Techstrong Gang – May 22, 2025

Hey, is platform engineering just for cool kids? You're watching Textron Gang. Hi everyone, it's Alan Humma.

Welcome to our Thursday edition of Textron Gang. You know, I love doing Thursdays on Textron gang 'cause it means tomorrow's Friday, and especially for me, I've got about another week before I head out to Italy on vacation. So I'm in, I'm in.

Let hurry up and let's get there mode. But there's so much going on. Everybody and their mother is having a conference this week, it seems.

We've got people all over the place at conferences. There's as usual a lot of AI noise, a lot of just a lot of stuff going on, and we've got a great panel to, to discuss it with you today. Let me introduce you to them first, I wanna introduce the new kid on the block.

He's not a new kid to me, though. He is been a friend of mine for, uh, 15 plus years, maybe more. Uh, he's a security person extraordinaire, uh, real like, and I've, I've seen him rise up from the, from the trenches of security to cso to entrepreneur founder.

I'm, and I'm thrilled to have him on here, my friend Fred Wilmont. Hey, Fred, how are you, man? Yeah, Great brother.

Thanks for, uh, thanks for having me. Just real, I, I'd hope I didn't embarrass you, but give people just a real sense. What are you up to these days?

Uh, so I have a company, uh, called a tech team. We're building, uh, a way to shift the detection engineering problem into a continuous and autonomous delivery mechanism instead of the static rules written by experts. And so think about that as reducing this problem of complexity and tech sprawl and, you know, uh, intellectual elitism down to 15 minutes of adversary to detection, deployment.

And, uh, yeah, it's, uh, it, it's been a lot of fun and I think, uh, always appreciate, uh, the, the support and the, you know, the feedback loop. So thanks for that, Shannon. No problem, Fred.

All right. Also joining us. I haven't had the pleasure of seeing her on for a while.

She's been on, it's just, I've been off. Um, our friend, I assume you're home in Austin and a whole ward. Hey, Anne, how are you?

Good, good. Great to see you. It's been A while.

It has, and I'm glad you're here. Then joining us from his perch up, uh, in Silicon Valley where he, he sees, he sees he's, he's the master of all he sees of his domain there. Our, our, our Silicon Valley editor, John Swartz, you just keep, keep Laying on.

Yeah, I, I'm good. I, Alan, I, I have to admit, I'm playing a little hurt today of this allergy season, but like the great Willis Reed, I'm hobbling out of the, the tunnel on one leg to perform today. The great will bring up those Knicks.

Baby. Let's beat Indiana. Um, from Kne, not from Connecticut, from Colorado.

He, he's the guitar guy. He's a future VP, analyst DevOps, Mitch Ashley Mitchell. Good to see you Fresh off the plane, uh, from Microsoft build.

Good to be back in Colorado. So Absolutely good to have you home. And then running the anchor leg in this six by hundred, I don't think they have a six by hundred, but he, he's also recently also home from Boston where he went up not to see a Red Sox game.

I'm still in Boston. Oh, you're still in Boston. Maybe you are going to see the Red Sox.

I don't know. You Benedict Arnold, you our chief content officer, Mike Ard. Yeah, it's a, it's a pleasure.

You know, I went to college up here, but I'm up here for this Red Hat event and I saw some old friends that I haven't seen in about 20 years. But I gotta say, you know, when I was in college, they promised me flashbacks and they haven't arrived. So how There, there you go.

There you go. Alright. Hey, let's jump into things guys.

So, um, Mitch, you, you did a report, uh, I think it's on the fu of intelligence panel, but it, it basically, you know, talks about platform engineering seems to be gaining real traction, especially in organizations where you've got cloud native AI enhanced cybersecurity. Mike, you wrote an article on it. I, I put up something too.

What's the story here, Mike? Yeah, I mean, it seems like from the report, Mitch, that, uh, to Alan's point, there is a lot of traction being gained, but it was interesting to me at least, the drivers are different. You know, you hear a lot of folks talking about cost and then a productivity, and it seems like it's just a smorgasbord of different things that are all coming together.

But what's your assessment and where are we on this journey? Well, you know, Mike, uh, platform engineering is, has a characteristic similar to DevOps, is that it's pretty adaptable to what you need and what you do in your organization. So I think it's, that's one of the reasons why there's such a variety of, of technologies that are involved in, in what is a platform, whether it's development tools for developers and how that shifts in the age of ai.

Um, but also, you know, Kubernetes and where we run that everywhere within the organization or, um, whether that's AI workloads or, or, or microservices workloads or other things, because of course we're using Kubernetes for so much. So one of the things this report did is, um, we, we do a, a decision maker server survey, excuse me, that we update frequently. And, uh, I, I did wanna really look in kind of a comprehensive look across the software development lifecycle, but particularly honing in on platform engineering.

And one of the reasons is because when you think about the platform engineering market, it isn't just tools that platform engineering might use like an Ansible or some from a Red Hat or someone like that. Um, it's all the technologies that they manage and work with and then have influence over buying. Um, our, our market estimates, just looking at platform engineering as a kind of segment itself for about 15, almost 16 billion, uh, in revenue by 2028, when you layer on the other technologies, you know, cloud native tools, software security, platforming, uh, platforms that they might support, you know, it just depends on what your layer in, but those being core things that they do, I mean, you're talking about 150, so really 10 x, um, of what, a billion of what they really influence, uh, in the organization.

Well, guess what? Vendors recognize that. I, I tell you Alan knows this, you know this.

How many people do we talk to that say, yeah, we, we we're, we're targeting or we're pivoting to talk to platform engineers. And that's why, because they play such, even if they don't do the purchasing, they're the folks that are gonna implement, they're the folks that are gonna be part of the selection, maybe strong influence over what those selections are, Especially cybersecurity. Right?

org. And I, I asked them this question, Mitch, right? What, what, what's the, what are the key indicators where you should be using platform engineering?

Is it should that you're doing digital transformation? Are you running Kubernetes? And those are kind of obvious, but he said, Alan, it's really the biggest indicator is your complexity, right?

If your infrastructure, and I don't care whether that infrastructure's in the cloud or on-prem, if your infrastructure, your IT processes, as the complexity grows and, and complexity grows, it's almost like the, you know, the lower of entropy in the universe, right? It, it's more complex lot not less complex. Uh, as your complexity grows, the more you have a need to bring order to chaos to try, uh, 'cause otherwise the complexity will eat you up, right, Fred?

A hundred percent. Uh, a lot of the challenges that, you know, we run into today, or, you know, uh, the last handful of companies doing this kind of work is really about the technology that we have, the technology we're integrating with, and also this thing called agility. And from a cybersecurity perspective, uh, there's a lot to be made of whether or not, you know, MCP does this and authentication au the off the authentic net.

That, but the complexity of being able to interconnect all the technologies to perform these functions is dramatically lower when you can, you know, sort of box it into a couple of areas and continuously redeploy it. So there's another set of hands on how many people touch that code. Um, and also the frequency with which you, you deploy.

So, you know, the historical thoughts on, well, we'll, sort of fang universe, we'll, we'll redeploy every day, we'll, we'll deploy 10 times a day. Um, I think the challenges as that complexity and there's drift involved in that complexity, uh, starts to take more shape. Uh, the AI portion of the universe that's starting to creep in is dramatically driving a shift.

And so my question is from a cybersecurity perspective, uh, you know, do we see the end of, of engineers and only have platform engineering? Hmm, I don't know you, I mean, you come up with a good question there because I'm having the same conversations here at Red Hat where if I describe platform engineering to DevOps engineers, they kind of listen patiently, and then they go, yeah, we do that. But they feel like maybe they've been doing that all along.

And I think it's different opinions about what the level of complexity is and, and what that mission statement is. At the same time though, it's pretty clear that there is an effort to centralize DevOps going on, and people are trying to figure out how not to have, you know, nine different CICD platforms running everywhere. So there's, uh, a drive from the executive side to kind of embrace something called platform engineering that brings some mortar to the chaos.

But I don't know, Mitch, what's your take on what is the line between what we're calling DevOps and platform engineering gonna be? I think platform engineering covers a wide variety of areas, and a lot of organizations are subsuming or incorporating platform DevOps, engineering or DevOps into their platform engineering teams because they're maintaining tools for developers now, and they're obviously maintaining pipelines, so things that are happening across the development processes. So it, it's a pretty natural fit, especially if you wanna centralize.

And I think one of the benefits, you know, we think about central centralizing and, and, uh, focusing on costs. And certainly cost can be a reason for, for doing that. Uh, another is that you can take a handful of, you know, talented people and they can in platform engineering who were doing DevOps too, or maybe a combination of both.

They, they can serve as a force multiplier in the organization. You know, if you're improving the productivity of, of lots of people, that's a force multiplier. So it isn't just cost, and yes, it's costing all the time.

All those other folks would've done the same thing or duplicate over chaotic kinds of things that overlap. Um, but I think that's one of the bigger advantages is that you can actually be more effective at delivering software, operating software, doing upgrading, addressing some of the toil things like that. You know, I, I, here's what this shimmy's take on it.

I always thought of DevOps, right? As you know, the event horizon, horizon is deployment from coding to deployment, live the land of DevOps, right? Middle earth.

Beyond that event horizon, we DevOps engineers didn't really see, or the DevOps folks didn't see too much other than, you know, uh, what we like, uh, what we used to call, uh, uh, application perform a PM application performance management. But basically it was feedback loops post that event horizon. We really didn't have a lot of control.

That was the land of the SREs. That was a different trilogy and you know, but we get some feedback loops that we would put back into our development process. But clearly the event horizon was deployment.

So from d and when we shifted left, we shifted left further into the point of where we develop code, right? And, and it went that way. I think of platform engineering as their event horizon is where we start coding everything.

Platform engineering does takes place before we start coding by putting everything in place for you to go code. And this is a very simplistic view of it, I realize, but that's their event horizon. They work to the left of where, where coders code, they're putting everything in place so that when coders code, it's in there.

I think another really good way of looking at this is, hey, soft, the software game has changed. How we develop, deploy and run software is very different than it was when I first, you know, Mitch and I are, it's still secure. And we, I met Fred back then today, software very much is in a factory.

And we got a lot of factory workers like you do in every factory. The platform engineering people. They kind of design the factory, if you will, where this machine's going to go, where this takes place.

So You're saying, you're saying the Shire is like, before we code, and what is CICD? Is that Mordor? Is that what that Kind Yeah, kind of.

Yeah, if you're gonna go with this Tolkien thing, you know, lot of the rings, but you know, but that's where it is. But DevOps engineers work in the factory. Developers work in the factory, cybersecurity folks, DevSecOps folks, they all work in the factory, right?

And that's, that's the modern, and I know a lot of us want to say we're craftspeople and it's still a craftsman and you know, where we're like, we're guild members or something. But no, it's a factory and I think that's where we are. You know, it's funny you said the factory because that's something that Microsoft was bringing up, um, about, you remember early on Gates talk about the software factory.

It has been in the eighties and nineties, and now that's kind of, it's coming back around of thinking about this with, with AI and agents and capabilities of it becoming even more of a factory metaphor because it isn't as craftsman like, um, as you were describing of, you know, slinging code and Things like that. No, it's not. I mean, we used to think of it as like a, you know, small craft shops and like, you know, the gilderman and stuff.

But it, it very much is, and just like in modern factories where we have robots and ai, you know, empowered robots building our cars and our appliances and everything else, we're gonna have AI in this factory too. I think the, the interesting part here is dev to, to me, right? Uh, DevOps and platform engineering are now the same role.

A hundred percent. You think They are think sounds a hundred percent the same Overlap. Absolutely.

Absolutely. And I think part of that, if, uh, I might step out of the Tolkien sort of analogy and maybe into, you know, if you think about an orchestra and a conductor and, you know, all my platform engineers are also all my DevOps guys. They help orchestrate all of the guys that write code from cradle to grave in the development lifecycle from, you know, the first steps of, well, you need to use these tools in order to get started here all the way through testing, validation, you know, production, deployment, and by the way, you know, operational response when your code doesn't work the way it should.

And if everybody's on the same page, right, to stay with this analogy here, then you know, you can actually make some music. Otherwise, you know, there's an awful lot of friction in, you know, development style or culture. And I think that's, to me, they're the keepers of the culture on how to do that effectively.

So you can have measurements like agility and, you know, you can time to, you know, remediate something by redeployment, by, you know, knowing everything that a particular type of service touches. And that's kind of what I would say is the bedrock for doing that successfully today. So my thoughts, they're all the same.

It's the same role, it's the same group of folks. And, uh, they, to Mitch's point, they have been doing this a long time. I just wonder if we're starting to see some more separation between what we call CI and cd, right?

Developers do most of the CI stuff, and maybe the platform engineers are gonna handle more of the deployment side of this equation. 'cause there'll be more standardization on the platforms, which is enabled by Kubernetes. Um, I think previously we had, you know, all these Snowflake platforms, so it was really hard to be a platform engineer.

But I think, you know, there's a reason why platform engineering and Kubernetes are going together hand in hand. And I think that that separation is a good thing. 'cause to be honest, I think we've sucked at CD forever.

So let me just be clear. You don't mean Snowflake the company. You mean snowflakes, like they're all individuals.

Exactly. Gotcha. Just, just check it.

Well, it, it, it certainly, you know, look, I, as I said, I was just on with Luca doing this platform engineering podcast. Look, they're gonna have 35, 40,000 people registered for Platform Cart. So whatever it is, it's pretty popular, right?

That big community big, Yeah. If I can just put a plugin, um, the, the resources behind this report comes out on a, on a portal or a platform platform, um, called Future in Intelligence. And we, we issue some reports that are a subset of that.

Um, and there also are vendors that subscribe to that to get access to analysts. And one of the things is that it's nice about it is this isn't like Gartner Forest or pricing, no dig at them, but we know they're not cheap. This is something that, you know, people in the tools business and the software, the Kubernetes or whatever, um, can easily have access to it.

So, you know, reach out to us. We're not salespeople. Reach out to us if you're interested in getting more in depth and access to analysts, and we'll, glad to help you.

Yeah, I, I think I have links to it in my, this platform Engineering for Cool Kids article on, uh, on the site. Anyway, hey, let's take a break. We're gonna come back and talk about, it's a Google AI world.

Well, it's an AI world for everyone, but what makes Google special here? But of course, they've got a conference going on too. You're watching Textron Gang.

All right, folks, we're back. And as Alan alluded to, there's more shows this week than you can shake a stick at. And of course, Google joined the party as well.

And John, you covered this whole thing, but I was reading your article on Textron ai and my eyes started to cross. I mean, they were sprinkling AI in everything and including the glasses, That was their goal, I think. So yesterday we talked about this, we talked about, uh, setting a narrative among all these AI shows or events, and we kind of came to the conclusion that Nvidia probably does the best job because they've, they've corralled this, this, this, this constellation of, of content creators.

So they form a story and you can follow it. Not so much with Google. I I actually came away from this event.

It wasn't just a kitchen sink strategy, Mike, but I think what they were trying to do is give their company an AI makeover, essentially, when you add it all up. And in a sense, I think what they're moving towards is, is making, uh, Gemini a AI operating system of sorts. So yes, there was a hodgepodge of, of stuff that was a fuselage, it was a fire hose of products.

So the, I'll just mention a couple of them to give you an idea of how it pinged and ponged. There was no logical, uh, flow to the conversation. It was a two hour morass of news and services and, hey, this is cool.

Look at this. Oh, wait, wait, now look at this. So there was, um, there's a new tab in Google search, something called AI mode.

So you could post longer queries, you can post up to 10 questions at a time and then follow up and then post follow up questions. 5 Pro deep think, which they're talking about as an enhanced reasoning mode. And they're making the comparisons to what it can do compared to open AI's oh one Pro or oh three pro models.

But they didn't provide more, more, uh, context than that. Um, there was the DeepMind, CEO who referred to this goal of making Gemini a world model that's capable of making plans and imagining new experiences by simulating aspects of the world, just like the brain does. You know, the, you're getting the gist.

They were the Google Glass, the return of Google glasses, um, something called an Android XR project. So they're, they're partnering with Warby Parker, Samsung, and others. Um, there was a, uh, new asynchronous coding agent, jewels that would conceivably compete with Codex and, uh, copilot.

So again, a lot of news, a lot of scattered attention, a lot of oohs and ahs. Um, it's, it, it, it was, I think the takeaway for me is that we are going to mix AI into every conceivable corner of our business, and we are going to fundamentally try to stamp ourselves as the IO company. Although I think in a sense, and I said this yesterday, I think Microsoft probably did a better job of, of making its its pitch as, as Nvidia did.

So, So, so I, I have, I have a selfish question though, upfront. So all six of the people on this show are wearing glasses. Do I need a, can this thing work with prescription glasses or is it just like, I'm gonna have to like put it over my, can you say that Again?

I didn't quite catch that. They never, ever do. And you, and in a lot of the headsets, you actually have to have a special bezel to, to accommodate glasses.

So like the Apple Vision Pro, I was trying a friends and I, you were there, John at South by Southwest last year. And I, I had to do it blind because there was no effective way to wear my glasses with it. But, but back to Google, I, I was genuinely curious how you were gonna summarize this conference because it was just like a series of overlapping announcements.

And some of the products were eerily similar to others and had fun names like Project Mariner. Um, I really love Casey Newton's one line on it. I'm a big Casey Newton fan.

He said at Google io everything is changing and normal and scary and chill Because last year they were pitching us on the web, we're gonna use AI to read the web for you. It's gonna read you, you just hang back guys. And it's cool.

That was last year. Remember this year was like, well, you're using AI and you're gonna use it for everything you do, and here's why. And they were obviously trying to catch up to chat.

GBTI was an early fan girl of Gemini because I liked the integrations, I liked that boom, it was right there at my Gmail maps, like all the stuff, and I'm on, you know, Google everything right for business. And it fell behind very quickly. So the deep seek announcement and the nod from the deep seek people were obviously them trying to say, Hey, we're as good as chat Bt now, by the way, it's not, We don't have any, we don't have any imminent news for you.

And it was like, the whole thing to me was like, ai, A DHD, it was just, it was, it was spattered. It was lazy. It was lazy in that they didn't bring it to a finer point for you to say, this is the killer thing because there was no killer thing.

It was a lot of, But ain't that the truth about AI in general? What's the killer thing? Yeah, that's, that's what everybody's in, in, in pursuit.

So I mean, look, the, to me, this is a kid to making a party. I love making a good party. But what they served up here was a veritable swarga sport, right?

But there was no main course, that's what you're saying. There was no entree. It was a smorgasbord.

Everybody was finger food kind of stuff. And it didn't all necessarily even come together. They had their pasta station here, Oriental station there, maybe lamp, you know, lollipop lamb chops over there.

But There was no, there was no gba goul. There was no gba Gba goul gba goul. Not gba goul are you saying?

But anyway, are you saying It was a, are you saying it was an AI Potluck? Yeah, it was an AI potluck. That's a good way of, but let me let, I mean, the glasses were the most interesting to me.

Yeah. No, but let Me just be clear. 'cause first of all, and I've played with the meta goggles, and they do have an insert that you put in, if you have glasses on that kind of keep it off you, but that's not the glasses we're talking about, right?

This, those are different droids. They're these glasses that just have a recorder in them. And I will tell you, you know, every time I go into a RayBan store, I get fomo.

I try them on and I wanna buy them, but I need readers in my glasses, sunglasses, otherwise it's hard for me. And you can't get 'em with readers. And you can't get a prescription lens on these Google, well, I don't know about the Google glasses, but with the Raybans, she can't RayBan's meta, by the way, it's a meta glass.

They did a, they did a demo backstage. Um, it was pretty choppy. And it, the, the latency wasn't very good.

And, and for some bizarre reason, um, for you, NBA fans, Giannis was part of this demo. He, he showed up and said something. I, Well, he's looking for a new contract and a team, so this, he's gonna be out there.

This, This is a conspiracy to make you get laser eye surgery so you can have 2020 vision. So you can these glasses. See, that's how it works.

No, But not the glasses just gonna read it for you. You don't need you. But seriously, that's one of the things I, I bought, like, again, what's the killer app here, right?

These glasses. So, can I go Ahead. I'm, I'm late.

Later that day, I went up to San Francisco to moderate, of course, you panelist and an autonomous, autonomous panel and technology. So I was on a panel with these guys who are, uh, it decision makers at Wells Fargo, Barclays, and then a guy who was a, at a startup. And I mentioned the Google IO announcement.

And, and they all looked at me and said, what was it? And I, and I said, this is why I want to ask you about this. How do you make rhyme a reason when you're supposed to be using AI agents?

Eventually there's all this pressure on you to, to, to enact these plans. And they said, this is adding to our uncertainty and our doubt is like, we're not getting a clear message from a lot of these companies. And to us, it was just like, as you said, Alan, a smorgasbord of, Hey, this is cool.

Wow, look at this. Oh, this is kind of similar to what we did a year ago or two years ago. We, we mentioned and, and, and, and that's, that was their takeaway from that.

I mean, I, they have different opinions about other companies, but I think we are, we, we, um, we're in this process of falling within this trap as the hype continues and companies feel this urgency. I I'm talking about the big tech companies feel this urgency to prove themselves. And I'm, I'm wondering, teeing me up to what, what happens when Apple comes out and ww DC in June, and what are they gonna talk about since they're so far behind with Apple intelligence?

That's gonna be interesting. That may be the vacuum they were attempting to fill, right? Maybe I they always are.

Yes. That, yes. So I guess the future is, you know, we used to say there's an app for that.

Now we're just gonna say there's an AI for that. Is that There's an agent for that, Right? And then I notice none of you have mentioned Grok has, have any of you even tried grok?

I I don't try grok on political grounds. Well, I separate the art from the artist. I, I have to, I have to because, uh, my, my profession requires me to be on Twitter.

So I, I've had to go that route. I separate the art from the artist, but I would say Grok is getting a lot better. Being late to the party was a key advantage to them.

They were able to benefit from the learning of everyone else. But I noticed they're not really inserted into these kinds of conversations that I'm having. Actually, Microsoft, to their credit, you know, has crock up on Azure now.

So there you go. Elon was in there, was in their keynote. Hey, I have a question for you, John.

It it, I was invited to go to IO and I just couldn't work it out scheduling wise, but from afar it seemed less of a developer conference and more of a consumer products. Is that true? Am I reading that right?

No, that you are after. You're spot on. Spot on, Mitch.

I think they're trying to, they're trying to, to pivot more towards the wider audience. Yes. And, and that's why they chose the demos they did.

Most of them were all, um, here's a vacation home example. We're gonna translate languages between two people in different countries. And here's how, here's how a video for content creators.

Here's a, um, a every example was a real life example. And it was, it, it struck me as trying to be, and I think you said this earlier, Apple-esque, you know, they don't, they don't twist ball Short. But to be fair, to be fair to Google, right, yes, Google IO has become a consumer show.

But Mitch, you had just said Google next, last month, right? And that was very much a cloud and developer orientated event That was, that was infrastructure developer to, I'm just noting that, and I'm not being critical of it. It's, you know, it goes where they want it to go.

But, you know, I, if I was a developer going to Google io, I might've been disappointed, might've been more set more satisfied by going to Google Next Cloud next. So, but that was much more platforms and building applications and that kind of thing. So it's just things are evolving.

So changing expectations of what IO is. I think my, my takeaway from reading everything I've been reading and I've been following it, Google is, is a key part of my day to day. I have to know what's going on at the company is everything is changing, but not that fast.

Yeah. Well, not, not, not as fast as the announcements come outta them. I, I, I'll call it as I see it.

That's, that's what I see. Anyway, Hey, let's take a break here on the gang. We're gonna come back.

Our next block is a report from our man in Boston, the Yankee ghost to Boston. Not a Connecticut Yankee, no, a New York Yankee goes to Boston. You're watching Textron Gang Discover Textron 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.

Well, continuing on the theme of, yes, There's another show, and yes, there was another smorgasbord of announcements and tons of stuff going on, but I'm gonna focus on two that seem to really make a difference or matter. One is, red Hat is up here and they have Ansible and they put an MCP server into Ansible. This is interesting because now an AI agent can call Ansible and automate a workflow or set up a task or get something going remotely that a human has vetted, at least as something that will have an outcome that is reliable.

So that's pretty cool stuff in the sense of it gives you a, a view into the future of IT automation in the gen AI age. The second thing they did is they announced that they're going to put together a project around a containerized version of something called BLLM. This is a, uh, software stack being developed, uh, I guess originally by the folks over in, uh, Berkeley in California.

And this is starting to gain traction as an alternative to Cuda. People are starting to say, I don't want to get locked into Cuda from Nvidia. So we're gonna build a software stack in the open source community that looks a lot like cuda, but does a couple of key things.

One is it runs on anything. So Cuda, as you may know, is pretty much locked into GPUs and I guess Nvidia plans to bring out some CPUs of their own in the future. But, um, right now you can have a stack that will run on Nvidia, will run on a MD, will run on Intel, and pretty much anything else that you want it to run on.

So I think that that is going to gain some traction. And then the second cool thing about it is, uh, the way Red Hat built it, it can run on a distributed set of Kubernetes clusters. So you can break up the processing and the loads across multiple CPUs and multiple nodes.

And that way, um, you know, you're not trying to build a mainframe type environment for every, uh, AI workload. You can distribute this out more aggressively. Um, now that all said, you know, God bless Red Hat, but I know I came to an open source Cons, uh, an open source event, but this is pretty much feels like an IBM conference.

So let me, let me pile on. When you say it's open source, was it released under a true open source license, Mike? Or is it this bastardized thing that they've been doing with Red Hat Enterprise Linux?

Yeah. Yeah. VLLM actually came to them through, it's a, it's a true open source project.

'cause it came Berkeley, a Berkeley license, neur Magic had a hand in it. It's a, the source code's up on GitHub. Um, but it, it is a, I don't know if they're extensions to it, but yeah, It is not clear that the container extensions that Red Hat came up with it will be under that same license or if they're gonna give that back to the Community if it's not an OSI recognized license.

It's not open source. That's it. They haven't, they haven't gotten that far in terms of how they're gonna position it.

So, I mean, basically they're just showing that they came up with a containerized version of it. Now, whether or not you need that, I don't know. I mean, some folks like containers, but they are saying it will be an open source project and a product.

So they were careful in their phrasing. Now here's another thing though, Mike, I wasn't at this event. I actually, I was in Boston as you know.

I left Sunday the day before, though. I saw them all setting up. My hotel was next door to the conference center.

To me, this very much looked like The final assimilation by the Borg of Red Hat. The red Hat that we knew that I've known right as the Shining City. Onto paraphrase, Ronald Reagan as the shining city on the hill of open source, right?

The example that we all strive to, where's there an open source success story? Red Hat be the first billion dollar open source company, right? That Red Hat's gone.

It's you, you're right. It's part of IBM. It's been Borg and, and, um, they still have really cool stuff and they still give out the Red hats as well as other stuff.

But it, to me, it just, it fundamentally feels different. Yeah. You know, who's not here?

It's, it's like the, and it's not a scientific poll, but one of the things I'm not seeing as I walk around is that classic kind of open source contributor, maintainer people, and, uh, all those folks, mainly the folks here are classic IT administrators, DevOps engineers, and you know, they all work for big companies. And, you know, we're not seeing this kinda like, I don't know where the cool kids went in the open source community, but it didn't come to this event this week. Red Hack Linux to, you know, yes, that's still there, but it's obviously OpenShift.

It's infrastructure software, software infrastructure, right. And the Ansible acquisition and folding that into MCP, which I, I agree with you for calling that out, that's important as well. So it, it, it, it resembles more of an IBM company because it's not just a Red Hat open source.

Yes. We've commercialized it and brought it to enterprises. So it's changed a lot.

Fred, I I stepped on You were gonna say something. Yeah, not at all. So I, I agree with you and I think my question here is, uh, so it's great that this validates the, you know, prolific use of MCP, which is kind of becoming the standard, but you know, it's expensive to run models, uh, locally.

It's expensive to run in your infrastructure. I really wanna know how this is tied to cloud infrastructure, which cloud service providers are going to adapt this. There are a million other, uh, infrastructure providers at that cloud providers already have.

So this feels very me too, but I just don't know if we're talking about IBM's cloud. Okay. Uh, I can understand that.

But for me, right now, if I chose to run on Google Azure or a WSI have all this functionality built into my cloud provider, I don't need any of this anymore, and they miss the window, They would argue that, uh, it's a hybrid world and you're gonna centralize the management of multiple clouds. So you don't want to get tied into too many specific cloud platforms. And so they're making a case for a horizontal layer of software, essentially, that allows you to centralize the management going back to platform engineering without getting too locked into a particular cloud service provider.

I don't know how successful that strategy is because they, in VMware for that matter, have been making the same case for a while. But I still see all these independent little stacks out there that people are managing with dedicated teams. So maybe that will all come together in the platform engineering era.

But to your point, I would agree right now everything is isolated. It would be really good to get some, uh, financial numbers on what it looks like to do the same kind of workloads across each of these environments using the same infrastructure. Because I think the, as far as I see it, the, the, the problem here isn't the adoption, it's the cost.

And in today's universe, the cost is super king for all the platform engineers. And there is this thing called token shock when it comes to ai, right? You gotta pay for the input and the output and it quickly adds up.

And so a lot of people are, um, you know, it's talk to some people and they're like, well, I built this LLM and it cost about a million and a half dollars to eliminate the jobs of two people making 50 grand a year. So it's, Hey, that's progress. There's another maybe subtle item.

I'm curious, Mike, how much they highlighted this, but part of what's kind of behind this V-L-M-V-L-L-M is the, our offering in the commercial service, I believe a curated model repository. So, you know, you're not going to hugging face or every other place in the world to get your models. You know, we had the same week, the same week, there wasn't a conference, but Docker announcer, harden, docker hardened images, um, repository as part of their docker hub.

So we're seeing more people doing this. We saw this at suse, they announced kind of the similar thing, right? Allen, we were there about SUSE images, curating images striped down.

That seems to part of that infrastructure now of where are we getting in, in our software and platform pipeline? Where is this stuff coming from? AI models included.

So the funny thing about that is that the Red Hat curation service actually resides on hugging face. So basically they're just creating a little segmented section of hugging face that they are saying, this is the models that we think work versus the, I guess maybe raff models on hugging face. But I, The good housekeeping seal of red, It's no, it's really, they call it the blue blood, as in idea, The blue blood.

Okay, big Blue. Uh, But I mean, isn't it, isn't it good to be able to mix and match hardware platforms, hardware and platforms? I mean, I think moving away from cuda based infrastructure is gonna cause some problems and compatibility challenges.

But I think that it's good. I, I think, you know, everybody being locked into Cuda and NVIDIA's cuda, I, I don't think that's, well, It's not good if you're a, a, you know, relatively short a guy in a leather jacket Mm-hmm. Who Loves waffles.

Yeah. Those Waffles Just say and pleased with it. It all depends who, who's, who's looking it all feds.

But to Fred's point, you know, I, we've seen this story time and time again. Nobody seems, you know, everybody gives lip service to portability and we need to not get locked in. And yet we find a way to lock ourselves in every time.

Absolutely. Absolutely. It's always the Optimization problem, or the theory of the optimization problem or the over optimization problem, Or we swinging, we swing the pendulum.

Yeah. It's always the pendulum. Like, I want one throat to choke.

I want best of breed. I want, you know, I don't wanna be locked in, but I want stability. You know, we want our cake and eat it too.

And sometimes you can't, you just can't be all things to all people. At Least, at least we're gone are the days of we're open and as long as you use our APIs, you're, we're open. So we, we embrace and extend it.

We embrace and extend, And some of this is still branding, right? A lot of people now identify themselves as Nvidia or something or other. And it's just like, back in the bad old days when people said they were a VMware admin or an Oracle DBA and they attach their job too closely to the platform.

Yeah. You know what's interesting though, though, with this move to platforms and, and all of these offerings in the, in here talking about Red Hat, but in, in Google and Mitchell you mentioned Seuss and, and some of the others, is they've all picked up the security baton and said, Hey, we're gonna take care of the security here. When you use one of my modules or you use one of my containers, or you use one of my diss, one of the reasons to do it is 'cause we're securing it, we're making sure it's secure.

And, you know, this was always, I think one of the things for me about cloud security, which is yeah, you are gonna tell me you are securing the platform, but, but when s**t hits the fan, it's still my butt on the line, right? So I'm glad that I use that docker hardened image or that SUSE image or the Red Hat thing. But, but when I get the breach and, and the, and the, the, you know, the bricks come tumbling down, they tumble down on me.

Mm-hmm. That's called shared responsibility. Yeah.

You have empathy for you. It's called hot potato. No, but, you know, but there you go.

This is a problem in this model, and it's a problem that I think, Fred, you know this right? In security, it's a problem that security people have grappled with for now, since 2005, which is, you want me to trust you to do my security, but I still bear, bear the responsibility for it. And, you know, and, and so that in order for me to do that, you gotta open the kimono a little bit and show me why this is so hardened, why this is so secure, why I should put, you know, my company on the line based upon what you are trying to secure.

Now, flip side is a lot of organizations say, well, you got a lot more resources to secure than we do. So we're, you know, we're, we're gonna put our eggs in your basket. But it's, there's A, there's an interesting conversation to be had when you look at the way that you containerize a platform.

Common applications, you know, workloads to say that, you know, Java, okay, well, the versions of Java my company uses, right? I want that in a containerized so I can optimize the patching, the consistency and delivery of that. And then I have the platform, so like Mitch was talking about earlier, so whether it's red hat or its use, whatever the operating system is, my question is, I appreciate that we wanna harden those containers, but you know, something more about polymorphic security cure.

Don't make me go figure out or deal with vulnerabilities. Go find them and patch them and do that automatically and bring that to me instead of, I've got this level of container security to maintain, then I've got applications, then I've got my infrastructure, right? The simplicity of that is, if your job is one thing, make a secure operating system, do that, do that dynamically, and then allow me to get the benefits of that for my hardened docker image or whatever the case may be.

I'd, I'd love to get there. That's gonna, when we talk about the fundamentals of security, the hygiene, right? Shimmy, this hasn't changed in 20 plus years.

The hygiene is still this critical moment here. So, I mean, all of this other stuff is really interesting. However, if we could use the tools today to solve the problems from 20 years ago, I mean, w why wouldn't we, Right?

But if I automate this vulnerability fixes, Mitch is gonna complain that you broke my application. 'cause, you know, we didn't test it and all this other stuff. And so, you know, we, You got Chocolate In my peanut butter.

Anyway, hey guys, we're about outta time. What a great, what a great, uh, gang today. Some really heading discussion.

I thought we were mature about our discussion. It didn't devolve into throwing names or anything. So Fred, you were a good influence on the panel today.

Can't wait to have you back on, man. Thank you for, gotta throw us some credit to Anne too. Well, well, Anne always brings a touch of class to it.

She keeps John in light. You know, we don't, We, we don't want Fred, we, we don't want Fred to run away. So we're, we're, we're on our best Behavior.

We're on our best behavior. Anyway, hey, I hope you've enjoyed this Textron gang. Remember, we've got another full, I don't know how many hours of Textron TV backing this up.

So go, don't go anywhere, watch that. If you're watching this on our daily feed, if you're watching this on our OTT channel that just came out on Apple or Amazon or Roku, or Android or iOS, thank you for downloading the OTT. You may be watching this on YouTube or any of our tech strong stations, channels, websites.

Thank you. Until next time though, this Alan Hummel for Text on gang. We're outta here.