Enterprise Agent Control: Copilot, Runtimes and Cloud

Hey everyone. Welcome to another week of Techstrong Gang. Today's Monday, so we'll call it Monday's Techstrong Gang.

But you got to start-- The news is so fast and furious that I tend to look at these things now as waves. I don't know if they're quantum waves, sound waves, light waves. Do you believe light is via wave or a particle?

It's something for us to-- I know Steven would love to jump into that. But welcome to today's Techstrong Gang. I'm Alan Schimmel, of course, and happy to have you with us.

Happy to have this great Monday gang with us. Let me introduce you to my friend Steven Foskett, friend Tom Hollingsworth, and of course, Mike Vizard. Looks like he's sporting a flannel shirt.

It's chilly up here in New York, Alan. Yeah, I guess it is. Well, how is it in Ohio?

It's chilly here, too, but it's beautiful. It is apple season. Ooh, I love apple season.

Yep. New Macintoshes and iPhones for everyone. Beautiful.

I was just upstate, and yes, it is apple season, and the leaves are changing, and it's all beautiful. Not here in Florida. But anyway, we're not here for the weather report.

We're here for Techstrong Gang. Gentlemen, as usual, it was a busy week last week, a crazy weekend with new announcements, and video's throwing their hat into yet another ring. Five and a half trillion dollars buys you a lot of things.

But Mike, I guess the biggest thing is Microsoft is again upgrading Copilot. This is such Microsoft, though, right? If you keep doing it long enough, eventually I guess you hit on something.

What do we got here? So Microsoft on Friday, which was unusual timing, decided that they were going to let everybody know that they were moving to the agentic AI user experience, and that's probably not much of a surprise, but the implications are profound. And what strikes me about this is, as we move into these AI agents as the new interface, if you look back at Microsoft, it was the house that that graphical Windows interface built.

And it was the assumption that they would have some sort of control over you as an end user because, well, you were used to that particular interface, unless you were a Mac user per se. But it was the thing that kind of enabled Microsoft to keep everybody in line. And now when I look at this, Tom, going forward, if AI agents are the dominant thing that we're going to use to interact with software, well, do I care if it comes from Microsoft or not?

Or is this an open conversation now? Because I can use any AI agent to access any data source, so who cares where it came from in the first place? What do you think?

Well, this one's more important, obviously, because it has Office built into it. Therefore, it's the one you have to use. And look, I don't know what we're going to do in the future.

I'm sure whatever it is, it will eventually be Microsoft Office-- or Microsoft Copilot operating system for Office and other stuff you want to do TM. " Microsoft has decided that whatever the future of AI is, by God, we're going to brand it as Copilot. I think that this is past due, and that should give you everything you need to know about what's going on when you consider that the big conversations that are happening in AI revolve around other companies, and Microsoft is still like, "Look, we still have an agent system.

" Dropping it on a Friday so that everybody could ruminate about it all weekend. Nobody was ruminating about it, as far as I'm aware, at least. I find it fascinating that this is the way that they're choosing to go about it, which I do want to make a point here, though.

Most of us. I have access to the Office suite because I have to send documents to the rest of the world in Word and open up spreadsheets in Excel. But it's curious to me because I'm now 48, and at no point in my entire life, if I sat down on a Monday morning and said, "You know what I need to do to accomplish today's task?

" But yet Microsoft is putting their application development environment right next to Office in Copilot. I'm like, man, if only instead of sending that email to Steven to let him know how things are going at this event, I could just write an entire app to pull all this data in so that he could query it on demand. No.

Nobody needs that. And Microsoft, if you think that people need that, you should probably get out of Redmond more and go talk to the people who are using your software. Well, let's talk to the people using the software, though.

Because like you, Tom, I have an Office 365 subscription. I have my Google Workplace or whatever the heck that's called, and I flitter and flutter between them, depending on what audience I need to share with. But I am constantly amazed when I talk to friends, families, neighbors, associates who are not in the tech industry.

A lot of them are in finance, healthcare, regular work businesses. We forget what an iron grip Microsoft still has in that market. I can tell you I know for a fact because I worked at IBM in 2001, where it was corporate mandate that we were using Lotus SmartSuite.

AmiPro 1, 2, 3. Hey, I was part of that. And as soon as I handed out a new laptop that was running a Pentium III chip, by the way, the first thing that popped up was a Windows 2-- or an Office 2000 CD.

" And that's the thing. But here's the extension of that. Again, we live in a bubble, us four, right?

We all play with every AI model we-- to our heart's content, probably, right? But for an overwhelming chunk of the audience, of the market, Copilot is AI to them. And look, ChatGPT or OpenAI was powering Copilot for a long time.

Now Microsoft has some of their own models. So they were getting ChatGPT through Copilot, so they-- It was the cat's meow. They don't even know.

And I'm not disparaging these people because they're not as AI cool as we are. That's not what it's about. Microsoft makes their living off of those people.

And they, as you mentioned, Tom, they will continue to well into the future. This is the way. Is that the Mandalorian?

This is the way? This is the Microsoft way. I'm going to- They iterate.

They keep iterating. I'm going to disagree with you. Okay.

I think that what you're describing is true today, but I think as we go along, and I'm already seeing people who are having this issue, right? " And if so, then, the Microsoft agent and the Adobe agent and every other agent is a sub-agent to ultimately what I want my AI agent to be the lead for, and that is the thing that I'm going to use to orchestrate all those agents and tell them to go complete a task. And whether the thing was created in a spreadsheet in Google Docs or Microsoft or whatever, I may not care anymore.

And I'll just double down here on the branding value that Microsoft has with Copilot. If you ask a lot of people to talk about agents, I'd say the word that they will use second most to the word agent will be Copilot. That's become almost the Kleenex of agents.

No, it is. But there's a reverse to that. I don't think Microsoft can trademark it because other people are using the word Copilot.

Yeah, I agree with you on that. But I would say that, I think that puts Microsoft actually in a pretty good spot. Even if it's not trademarked, at least users know what it is.

At least it doesn't have some weird wacky name that they have to try to identify. But I'll take the over on this. I think Microsoft actually is in a pretty good position with Copilot This despite the fact...

So first off, I really want to put a shout-out in the episode here to my Mastodon friend, Kevin Beaumont, @ossiethedog, who single-handedly cost Microsoft a billion dollars by posting a years-long Mastodon thread about how bad the Microsoft recall feature was. And so Microsoft finally killed the Copilot Plus branding. And according to internal Microsoft employees, it was because of Kevin Beaumont's Mastodon thread.

So power the Fediverse. But that being said, what they've announced actually really is kind of a Copilot Plus PC, because essentially what they're doing is they're pulling more and more into the Windows operating system. They're allowing Windows to do what Mike suggested, which is to sort of be the one-stop shop, the interface for agents.

And frankly, I feel like more companies should be, like Microsoft, like Apple, like Google, should be doing this sort of thing, because otherwise this stuff's going to get out of hand, and it's going to be some other company that's going to do it. And I know we're not talking about it today, but I would so like to talk about Facebook and Muse, but there we go. You can go throw that in there if you want, because just today they were talking about their enterprise edition of that platform, and they have some ambitions in that particular space.

But I come back to the simple fact that there's too many damn agents, and I can't keep track of them all, and I cannot handle the cognitive load of managing all of those agents as a user. But Mike, so what you're really saying, though, is that the catbird seat is the company who makes my master agent that controls the other agents. And I think what Steven is saying, especially if you're a Windows user, there's a betting favorite on that catbird seat winner, and it starts with Co.

Right? And if Microsoft can pull that off, great. More power to them.

I'm sure Nvidia will throw their hat in the ring on this, and some of the other usual and likely suspects. But who better than Microsoft? Maybe Apple.

You could see maybe Apple, they are the sentry at the entrance to the walled garden, and everything has to go through them. Perhaps. Perhaps.

Or perhaps I do something radical and Perplexity becomes my master orchestration for all these other agents for data. I'm not on Kalshi or anything, but I would say that's a long shot. I totally agree with you on this.

I think that that's the battle. That's the core battle. Wait, could you say that again, Steven?

I said, well, because we're not talking about data centers and Facebook. Oh, okay. But I completely agree with you, Alan, that that is the core battle here.

The battle is for ownership of the master of all agents, and Microsoft has a really great start here. Google has a really strong potential, though I haven't seen them really doing this. But then, yeah, the battle is open.

Anthropic and OpenAI both have master of all agents that they're kind of baking into their platforms. We've got Muse, we've got all these other, Perplexity. Everybody wants to win this battle because they know that that is the battle, the whole thing.

That's the whole thing, the whole can of worms. And here's the other thing. Remember, Tom, back to your Lotus thing.

Remember the saying when you were at IBM, you know what the saying at Microsoft was? Windows ain't done till Lotus don't run. Right?

And if Copilot has some sort of inside track to allowing people to use Windows, and that's still 70% of the market. I don't think AI agents give a, pardon my French, s**t about whether or not the thing is Windows or not, or whether it's a Mac thing. It's all headless back-end services as far as they're concerned.

Yeah, but no, you're giving the wrong s**t about something. It's not the AI agent, it's the end user. I don't think- And the end user knows Windows, it knows Microsoft.

I think the end user's not going to care. Or, okay, here's a curveball. Is that actually not true anymore?

And is the end user now Facebook and Meta and maybe even ChatGPT? I think that the ballgame might have gotten away while Microsoft was playing. That's a possibility.

Because ChatGPT, between their commercials and everything, I don't know if you guys see a lot of their commercials, like on YouTube or whatever, and they clearly have a consumer focus. So does Meta, the Muse thing, where Copilot, Office 365 is still work. I'll be honest about it with both Google and Microsoft.

The Docs experience from both of them leaves a lot to be desired. So it's not a great experience. So I think there's an opportunity here for a different conversation to be had about how we're going to engage with software in general.

And for the first time in, what, three decades, we're having a moment here where people can think about it because they were locked in for the last three decades. Not so much anymore. So let me, I'd like to just say one quick thing on that.

It's funny, I use Google Doc to share a doc that I don't expect you to do much editing on, light editing at best, but at least you'll be able to read it. And you'll be able to read it on the web, and you don't need an app, you don't need a subscription, you can read Google Doc. When I want to share things like with editorial, Mike, or I noticed this with my book when I was going back and forth with the editors.

The editor making changes and tracking changes and accepting, like real editing. Google Doc came up really short. Microsoft Word is still the king of that castle.

And the Copilot integration into it was pretty nifty. And I'm not a Copilot fan. I don't use it very much at all.

I do use Perplexity and Chat and Claude. But they still are at the same game. They've got two big horses, Windows and Office.

And as long as they put everything else connected into that, that's a damn good business. Anyway, let's move over to, we'll leave it on that, move over to segment two for today. Mike, what do you got?

Well, it's more about how these AI agents are going to run, but safely this time. For sure. And I think Nvidia today stood up and said that at some point in working with not just themselves, but other providers of silicon, that they will describe a way that they're going to create a secure runtime for AI agents.

Now, this comes on the heels of Docker last week saying that they have built a secure runtime that is more software-oriented and can be deployed anywhere. And all the underlying framework for that, not the actual runtime, will be donated to the CNCF, and then that will become the thing. We're going to build an ecosystem of secure runtimes for AI agents that are processor independent and can run anywhere at any given time.

And it kind of seems like, to me at least, that the Docker thing has the benefit of existing today, versus the other one is waiting on chip companies to actually do something at the processor level, which is going to be, I would argue, a day late and a dollar short. But Alan, what do you think? You know what got me with this?

Nvidia made a cheap huckster marketing claim, which kind of just made the hairs on my, I don't have many hairs on my head, but on my neck stand up. And that is, if this was around, it would've stopped the Hugging Face incident. And you know what, Tom?

You're a security guy, right? We've been hearing that same argument. If you would've used this in Troy, the Trojan horse would've never got in.

And we've been hearing it ever since, right? Had they used this, this worm wouldn't have worked, that virus wouldn't have done, this attack wouldn't have happened. " Yeah.

Don't waste my time. If only you'd been wearing your seatbelt, you would've survived that car accident. Exactly.

But in the auto manufacturer's like, "But I provided it for you. " And that, I think, is where the definition point here is different. Nvidia is saying, "We are giving you the capability.

" I'm just curious, have we heard anything from Sam Altman about how he's so sorry that this happened, and it's completely our fault that we didn't write code in here that would've prevented this from happening? Or is it the same smugness that I get from him normally of, "Well, look how powerful our AI is. " Like, I could've stopped OpenAI from hacking, well, we're up to now what?

Hugging Face, and the Australian government, a bunch of other people. It's called a power plug. Yank it out of the wall and suddenly it stops because- Well, no, Tom, they did.

So I read this morning-- I started a new podcast, audio only, I'm doing every day, 15 minutes myself, called Techstrong Daily. They did pull the plug on their latest model. Oh, I'm so glad to hear that.

Because actually- Sam didn't make an announcement, but OpenAI said they pulled the plug ... So they put it in time out, and they're rebuilding the guardrails, and they're making absolutely sure that it can't escape containment this time. Sure.

They're going to a different model altogether. Hey, Rocky, watch me pull the rabbit out of my hat. This time for sure.

And that's the problem is they keep shoving this thing under the covers, hoping that nobody's paying attention to the fact that this has nothing to do with Nvidia, this has nothing to do with Docker, and this has everything to do with you guys letting your teenage model run wild everywhere, hoping that it might eventually develop some kind of a conscience and not do these kinds of things. Yeah, because every 18-year-old out there knows that if there are no consequences for you breaking curfew, that eventually it'll kick in. So, but let's not just leave it at OpenAI's doorstep, right?

Google's done had these problems. Anthropic's had the problems. I believe DeepSeek had.

Everybody, all the model makers wanted to join the 18-year-old adolescent parent club. And so they've all had these kinds of things, which to me is a bigger story. I think the bigger story is here that AI, when it reaches a certain level, and it's not consciousness or anything, but at a certain level, AI agents do these things.

They push the envelope. They bend the rules. They have a mission, and they don't know where the guardrails and when to stop is.

And that screams to me that we got to do something about that. Alan, there is a secret trap door built in that will make the AI agent stop. I don't know if you're aware of this or not, but it's the same one that works on board members and CEOs.

They stop when the stock price goes down. Suddenly we're stopped. Yeah, but in this case, it's opposite, Tom.

First of all, most of them are in public anyway, but they're doing it to boost their creds, to boost their valuations. Yeah. So if you're billing marketing, it works.

So again, the secret is that if we punish the companies by lowering their stock price or whatever their valuation is, when they do something they're not supposed to, then maybe we should stop rewarding these stunts. I was on a panel last week and jokingly said to the members of the audience, I said, "Hey, has anybody figured out if OpenAI hacked anybody today? Because I feel like we're due.

" Little did I know that was when the news came out about the Australian government hack. And I said this on our episode of Security Boulevard today that is going to publish next week. " 2% of your valuation in a fine until you fix this.

Sam Altman is going to be hauled into the- Tom, you're starting to sound Bernie Sanders to me here Oh, I know. He's starting to sound very European Union Commissioner. Yeah.

I missed his campaign. I'm going to haul Sam Altman into a meeting every day until you fix this. Because if you are the public face of your company...

I know that corporations insulate the executives from being able to be held liable for stuff. But if you're the public face of your company telling everybody that if we just let it do whatever it wants, eventually it might find the cure for cancer, cool. You can sit here until it does.

You can pay us 2% of your revenues and/or your company valuation because let's be fair, you're OpenAI, you haven't made money yet. We're going to do that until somebody inside your organization says, "Hold on a minute. " Because Sam's getting awful mad that he has to go sit in this meeting every day.

Tom, we're going to give you five minutes. There's a guy from the Justice Department at your door. This goes against the AI force.

Yeah, exactly. Yeah, you're right. And if I wanted to make up a military, I have all of my anti-AI force people here too.

You just can't see them because they're invisible. They have really good stealth technology. They're totally not imaginary.

So Tom, I have a question for you. If you fine them a percentage of their revenue, and they're actually losing money, does that mean we have to pay them? No.

It means that it just goes against them so that the stakeholders don't get paid out of the IPO that they bought low buy. Yeah, but what about if the American government's a stakeholder? You're costing us all money.

Exactly. We're all in the boat with Sam Altman. Yeah, but the difference is that the American people- Too big to fail ...

the American people can vote out the board of directors every two years, unlike these companies. Yeah. Not only that, I'll have you know that in the Anthropic IPO, they have not a poison pill, but they're trying to put a clause in there that the board directors have to be picked by a governing secret society or whatever, who will pick the board directors to make sure that the founder's vision for AI and everything stays pure.

Right? So that's a little... Think about that.

It's pretty unusual. So besides me, though, does anybody find this very suspicious that Jensen Huang is standing up after last week saying that AI shouldn't be regulated, and we can take care of it ourselves, suddenly is saying we have a magical cure that's going to be an extension of our silicon firmware, and we're supposed to believe that? No, because Jensen's not regulating AI.

He's giving a tool to people who wrongly assume that you should regulate AI. He's not enforcing it. He's just building tools in for you to do it, if you so choose, but you're wrong, so don't do it.

We live in a world of personal responsibility. I'm in Florida. I know from this.

Didn't you once say that personal responsibility is something to the effect of it ends at the tip of my nose? Exactly, it does. But don't ask me why I'm still here in Florida.

But I don't think it works there, Mike. I don't think it works, Mike, in holding up our constitutional rights, and I don't think it works in securing AI agents either. Yep.

I will agree with you on that one. If I can jump in on this, too. First off, so far agents have proven really adaptable and able to escape containment.

Some of these things are woefully terrible security practices. Some of them have actually been pretty good security practices. I like Docker's approach, but I'm concerned that it won't be enough.

I like this Nvidia proposal, but I'm really not sure that this is enough. The truth is, if you look at what they're doing right now, all of the AI security stuff is turtles all the way down. Again, I'm bringing it up again.

So Muse, another great Mastodon poster, Johnny over there, who's been a real thorn in the side of all this stuff, has deconstructed Muse because surprise, they give you root access on that magical VM in the cloud. And so he's read the whole thing, all of their prompts, everything. They are actually using AI agents to prevent prompt injection.

They have an AI agent that is supposed to inspect the prompts to keep prompt injection away. And in case that doesn't work, they have an AI agent whose job is to keep the AI agent whose job is to keep the prompt injection away from getting hacked. And if that's not enough, they have another AI agent who- Which is really interesting ...

and you get where I'm going with this. It's all AI agents, and none of it works. None of it.

No, but so Tom, let me soothe your troubled soul today. Okay. Yeah.

I remember 20 years ago, I did a podcast with the CEO of Couchbase and MongoDB, asking them if NoSQL stood for no security. And you know what they told me? " And I'll leave you with another good quote, Alan, from "Battlestar Galactica," all of this has happened before.

Yeah. All of this will happen again. When we demand security- Yeah ...

magically it will appear. By the way, the CEO of MongoDB announced today he's going to work for Meta or something. Really?

He's leaving Mongo, though. They didn't buy Mongo with him. He's leaving Mongo.

Mongo's stock has taken a smack in the face today, and he's going to be running, apparently, a lot of this Meta enterprise stuff, so there you go. That's interesting. Mongo's had a...

Well, that's a whole another segment. Another day. Which database companies have capitalized on AI to the fullest extent?

I would say it's definitely not Mongo. Those freaks? All right.

Have we beaten this one to death? Are we ready to move on to segment three? I believe we are, and I don't know how much AI will be in this topic, so we'll see where it goes.

Luckily, because Tom's going to... I'm worried about him today. Okay.

But Steven, take us through, you guys had the Cloud Field Day, and you had a bunch of folks come by and have a chat about where all this is going. Give us the highlights. Yep.

Yeah, so Cloud Field Day was last week. It's always great to do something that's not AI Field Day. But let's be honest, they're all AI Field Day.

Hey, Tom, how's Networking Field Day looking? AI Field Day? All right.

So, yeah, Cloud Field Day was last week. We had some great presentations. Unfortunately, one of them I can't talk to you about yet because it's under embargo, so that's going to be a lot of fun.

I can't wait to share that later. But yeah, I know that this is going to sound like a newsflash, but there are things other than AI in the world, and that was pretty evident. Now, of course, there's AI flavoring in here.

Every one of the presenters talked about AI, but there's a lot more that's happening, too. So, we got to see an awful lot of really in-depth information from Clumio at Commvault talking about data protection for cloud native. We talked about ransomware.

We talked about granular recovery, because actually they can do some really cool stuff, like on a per bucket basis, rolling back to different points in time based on where the ransomware hit and where it didn't. Things like that are pretty powerful. And of course, the Clumio team really had a great understanding of cloud native architecture before they were acquired, so it's really neat to see how they've been able to bring that sort of really next generation platform background to Commvault, which, let's face it, Commvault is a long-running company, and of course, it's always hard to change the course of the ship.

But I think Clumio has done a really great job of bringing that kind of understanding. We also had another company that you've probably heard of, Broadcom came in with a little product called VMware, if I get this correct. And they talked about where they're going now with basically everything that was everything a few years back.

And as I said last week on the The Techstrong Gang here, I actually think that what VMware has done, even though it's been hard, it's been hard for the employees, it's been really hard for customers as well. I think that it had to be done because the VMware platform was really rooted in a decade or more of technical debt. They've had to wipe away a lot of things.

They've had to wipe away a lot of capabilities. Unfortunately, they wiped away a lot of employees and customers, too. But the remainder is actually a pretty compelling, unified, next generation architecture for a few of these major companies that really needed something like that, and we got to hear a lot about that from the Broadcom folks.

And then finally, we had another favorite company of mine and Tom's, Selector. No, not the ska band from the 1980s, but in fact a modern-day company specializing in really understanding what's happening, what's really happening in the network, and how do we figure out where there are problems? How do we make that more usable, more friendly?

I feel like they're completely rethinking this whole platform. And Tom, I want to throw that to you first because you're the most familiar with Selector. I have to admit, some of the stuff they say goes right over my head, but man, I love the fact that they're really taking a fresh look at all of these network troubleshooting issues.

The thing to me that I think that a lot of people don't understand is that this data has been sitting in the network forever. We're not manufacturing this stuff out of thin air, but there's so much of it. Think of UHF, right?

You get to drink from the fire hose. If you can't make sense out of the fire hose, then you can't start to know where these problems need to be fixed. " And that's super valuable because if you can disregard the background noise and focus on the important parts, then that makes your operations teams run a lot more efficiently.

And to me, that's always been what Selector has been about, is the operations mentality of we're not going to try to boil the ocean in a seam. We're not going to try to automate that ocean boiling with a sore. It's like this is the thing that you need to be paying attention to because this graph over here that we've run says that this is the cause of all of your problems.

Anybody who's ever dealt with network operations, remember how transformational it was when the system would detect that a WAN link went down, and so then it wouldn't send you 55 alerts that all of the devices on the other side of that WAN link went down, too? That it was smart enough to know, oh, if that link went down, that's probably the root cause of all these other outages. Imagine if that could be applied to every problem in your network.

I would really like that because every time my WAN goes down, I get a million website down notifications. And I got it. Also, kudos, Tom, UHF reference.

I've got a ska reference. Alan, can you talk a little bit about ransomware protection, and can you please put in a 1980s media reference to your discussion? 1980s media?

It's critical. I'm more of a '70s music guy than '80s- We'll take it ... but we will.

But ransomware protection. Boy, you threw me a fastball here, Steven. I was still in the on deck circle.

To me, you look at what's going on with Cloudflare and Akamai, where they're trying to cut this off at the knees. I can't think of a 1980s media thing. I'm so sorry, guys.

Come on, man. Put the lime inside the coconut and do it. No, that's '70s.

Come on. " I'll give you that. Go ahead.

Take it. Is that '80s? All right, "Money for Nothing," one of my favorite Rockeffeller songs.

I always thought of it as "Chicks for Free" but whatever. But more importantly, is ransomware still a thing? We're not hearing about it as much as we were.

Has it just become in the woodwork? Well, that's actually a good point, Alan, because AI is drowning out everything, isn't it? Yeah.

Ransomware has not gone anywhere. It's just as bad as it ever was. We're just not talking about it because we've got- No, they're using AI for the ransomware, but- ...

more hidden AI hitting us every day. Right. It's going to be ransomware at machine speed next, is what we're really looking at.

No, I think the issue with ransomware is that the attacks still happen, the repercussions aren't as bad. A lot of companies have gotten wise to having backups, off-prem, non-connected backups. The cyber insurance industry has had a huge effect here in getting people smart about how to survive ransomware, and they're also very good.

I don't know how many of you have unfortunately ever been involved in a ransomware with a cyber insurance company, but they generally know who the What kind of attack it is, what's the normal customary rate for these things, whether to pay or not to pay, how to avoid the most damage, and all of those things. So, while we haven't eliminated the scourge of ransomware, I think we've made it livable. What you just described is we put a customer service desk in front of our extortion people.

Come on. No, but you know what? That's- And look, if you think that this has gotten better, tell somebody who works for Jaguar Land Rover that it's gotten better.

They lost half a billion dollars and had to furlough thousands of employees because of a ransomware attack. Yeah. I don't think it's gotten any better.

Shame on them. I think we're talking about it. Shame on them.

I think it may get worse before it gets better. I think AI agents will make it better to defend, but there's going to be a gap because the bad guys have more money and more time, and they're going to use AI agents to launch these attacks with more sophistication and faster, and it's going to take us a little while to catch up. Fair enough.

I still can't think of an '80s media reference. It's okay. Tom and I will fill in for you at any time.

We've got very good questions. I have one last question on this cloud thing, though. So does cloud kind of erase the silos between what we used to call compute networking and storage, and all that stuff just kind of now is in the background somewhere?

Or are we still managing these things in some semi-isolated kind of way, Tom? I don't think it erases the boundaries. And the reason why is because, with due deference to my good friend Justin Warren, who wrote this presentation many years ago about why your organization wants cloud.

Cloud isn't a physical thing. Yeah, it's a data center somewhere in the middle of nowhere. It's an operating model.

And I think that applying those operating models to different locations, whether it's on premises, off premises, what have you, is the real value. So yeah, there's some storage, there's some compute, there's an insane amount of RAM over here. It's like the GDP of a small country in RAM.

But the cloud in and of itself is not magical. It's how we apply those tools to be a better way to operate the resources that we have. Because we're starting to see, especially with the rise of the AI data center, that stuff is being repatriated back into the organization for security reasons, but also because trying to run that much stuff in Amazon is reassuringly expensive.

Yeah. It's a real good point, Mike, and I have to agree that it does sometimes seem like these once important disciplines have faded. Which breaks my heart as Mr.

Storage talking to Mr. Networking over here, that people think that these are not important disciplines anymore, that they're just elements of a platform. In my mind, they have not decreased in importance, they've only decreased in attention.

Just like what we were saying about ransomware. People think storage is easy. People think networking is easy.

People think compute is easy. People think everything is easy because ChatGPT and everything makes everything look easy. But you know what?

It's actually really, really not. And the companies that are doing the best, like AWS, like Cloudflare, companies out there that are doing all this platform, really nuts and bolts platform work, are doing it because they have specialists that are focused on these little areas. And those specialists are doing well.

Stephen, I read an article this morning, I think it was, about S3, right? How fundamentally S3 changed the storage game. Yeah.

And the historical comparison of cost for storage between, let's say, real disk versus solid-state drives, right? And believe it or not, it's not that much more expensive anymore. And S3's kind of led the charge there.

And there's still a world unto itself of S3 experts, and then the S3 wannabes, right? If I really want to go downstream and do storage as- Yep. Backblaze, Cloudflare, Wasabi, all them.

Yep. So, I don't know if we busted down the silos, but I think it's still expertise that's required. I don't know.

" I don't know. Am I dreaming, Tom? I hope you're running that model locally, Mike, because if your networking agent- ...

I don't know, decides to upgrade the code on US East One randomly on a Thursday afternoon and knocks out cloud access to the entire eastern seaboard, have a local backup model to write your resume from that event just in case. I will call up my AI agent that manages incident response and take care of all that. My question is, though, is it called Copilot?

I don't know what the future of this all is, but it's going to be called Copilot. There we go. There you go.

All right. Hey, we're about out of time. Steve and Tom, thanks for coming up and making our Monday gang a great kickoff to our week.

Mike, good stuff as always. We'll be back tomorrow with more gang, more talk, more news, more everything. Until then, though, thanks for watching.

tv, TechstrongTV YouTube channel, the OTT app for Android, iOS, Amazon, Roku, and of course, Apple TV. But for now, on behalf of Techstrong and Tech Field Day and Futurum, Futurum Media, have a great day, everyone. We'll see you tomorrow.