place where people can use and operate that, where it's not, you know, at the behest of whatever technology organization is running it.

So there some standards are good, and a technical standard with open source projects is a, is a terrific way to allow people to, you know, get their feet into something they can measure between businesses. I think too, like what is the commercial? Is there a commercial, um, undertaking that is gonna be the counterpart to the open source?

And right now, I don't think there's anything, 'cause everybody's trying to learn everything. So I think it's pretty smart to put it all in one place, like Fred was saying, because if you don't, then someone's gonna come along and from a commercial perspective and see how to do it, and that's gonna be the next big thing. And that becomes the standard, but then it's locked up in private.

Yeah. Um, I also agree with, put everything in one place and also get all those supports from the most significant players in this ecosystem, right? You can see that he, historically, the biggest technology revolutions were never, um, about the products, right?

It's about the shared standard infrastructures. It's kind of like internet has a TCP IP or like smartphones has like OIS or like Android right? For, um, agents.

Uh, it's, it's, it's about the point. Like no one can avoid and we do need to have a standard way to regulate it. So, uh, so that in the future, the corporate adoption can make it better.

Yeah. I I I'll say this though too, at, at some level, this is yet another milestone in autonomous AI agent AI's, um, you know, move to, to, not relevancy, but to dominance, to, to visibility to the top of the heat. The mat maturation is part of the, the maturation.

But, you know, it's, it's a thing when the Linux Foundation creates a foundation just for you. That's a, that's a statement. It's a big deal.

So I am concerned about one thing, and we've seen this in the CNCF, right? I can't help but feel like there's an artificial cap on innovation when these projects go into the foundations because suddenly the vendors are there. Well, you know, we want this in the project, but not that because we monetize this as part of our quote unquote support service for that thing.

And, you know, it winds up ultimately being a force for stagnation. So I'm hoping, you know, it's still early days here, but I'm hoping this doesn't play out with MCP, which is kind of critical in my mind. Well, it's politics, but, and, and that's, you know, you get into these foundations and, and that's part of it is the politics Yeah.

Of a, of a multiparty Group. They're always ting. These, these foundations and projects always seem to be tinged with politics.

Oh. Especially based on the players involved, right. And what they, the standards they set or where we go.

But I, yeah, it's, it's interesting. This is kind of a transitional transformative moment for genic ai, which is not, we're gonna talk about this later, but it hasn't gained traction within companies as quickly as we had thought or we're told it was gonna happen by the vendors. Well, I, I think this actually is the, the wind in the sails here, right?

So if you're not sure what to standardize on, right? Having this as a defacto standard, uh, by proxy right, is a terrific way to allow organizations to be able to adopt it. Otherwise, you know, I think on the innovation front, Mike, I love it.

The, the, let's make sure we don't stifle innovation here. I think we're not gonna stifle it yet. I think what we've done is we've said, okay, look, we're putting a stick in the sand.

These are required. There's probably a lot more innovation. And every company has the opportunity in this day and age, right?

There's the saying that there's gonna be more millionaires made in the next three years over ai, you know, as opposed to the 20 years of the internet, right? Is a, is a statement that's been made. So I think innovation's probably still gonna flourish.

But the challenge is, is if we look at that rate of adoption, which we're gonna talk about, you know, that's, that's the worry. If we don't have the standards here, then, then we're at very worst not going to be able to adopt the things we have now. Yeah.

I mean, if I took away their phrase open source from this thing, and I said, you know, six vendors are gonna get together and form a governing committee to decide what think is gonna be the standard, everybody would help. So, Sounds like the Java Foundation, Well, at least they're All invested in each other. That's good.

But like people, I'm sorry, go ahead. Wiki. Yeah.

In reality, right? MCP is one of the most acceptable standard way to do things, right? No matter is like get into the open source or like nonprofit or something.

People support MCP in the best way. So I think it's easier to use MCP to standard to make it standard. And what will happen is the same is is the politics, right?

So you'll have all these vendors in there, and they'll start arguing that they need this and that, and when they're told no, they'll go and build a version of it of their own and they'll come up with something new and other people will come. So that, that's kind of where traditionally this has happened and what ends up being the standard is often the result of politics. Yeah.

And to your point about that, you know, Alan said mc P'S a year old. Well, from what I can see, MCP doesn't look anything like it did a year ago. Today it's very different and it seems to be updated every quarter.

So a year from now, I'm not sure what we're calling MCP will be the same MCP that we have today. And I'm not even sure then we backwards compatible. But who knows, If you wanna see politics go to a standard body that's real politics.

These open source projects are babies compared to that. All right. Hey, we're about outta time for this segment.

We've got a lot more to talk about and, and, uh, you know, do you trust your agent secret agents? Maybe not text on gang. We'll be right back.

You've earned it. The spotlight, the responsibility, the weight of teams, companies, and entire industries fall on your shoulders. Lives depend on your decisions, your home life included, that work your protected physically and digitally.

Nothing gets through your team without a fight. But in a globally connected world, everyone sees you, including those who mean to cause you and your organization harm. And now home your sanctuary attackers see an opportunity.

Your digital front door is wide open. And what compromises your home can breach your boardroom. Because the devil's greatest trick isn't targeting your workplace firewall.

It's convincing you that your personal life isn't at risk. Black clerk, digital executive protection, defending the new attack surface your personal life. Hey folks, we're back and we're talking about AI agents some more.

There was an AI conference this week in New York that I was at, and everybody was talking about, well, AI agents, including a fellow from NS two, got up and told everybody that, uh, from what they've seen so far, that maybe you should not give any data that you care about to an AI agent because, well, a hundred percent of the models and agents that they tested were all hacked. So it's still early days. And John has a couple stories talking about the fact that, well, AI agent is a lot of enthusiasm for it, but adoption remains relatively low.

That may not come as a surprise, but wiki, I'm starting to feel like maybe people are cutting onto this whole AI thing and that they're, you know, going slow because they're not wholly confident in what's gonna happen. Yeah. Uh, actually recently, we just had some discussion.

You our Yeah. Infrastructure, uh, gathering, right? Uh, we talk about this AI agent, why the adoption is, is so low at this moment, right?

Um, there are three things currently happening. Um, number one is people are not very sure, like, uh, if, if they adopt AI agent, right? Is does the AI agent can keep consistently perform what they try to do.

And then second one is, uh, there, uh, is related to identity and, uh, um, um, authorization concern, uh, 'cause currently majority of the, uh, AI agent, if you try to create one, uh, it will be based on the person's identity who created it. So, but, um, after a while, you see the, when AI agents start to work with each other, it's a little bit hard to identify, oh, if, if this is a personal, uh, behavior or like a, uh, it's a AI agent behavior. So there are a lot of, uh, new startups try to address this issue.

You can see also, uh, ServiceNow also acquired a, a cybersecurity company to address identity issues so they can adopt the ai ai agent more easily. Right? Um, and then also there's some like observability, observability issues, uh, like lot of, lot of work, yeah.

Agent do still in the black box, even though people keep the logs, but it's still hard to predict where's the fault, where something wrong. So that, I guess that's the three main reason it's a little bit hard to adopt AI agent in the co corporation environment. Plus, related to our, uh, first topic currently, we don't have a good standard and the regulation is on the way.

So everybody's trying to figure out, oh, if I do this, is it my responsibility? If if something happening and what's the consequence? No one know exactly.

Alan, this reminds me of your, the cloud's not secure. We can't use it. And analogy kind of, we're at that phase, right?

Yeah. So I'm glad to see all those, uh, new things happening, like the, uh, US National Institute of Standard and Technology that try to build up the, uh, uh, securing, uh, artificial intelligence agent dis discussion, right? And also for, uh, the, also in the, like hat Europe 2025, uh, O-W-A-S-P, they also publish the guidance, which gave people a little bit light, but this is still long way to go.

Yeah. Hey, friend. Um, one of the things that struck me about the Oasp Blist is that a lot of these hacks for AI agents are pretty trivial stuff.

I mean, it doesn't look like it's very hard. And, you know, it seems to me that bad guys can just kind of create a prompt injection and take over an AI agent, and they're off to the races in split seconds. Uh, there's a lot of it just the same as you would think about, uh, imagine that we didn't have any, uh, armor for our identities as humans, and now you have non-human identities that don't understand the context of why armor would matter, right?

Not, not relevant. md and the way that my, the way that my agent should operate, let alone what other influences can change that. So CEEs are on the table, uh, you know, the, the human boundary trust exploitation on the table, uh, supply chain.

There's, you know, there's more than 20,000 MCP servers out there. So, you know, direct compromise, indirect compromise adjacency, supply chains, like memory and context poisoning. Uh, there's tons of things available.

And part of what Wiki is referencing is really where we think about, you know, today we have this notion, uh, we talk about this thing called zero trust. And, you know, we talk about this other thing called identity and access management, where we have authorization and authentication that is in the just infant stages here with this. So even with those things, when we've delegated our credentials to an agent, that is also non-deterministic outcomes, right?

It's an absolute recipe for that. So there's a lot of people talking about it, for sure. And it is a widely, uh, utilized vehicle.

If you'll look at, uh, you know, like N-C-P-S-O has, you know, probably around 20,000 servers on it, and, uh, your guess is as good as mine, Mike, how many of those are real valid, uh, servers that are available for you that have been trusted, let alone what they do? So, uh, completely valid concerns on that. And I think that's part of why some of the relation on ways to do some of the CTF work around this becomes very important.

Mm-hmm. You know, Alan, I would love to get your thoughts on one of the things that the fellow from NIST said was that in effect, there's no perimeter anymore to defend then. So, Well, there hasn't been a perimeter since the Jericho Foundation went back in the box, but let, let me give you Shimmy's take on this.

Just, just as I, Mitch mentioned in the last segment, this is a sign of maturity. It's a sign of maturity. When you got all of this news about securing AI agents, this is like no one, no one, no one ever wrote about Max security when it only had 3% of the laptop market, right?

Max was in vulnerable because no one gave a crap. And it was only, they weren't a target. They weren't a target like Windows.

Yeah. Right? Now, Mac gots 20% of the, of the laptop of the PC market.

Well, all of a sudden, there's some bugs in that Mac os. The fact that this many people are, are writing about it is a sign of the success of Agen ai. Ai, right?

Secondly, a lifetime of learning in the security company, these guys could stamp their feet, hold their breath, and turn blue in the face. No one's stopping trains left the station, either get on the moving train or get left behind. So these are, these are, as we say in Vegas, a fine beginning, but you know, it's not gonna stop the, the forward progress of agentic ai.

And the problem is it's moving so fast that many of these things are kind of obsolete by the time we, we, you know, they decide on them and we report on them. Mitch, to your comment about the cloud, right? We shouldn't use the cloud because of security.

So, Although I, I wanna ask, but still, there might be some folks at ServiceNow in Salesforce and Adobe and whoever else is carting out these genetic ai, uh, projects with a little bit of concern. When I look at the 6% figure, the 6% figure that Harvard Business Review came to the conclusion that they found that only 6% of companies fully trust the agents to autonomously ha autonomously handle their core business function. So there is widespread enthusiasm, Absolutely.

But Judge, here's, here's The, but the confidence is scarce for now. It's the hippo. So that's, that's one of the underlying hypocrisies in the whole AI thing.

90% of developers use AI to develop code. 40% of them don't trust it. 65% of them think it introduces instability, but still 90% of them use it.

6% of the people using ag agentic AI don't trust it. Think it might be not so good. Does it stop them from using it?

That that's the, at the, at the nitty gritty core of the equation? Yeah, that's what we're dealing with, Doesn't stop. But it slows, I think it slows the adoption.

That's, I guess that's was my, is my point, is that we anointed 2025 as a year of a genetic ai, yet there are a lot of concerns that are cropping up. And you see them from studies, you see them from real life examples. So I'm just saying it, maybe we we're just maybe being more cautious, But you also see studies how many, how many enterprises have, whether they trust them or not, are using ag agentic ai.

How many agents are out there? That would be the question, right? Because I think number one studies are, are they're just numbers and anybody can make a number, say anything they wanted to do by how they represent the data.

Number one, I think what's happened, yeah, maybe 90% of developers use AI for whatever, but when it got back to operations, and we're the ones that are responsible for that data, and to make sure that things didn't get stunk, it stolen, that we can recover it, that it, we don't break any regulatory rules, all the rest of it, that's where it's slowing down. Yeah, it's cool. Yeah, it can make it do really cool stuff, but we wanna do it within a box that is protecting the organization and not just saying, we're doing ai.

It's amazing. So I don't believe the numbers for one thing. Uh, it, if you dive into the surveys, some of these surveys are kind of questionable.

Yeah. So I mean, yeah, there are, there are some infamously Lies, damn lies and metrics I know Was the, the, that was the worst. The worst.

But I mean, I, I, I also wonder though, like when you press companies to try to get examples of like significant use, it's usually at the very, very low level for now. But again, maybe that's just the process and they will accelerate and they will graduate. The type of uses, Uh, I completely, go ahead, Lina, please.

Just really quick. I just wrote a blog post about this. People are starting now to lay off and let go some of their low level, you know, like, like L one support, these kind of things, and what do we get?

We get these stupid agent bot that can't tell us a que answer a question and can't do anything. Even an L one could do Madden. So that's mad maddening.

I hate it. Sorry. No, that's okay.

I think in addition to that, and I, I agree, I think we've all had some occurrences where that, that that will naturally snap back. I think that, uh, the customer interaction with, with AI is a very defined thing that OpenAI thinks they're gonna own. But anyway, the, the, the behavior of utilizing, implementing and then fully adopting, I think is an enterprise.

I mean, it's a very long curve as it is. And if we see right now that, you know, 86% of those organizations pulled know that they're gonna make future investments and so on. I mean, they can't do that without standards.

That's now something they have. And now we can start going through an adoption process. But, you know, whether or not it's the core functionality of the business, I mean, I think we treat everything with the same kid, gloves of the core function of the business to hand over in any way, shape, or form.

But I, I, I see rapid adoption in a lot of the areas that are, you know, the workflows and the behaviors, writing code, testing, validating code, validating vulnerabilities, uh, you know, there's a, a massive amount of multiplicative, uh, benefit from being able to do that with agents. Uh, there's a, a guy that, uh, used to, used to work for me that operates like 10 engineers now, and I mean 10, right? With the number of agents and tasks that he is able to perform at the same time by coordination.

And when we think about the effect of what it means to be able to do massively more work for folks that are capable of doing it, it only can have that type of effect down the road. So it's exciting and it's early days. You guys are all, I think, a hundred percent on point.

We should question all of the process to get there, but next year is going to be that year, next year. Yeah. I, I guess that's why they Start the foundation now, In part, you know, at every one of these tech vendor conferences, whether it's Reinvent or Google's or who you, you name it, they spend an inordinate amount of time with two, three customers essentially doing a use case of what they're doing with agents.

And of course, you know, they, they're positive stories. I'm not gonna put the bad ones up there, but the vendors know the industry needs some examples, successes to build on, and we will see more and more of that. Yeah, I think, you know, to Fred's point, there's a lot of people who have awesome skills that to do that, but I think that they're a small percentage of the overall population, and it's gonna take a while for the mere mortals of the world to get the cognitive capabilities to manage 20 agents to run a process.

So that's just gonna take time. Yeah. I I also feel like there's a huge market about this agent, right?

Even though the adoption is low, you can see there are a lot of companies, they have huge needs, but they don't have enough budget. That's where the agent will play the role. What's a block here?

Currently, the regulation compliance and standards are not clear, right? Yeah. So I'm not sure, I'm not so sure how cheap the agent is either if I can have to keep paying for all the tokens to use it.

But we'll see. But I think also, Mike, to your, your statement, I think there's a lot of people that are stuck in jobs where they could do a lot more, but they're, what they are actually able to do is just burying what they're cognitively capable of. So if it's, if it's this very low level stuff, if, if people can wrap their heads around that and, and elevate people to the point of, okay, this boring bit we've automated, now you use that to do the things you've always wanted to do, then that could be pretty amazing.

Fair Enough. I just want an agent that says, I know you do this a lot, a lot of this is action. Here you go.

I've, I've automated it for you. Thank God, thank you very much. That's behavior number one.

And, and I would love that. But if I ask you to do the same thing and then tomorrow it will do it differently. So Yeah, it is a little bit of a memory problem, doesn't it?

Persistent agents. All right, let, let's take a break here. We'll come back.

We've got our C blocks still to go a little bit off of the ai. Well, it's still ai. What can I tell you?

It's Techron gang. We'll be right back. 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.

Hey, everybody, we're back and, and write this date down because I'm about to say something that I rarely say, but it looks like Alan was right. And we are seeing this innovation in the AI space with data centers and energy and consumption, and we've seen some stuff from an outfit called Palantir that's pretty famous in Washington circles, at least for building out AI applications. And they have a chain os that's gonna help us maybe consume power more efficiently in our new modern AI data centers.

Gina, what do you think? Well, let me tell you what it is first. So volunteer is partnering with Nvidia and CenterPoint Energy, which is really as a Texas resident kind of gives me a pause for concern.

But, um, they're to build the, to, to use something called chain reaction, which is, uh, I guess a service forum, um, from Palantir to, I'm just gonna read what it does. It will accelerate Nvidia AI infrastructure in installations across the US by streamlining the complexity of managing the complex supply chains, supporting what power is needed for the gigawatt ai, um, data centers. Um, and the, the thing it will do, hopefully is look at where the, the blocks are in power distribution, construction, all the, all the, uh, the, the, the legal things that, that the municipalities have and data center operations.

So this reminds me of a, and I couldn't find it. Um, of course I wrote a whole bunch about it, and it was probably in a chat about something. The US government just had an, uh, uh, the president signed an, uh, executive order to get this done, to get the data centers built to, to do better with data center.

And it's all resting within the Department of Energy. And of course, the Department of Energy is all run by people who are oil and gas. So my, I'm really concerned, like as, as just a Texas resident, um, who their, uh, center point is in the Houston Cold Coast area, and they have been horrendous, just like the entire Texas grid is at keeping power on during, um, emergencies.

I'm in Austin and I lived through the ice storm ice apocalypse a few years ago, and it's crazy. So the last hurricane, big hurricane, um, they had in Houston, they lost power for weeks, some places, um, they just weren't able to get it back up. So this all is not, does not seem to be about that, though.

It doesn't seem to be about keeping the local, um, grids, firming them up, modernizing them, all. The rest of it, it seems to be about having a separate, um, roadmap to building the AI data centers. And so I a little concerned about that because it doesn't mention anything about heating, it doesn't mention anything about cooling, the amount of things that'll be pulled away from the resources that'll be pulled away from local, um, in local areas.

It's just about building AI data centers, which are important and we all want our agents to happen. We want to move forward into the next century. I'm not sure this is the same as, uh, building the atom bomb, which is what it's been compared to at Palantir levels and government levels.

Um, and I'm not sure if you look at all of the information, which is very kind of scarce actually about what they're planning to do. Um, it's, it, it's not very, and it's not about anything but building out for AI data centers. So I was probably the wrong one to throw this to because I have a totally, I do not have a, I do not have a, um, I don't have really a technical response for this.

I know we need the power. I know we need everything we talk about, but it's being put, all of these data centers are being put in areas that are either very, very poor or very, very remote that in Texas do not have the infrastructure for this, that it will have to be, will there a second, infrastructure will have to be built that all of the towns and cities that these are affecting are being promised the moon with jobs. I hear one more meta commercial about how many jobs these data centers are going to bring.

I might shoot my radio, so it's not going to do it. It's, it's dangerous to our water. It's not supporting anything local as far as power grid, which in Texas is a really, really big problem.

And I don't like it. So I think we should bring out into the open, like, what is this doing? How is this, I, we know there's all sorts of things that AI is impacting, but, and we want to, we don't want to stop ai.

We want to build the data centers. We want to have the things work, but are we doing this in a way that is going to disenfranchise the people that are most disenfranchised now because they're not being on the level and the marketing is so slick to make it seem like this is a wonderful thing. Tina, you're in good company.

I don't understand it either. I, I read it over. I'm like, what exactly is This?

Well, let, let me see if I could, let me see if I could illuminate it for you, Mitch. Gina, please do, please do. This is Peter thi who owns Palantir preaching the gospel of whatever frigging gospel he's been preaching lately to ride the, the teat of the government's dollars as we bite around the edges trying to squeeze out one or 2% more efficiency by burning our fossil fuels to, to power these AI factories because they don't wanna acknowledge that solar and wind and renewable energy is the way to go and that these aged and that these data centers are gonna be manned by robots and they're not really gonna have a lot of jobs.

So if you want the truth, that's the truth. Gina, you weren't strong enough in saying what it is. Thank you.

You know, I'm glad you, I'm glad you pointed out your, your your I'll leave it right there. Yeah, because This is like, this is something that's happening in community after community across the country is Ohio, Pennsylvania, Texas, what, what have you. I mean, it's the same old story and I think Al nailed it.

This is just a very cynical approach. Of course it is. I know It seems, seems to me we need another foundation here to solve this one, Need a revolution.

Field foundation will take care of Everything. Well, but from like a in investment perspective, right? Currently we do need some sort of new infrastructure to solve the problem, which, um, plan plant, uh, planter solving, right?

We need the full, uh, whole set, um, on the, to, to make things quicker to solve the bottom neck problem. It's no longer the algorithm, it's power land, the transformers, chips, construction materials or like great inter in interconnections. I think they're doing, uh, the right product for the, for the market.

But you can see like different companies, they react differently to try to address this needs, right? One is like this way they build up the OS to get a, like a new different, um, physical backbone on the ai uh, AI stage. And you can also see company like x they try to go to space, make a new different market to address this problem.

Everything, uh, every, everybody react differently, which is very interesting to say. Yeah. So, so lemme throw, lemme throw a couple things on here 'cause uh, it's a contrarian perspective from both what, what both you and Alan shared.

So, uh, the first thing is, is, uh, I I want to make sure everybody's super clear that we did not build the power infrastructure for today's usage alone today, right now, right? As a person that has spent a lot of time on operational technology and security of such things in planes, trains, automobiles, cars, ships, all the things, uh, it is currently not built for what we do. The second part of that is we talk about the resiliency requirements of what power is now, let alone what power needs to be in the future.

It's not optimized at all. If you look at the amount of electricity and power loss in major cities, 40 to 60% of that, you look at fifth wall statistics is caused by old buildings, old lighting, old electronic infrastructure that is not managed and is just electronic loss. So when we think about resiliency of the future and the optimization efficiency requirements, it only makes sense to do that.

Does a a rising tide raise all boats? Uh, we would like to think it would, right? And, and part of that is a requirement to do such a thing.

The US Navy signed a thing with Palantir in the same way to have react to some of the, the reacts to some of the ship Os for them to build the supply chain parts of that. 5 trillion. So doing that effectively, efficiently with operational, uh, crispness is what builds the resiliency we need for the future.

I don't know whether or not Palantir's gonna solve all those problems, but I guarantee you it's a problem that needs to be solved. I agree with that. I don't disagree with that at all.

But what I disagree with is, number one, the marketing for this solution. 'cause the places that I found the most information, um, were on the finance pages. So you're absolutely right.

There's obviously a huge, um, a huge, um, market for the transfer of energy, which of course, Texas is not part of that market, which is interesting why they put CenterPoint as their main partners. That's what I'm trying to say, right? So, um, but yeah, I mean the en the energy problem, uh, all of us know that.

All of us know that, especially if you live in a place where we're not on that grid. Everybody I know has a generator. Everybody I know has portable backups.

Everybody I know. I mean, like when you go for a week in Texas dealing with an ice and snowstorm with no electricity, it gets pretty hairy. Like really, really quick.

It was deadly. So, absolutely. And, and that's just, that's just like one place.

This is happening all over. So I agree with you, and that's kind of my point, our entire grid is it needs to be updated, things need to be up updated. You need, um, all of the municipalities needs something.

I don't know if, and Palantir os is like, there's, there's no competition to it. That's the other thing. And that's the only thing that's being bought.

Hang on. So, so, so why aren't we fixing the entire grid to also provide, to make sure as we fix the entire grid to provide that, make sure that it can provide the excess for what we can see plus what we will see 20 years, 30 years, 40 years down the road. That's not what's happening.

They're concentrating on building AI data centers out. That's what they're working on. That's what this is for.

And that's not okay. It Needs to be, but if that's the engine that pulls the train, that does improve the whole thing, that's not a bad thing. My my point though is, Gina, what you're describing, Fred, what you're describing is not gonna be solved by a smarter os helping us try to manage what is an inadequate system, right?

That was the whole point I thought under the last administration, this whole infrastructure package, where we're actually going to address the fundamental issue, which is we don't have the right infrastructure, right? You could, you could marginally improve, you know what, what, what we is a generation or two old right now, or you could, you know, it reminds me when, when, when cell service first came out, right? And internet access here in the us we had pot slides, right?

And once a matter with a plain old telephone service, right? Where like places like, you know, the tigers in, uh, in Asia and Korea and Singapore and some of these places, they didn't have that existing infrastructure that they tried to speak every last dime out of. They went and built fresh and new.

So broadband was much easier to deliver there. Meaningful broadband, not, not DSL and stuff like that. And we, we almost need that kind of effort, I think on our energy grid where we, it it's not just incrementally squeezing another percent or three outta what we have.

We, we need a fundamental infrastructure built here for a gener for the next generation, not last generation, not this generation for the next generation of power consumption. I Agree with that. I a stupid question friend.

Why don't we just build a grid for the AI data centers and let them have their own grid? That's What they're doing. Uh, well that it's starting this way, but I mean, I I would also caution there too, right?

So when the internet was invented, right, it started off in very isolated points, the way that organizations like IBM and, uh, and, and some of the other bigger companies that drove both adoption of servers and infrastructure and connections. When I ran a data center at IBM, right? I would be out at every Sears store in Washington, New Jersey, New York, all of the things ran through those transits.

And those transits were not owned by the federal government, right? They were owned by an organization that invested in the outcome. And Palantir is no different whether or not they support the infrastructure for building these data centers, right?

The same thing. A rising tide, uh, raises all boats. What happens is democratization of what happens to, you know, the access to those things.

And I fully agree with you that that's exactly what should happen. And super sensitive to the fact that you also went through something very specific in a place that is the origin of energy, right? Uh, especially if you watch Landman.

Uh, so when you think I love Landman, It's great. Uh, the, the, the biggest challenge I think is, is what is the rate of adoption? Let's, let's let, let's let them get a prototype out, right?

Let's get a prototype out and see what that looks like and see what can be adopted there. But I think the hard part is if, if they're going to build a net new grid, you can't just start off with that intent, right? We would all agree that if you started off to build an entirely new national grid, you'd probably fail in today's what?

That's too much. And it's also too much money. And so the way the grid works today, right?

May not be the way the grid needs to work in the future, right? So today it's not a mesh, right? It's, it's kind of a mesh, but controls flow down in this particular case, maybe they're all adjacencies in the same way we talk about MCP servers.

Maybe. I don't know, Fred, I I think that's a great way to end this. You brought some reason to the chaos here.

Thank you for that. But guys, we're outta time. I know we can talk about this stuff for days, but we, we can't, um, we all got work to do, including you.

But hey, what a great way to end the week here on this Friday. As usual, we have Techstrong TV immediately following this. You could check it out.

Or if you're watching this on demand, maybe on the OTT app on a big TV or wherever you're watching it. Thank you very much. Stay tuned for Textron TV gang.

Thank you. It's, you've been great today. We'll be back Monday with Fresh Textron Gang.

In the meantime though, have a great weekend. Get your shopping done and enjoy. This is Alan Shimel.

We're out.