The Post-Cook Era, Copilot’s Cost Crisis & Google’s AI Arms Race | Techstrong Gang

Thank God it's Friday. Welcome, Techstrong gang, and another whirlwind week of tech news. Boy, it's just coming and going as fast as we can process it.

In case you haven't noticed while you're traveling, we're in the midst of conference season, which explains the absence of Alan and Mike, who are in Prague this week for the SUSE Conference. I drew the short straw, I guess, and attended Adobe Summit in Las Vegas. Lucky me.

Anyway, we have a great gang assembled here today, and we want to introduce you to them. As usual, on Fridays, we have Tracy. Hi, Tracy, you're in New Mexico, and how are you doing?

I'm doing great. Glad to be here. Glad to have you with us.

We have Jack in North Carolina, Jack Palmer. Good afternoon, folks, or morning, depending on which half of the country you're on. Jeff Ritch is in South Texas.

Woo. Hi, Jeff. Hello.

Hello. And we have a newbie. We have Andrew Storms, who's in Northern California, who's fairly close to where I am.

And this is Andrew's first show, so welcome, Andrew, and as part of our hazing ritual here at Techstrong gang, we want you to talk a little bit about yourself and introduce yourself to the audience. Oh, thank you. Nice to be here.

I don't think that's much of a hazing, but more of an- Eh ... introduction. Yeah.

I was expecting something a little bit more intense. So 30-something years working in cybersecurity, either as an individual contributor manager kind of thing. Really kind of all the gamut of maybe pen testing, running teams, policy, you name it.

Predominantly B stage level kind of companies where they're looking for hands-on folks to come in and just right size the security organization and mature it. Right now I'm between jobs, but doing some contract work for a company called KiloCode, where I'm helping them build a hosted OpenClaw infrastructure for them, so fun AI stuff there going on today. Yeah.

Well, welcome, and we're glad you're here, and we hope to have you on many more shows. So while Mike and Alan were in Europe, Apple dropped a bombshell earlier this week. And I'm going to go to you first, Jack, but I'm going to drop a bombshell on you and have you go first.

And we're going to start with the end of Tim Cook's 15-year reign as CEO of Apple. John Ternus, who's currently the senior vice president of hardware engineering and has been with Apple for 25 years, is going to become the company's eighth CEO effective September 1st. And I'm going to kick it off this way.

While there's no debate Cook has been exceptional and fantastic for Apple investors, pushing the company to a $4 trillion valuation powerhouse, offering a steady hand, he wasn't exactly a dazzling innovator, and there's a growing concern that his reliance on foreign manufacturing and global supply chains didn't always serve American economic interests well. So there's kind of a mixed bag. But I'm going to start off with you again, Jack.

How do you assess Cook's legacy? Well, I think he had some very big shoes to fill coming in at the end of the Steve Jobs era. And there was always the question of when the founding personality leaves, what is going to happen to a company like Apple?

And I think Cook did an amazing job, as you notice. He grew the company an extraordinary amount of growth. It's one of the biggest, if not the biggest, companies in the world.

He also turned it from a computer company into a consumer electronics company. And while Apple is known for the Apple computer, most of what they do today is really much more consumer electronics oriented rather than traditional computing oriented. And alongside that, he built up an entire media juggernaut.

So I think from a pure how did he do in his reign, he did great. It's a track record that any executive would want to have and a legacy that every executive in the world, I think, wants. Yes, I think he did offshore all of Apple's manufacturing, but that was the trend over the past 15 years, and it's only recently that people started thinking that maybe that's not such a good idea.

So I don't think we can fault that, put that solely at Tim Cook's feet as much more that that's the entire world was doing that and offshoring. And I think had Apple and Tim Cook said, "No, we're not going to do that, and we're going to maintain the costs of US manufacturing where all of our competitors aren't," that would be a hard pill for people to swallow. Yeah.

So I helped, or I worked with Daniel Newman, who is the CEO of the Futurum Group, on a column that he wrote for MarketWatch about his takeaway about Cook, and he referred to him, and I think it was incredibly accurate, as the greatest caretaker CEO in the history of technology. And he meant that as a compliment, given the time and the space Apple was in post-Jobs. But I guess now going forward with Ternus, who was involved in iPhone and some other major products at Apple, we're at this AI inflection point, so to speak, and I'm wondering what would we require from Ternus in terms of product vision, supply chain execution, or even ecosystem diplomacy.

I'm wondering what you think about that, Jeff. Well, a few things. First of all, I haven't heard anyone say yet what would Steve Jobs be doing now?

And I think it's good that that question isn't being asked. That being said, Tim Cook is a wonderful executive. Financially, he couldn't have done anything better for Apple than what he did, and I don't think anyone will deny that.

But from an innovative risk-taking, cutting edge sort of thing, which Steve Jobs really did a lot of, we haven't seen that for 10 yearsAnd even though there are some good products out there, a lot of people are starting to feel maybe some of them are getting a bit stale. And is Apple going to be able to stay where it is without something new? Because it can't simply be executive financial leadership, which Cook did a wonderful job of doing.

Now, I think that's why he's a good caretaker CEO, to your point. However, I think it is time, for a lot of reasons, for Apple to start pushing the envelope in some areas, maybe even finding ways to fail so that it can grow more from a technology and innovative perspective. I don't know what anyone else thinks about that.

I think- Yeah ... Apple, it can't be just a consumer products business. It can't live on its iPhones.

It just can't. And they're at a point now that they're going to have to do some kind of a pivot to an AI strategy. In particular when you see...

We talked about this last Friday, how all the giants are starting to really operationalize AI and build these platforms, and Apple hasn't been in that conversation yet. So I think it's a transition period for Apple, and they're going to have to build a strong AI strategy, and that may be an on-device AI. Who knows where their specialty will be, but they've got to move forward.

Kind of like the Microsoft changed and led the cloud pivot under Nadella. Similar. That's a good point.

It's interesting. So we have these almost three eras of Apple in the last 25 or so years. So we had Jobs, who was the marketing genius, could sell anything.

Also a visionary in a sense because he could piece together ideas and concepts, and then he went to the operations guy as the transition. And now the operation guy is making the transition to the hardware leader, which I've heard internally through Apple, a number of people said this should've been done maybe five years ago. That it was getting long in the tooth.

How many iPhone announcements can we have where it was an incremental camera improvement or memory or a different color of the model? So I'm wondering about that, the significance of having a hardware leader in a sense and maybe somebody who has ideas about products because under Cook's reign, the last product I can think of was the Vision Pro which just landed like a Led Zeppelin. Any thoughts on that?

I wouldn't be necessarily surprised if they did do a hardware offer. That Mac Studio really is intended for maybe designers, but wouldn't it be cool if we had something more tuned to local AI agents running at home? I don't see them necessarily going after the B2B place.

I still see them in the B2C, and I think there's an opportunity for people who are conscious about data exfiltration that would want to have something run local and not in somebody else's hardware, for example. Yeah. I guess there is kind of that open lane.

I remember all the ballyhoo over Jony Ive, his company being acquired by OpenAI, and this idea there were going to be some sort of hardware gadgets. So now that that looks like it's gone by the wayside and maybe is not a priority, maybe does that leave an opportunity for Apple to eventually fill the void once AI gains all sorts of traction? This is what Apple does.

It waits out an industry, whether it's the cell phone or the whatever it is, tablet, and then it moves in and establishes itself and becomes a market leader, for consumers at least. When I started prepping for this, when I looked this morning, I thought my original impression was that Apple was really a media company because they do so much on the media side. But it turns out that they get 75% of their revenue still comes from hardware, so they are a hardware company first.

Which I don't think of them that way, but that's what they really are. So having a hardware leader take the helm is probably a good idea. I think the challenge is they try to bring in AI hardware onto the iPhone and the last versions of the iPhone and the Apple Macs were supposed to have AI hardware, and that sort of fizzled, didn't went anywhere.

But there is this opportunity with a condensed home server, and people went out and bought the little desktop Macs, whatever they were called. I can't remember the little tiny ones. Mac Mini.

The Mac Mini. Thank you, Jeff. Mac Mini.

For running OpenClor, whatever it was. The original version that everybody went out, and I just found this article like two days ago that says you can't find them anymore. There was such high- Yes ...

demand and you can't find them, right? So maybe there's an opportunity there for them to build custom hardware for people to bring this stuff into their house. I don't know, but- I agree.

I feel like it's like a generational shift from this execution leadership that Cook was so good at to platform innovation. Mm-hmm. And this is, I think, what they've been missing over the course of the last five years is leadership in that platform.

Even though they've had some successful products, but there's so much that could be done, in particular, around platforms to support AI, even if it is at your home on your Mini. I do think there is an opportunity there. I have a Mac Studio here sitting next to me.

I'm real happy with the machine. I wish there were more in it, too. Andrew, you took that by storm, by the way.

That's part of your hazing. Love it. And I actually have a name for the product, too.

Rack in a Mac. Copyright it- Yeah ... as soon as possible.

So I don't know if this is a fair question, but I'm going to ask it anyway. Do you think Apple is... There's always this debate whether Apple waits too long or waited too long in terms of AI strategy.

We always talk about this on the gang, and I'm wondering if it would've served them better if he'd passed the baton earlier to someone else and they went harder after AI, or is it because it's Apple, they can wait longer than others? Well, they certainly have money in the bank, right? We've seen that before, and they've sat on it for quite some time, and we haven't seen it really put towards innovation in quite a while.

Coming from working for VC companies for so long, if I was at a company where we didn't spend that inflection or the induction of money, the board would get mad. And so I am surprised that they've sat on that cash for so long and really didn't do something more meaningful with it. Yeah.

They did a lot of buybacks. I think it's just- Oh, sorry. Go ahead, Tracy.

It could be just the right time. Mm-hmm. It could be just the right time now to start talking about AI native devices, and who is better positioned to deliver that than Apple?

So maybe they're just exactly at the right time. But they do suffer a little bit from the innovator's dilemma problem, too, is that their existing product set is so successful and generates so much money that innovating in a completely new space or doing something that might cannibalize the existing product lines is a very difficult decision for companies to make. And Apple, it's not just a question of Cook's leadership, but he's had 15 years to imprint that on the entire leadership team.

So is, Termas, I'm sorry if I got his name wrong. It's one of those that I can't pronounce. Oh, Ternus.

Yeah, you're right. Ternus, thank you. Is he and the rest of the team, are they so embedded in the conservative, go slow, test the waters approach that Apple's been doing for the last 15 years?

Is he just going to continue that legacy, or is he a bold thought leader that's going to put a stake in the ground and say, "I want to put my own personal stamp on the company, and I want to turn it slightly in a different direction"? And if he's going to do that, is that going to be successful or not because we have 15 years of legacy of being- Yeah ... very conservative and slow-moving.

Yeah, it's interesting because I think one of the projects he worked on, which really never saw the light of day, so to speak, was the Apple Car project, the much ballyhooed, mythical project that we heard about and read about for years. And he's been there for 25 years. He's 50.

He spent half his life there. So I would think he would probably defer towards the way things have been during his 25 years there. Or maybe he convinced them that he wants to take this company in a bolder move or bolder direction.

We don't know. But I will say this, I think Cook's... And a friend of mine, who's an analyst, Bob O'Donnell, put it pretty well.

He said Cook's real genius was figuring out not only how to create a global supply chain for a wide range of products, but how to make all those products work together as a system. And that's no small feat. That's part of his growing legacy, and he grew this company beyond anyone's imagination.

I would almost argue that he probably grew it as fast as, if not faster than, what Jobs would've done. Oh, in terms of expansion, he was amazing. Right?

Yeah. He was incredible. Yeah.

But we need an Apple bot. We need a company that's really focused on edge devices and AI, so I'm encouraged by it. I think that he's going to really try to usher in a next computing wave for Apple.

I really do. He comes from- That would be exciting ... the bottom side.

Yeah, it would be. Yeah. I would- I don't know about you, Andrew, but out here, there's this kind of sense of, I wouldn't say disappointment, but kind of lethargy or malaise around the Apple brand.

It was so exciting for a while, and it- I'm having a good time. I don't know about anyone else. But it's like he needs some sort of jolt, and I think this might be it.

I'm hoping, anyway. It just needs something new. Something new and different.

I would love to see- We need robots in the house. I was just going to say that. Let's get some Apple bots.

I would love to see them do robots. No one hears you. And get into an i- Yeah ...

an iOS-connected and iOS-driven robot set in the home. I think the home space would be great. I think we all need to be careful what we ask for.

Not that the robots are going to take over the world and kill us all, although I do believe that will happen. I don't think I'll be around to see it. I would like the robots to open the door for the dog, go get me a coffee.

I think it would be fabulous. Yeah. I'm ready for that.

And you know, Jeff Bezos has that company that he's invested in, that company that's out here that does the squashing robots. I think what we're looking at, though, as far as innovation, because I know Jack was talking about the innovator's dilemma, and you can look at the technology graveyard of every organization that took the wrong turn at that dilemma. Mm-hmm.

The advantage Apple had, I think Andrew mentioned it and someone else may have as well, is that they have enough of a cushion that they can actually make a pretty big mistake and it won't be great for them, but I don't think it'll hurt them. And I believe it's a time for them to do that. Not that, oh, you have the capital, go spend it.

But boy, I'm not sure there's going to be a better time for them to do it. Yeah, they have the cushion, and they have the, as you said, that war chest. So in a sense, it almost is like a- So I'm assuming John and Jack- ...

opportunity ... are somewhere in the ether. See?

I- Oh, go ahead, Tracy. Sorry. Like I said before, I think they're at the right time.

I think that we are starting to think about these AI devices in a different way, and we are behind China. China is way, way farther advanced inAs Andrew pointed out, B2C products in AI than we are, and I think that Apple's perfectly positioned to take that market on. Yeah.

That's good. Thank you. Well, we'll see what happens.

We're going to move on, though, to our next topic which Mike from Europe sent, which he referred to as AI indigestion, which I think was a great headline. Basically, GitHub suspended new sign-ups for several of its Copilot subscription tiers following a surge in demand driven by agentic coding workflows. The company confirmed that new subscriptions for Copilot Pro, Pro Plus, and Student plans are paused while existing users face tighter usage limits.

So Jeff, is this Copilot demand issue really a capacity issue, a margin issue, or a pricing model mismatch? I'm not sure I heard you. Oh, can you hear me, Jeff?

How about... Okay. Well, let me do this.

Let me pass this over to whoever wants to answer that question about the- I'll take it to start with. Yeah. Oh, sure.

Thank you. I think it's all of the above, quite honestly. Okay.

I could tell a story when at a former company, we were talking to Anthropic, and we kept running out of credits, and we were just like, "Here's my credit card. I don't care. " And yeah, it's indigestion because you want to use it, and in fact, I think we're being limited because there is a capacity problem overall in all the suppliers here.

And it's just not necessarily like that front end, like Claude code, but it's literally all the back end, too, that comes with it that's required. There's a huge demand. We're going to continue to see that kind of supply chain problem continue until they figure that out.

I don't think it's just GitHub. I think these coding assistants are kind of starting to behave more like, I don't know, a shared platform infrastructure more than an individual kind of IDE or productivity tool, and that kind of changes how we manage the risk and cost. " And that's kind of a ridiculous way to work, right?

So something has to be done. I was going to say, Tracy, didn't we have a conversation maybe three or four weeks ago on the Techstrong Gang where you talked about, we sort of alluded to the costing and managing your developer team and understanding how much they're spending on the AI stuff over time. And you are much more involved- Oh, it's a real- ...

in the development world, so it's your problem, right? It's a real problem. It's an annoying problem.

Like I say, you have to wait to spend more tokens, and it just seems like a really weird way to work. But we are looking at these products more as a platform, right? They're not just something that's generating code for us.

So we're looking at them even to start thinking about CI/CD infrastructure. So they're bigger now than just something that's generating code, and the cost factor, it can be pretty dramatic in terms of trying to manage it. Kind of like it was hard to manage code computing costs or cloud computing costs.

How long did we talk about that? We're in the same exact boat. Exactly the same place, and it's really hard to manage it.

And Jeff, you're- Unless we wait around for 30 minutes for new tokens. Jeff, you're back with us. Did you hear the question I asked you earlier, or do you want to weigh in on this?

Well, no, definitely. I'm sorry, I didn't hear the question. Oh.

But as you may guess, I clearly have an opinion- Okay ... on this whole- Share your opinion, please. I'm sorry, we lost you when I was asking the question.

Yeah, and I'm not sure what happened, but in any case, I'm back. That's good. I think what you're seeing is the beginning of a crustacean-based economy- Crustacean ...

in AI, and does anyone know where I get that reference? Crustacean-based. Yes.

How about- Yes, you do, Andrew ... from Red Lobster? It's- Claws.

Yeah. Yep. If you remember back a couple of years ago, it was two years ago, Red Lobster had to go into bankruptcy.

Yes. And a lot of people think it was maybe poor management, but a lot of people think a lot of it had to do with this endless shrimp special promotion that it always had going on, saying it was losing money by just letting people eat as much shrimp as they want, which I think probably, in many cases, caused indigestion. And I think you're seeing the same exact model play out here.

Now, I think where we are is GitHub may have learned from Red Lobster's mistake. No, you can't do endless shrimp forever because it is an bottomless pit that you have to continue to throw money into. Now, Red Lobster's coming back, and if you haven't heard lately, they are reintroducing endless shrimp.

They're reintroducing. I was just going to say that. They are in financial straits, and they are reintroducing the endless, yes.

So let's see how that works out. I'm not here into the food service analysis phase. But I really think there's a direct parallel here that allowing people to say, "Let have AI determine what AI can do to produce more AI," which is really what this ends up being.

" So the gates are going to have to be there, whether it's tokens or, I think the SaaS model doesn't play here anymore. Buy and you see. I would agree 100%.

It doesn't. The SaaS model doesn't, because if the usage-Um, throttling these compute costs, and that's exceeding subscription revenue, then we have an issue. So it makes us think things like, well, where should AI run?

Local versus cloud, or maybe, some private models. What workflows deserve agent support? Where should we use it?

How do we tightly control how to remediate where the AI and automation is prioritized? So they've got to get in front of it before we figure out a better way to do it. Mm-hmm.

I think the irony of this is that the latest buzz term that came just after vibe coding, which is a hideously bad term, is now token maxing, right? Oh. Ugh.

Where the engineer is told to shove as much context as possible into their interaction with the AI to essentially spend as much money as possible, the theory being that the more context you give the AI, the better the outcome is. So we now have two things fighting, two opposed forces fighting each other of how do you limit the usage on the service provider side, while how do you maximize your usage of this very limited resource from the developer side? Jack, could it not be that if you shove a whole bunch of garbage by token maxing, that you get processed garbage out of it?

I said it was a bad term. Oh, 100%. It's a bad process.

But, it's great- It's kind of bad ... for people who aren't software developers. They can go do vibe coding and feel nerdy, I guess.

I was chatting with some friends yesterday, and we're talking about can I resell my unused tokens? And clearly, it's against the terms and conditions of, say, Anthropic, but it's an idea, right? Is there going to be a secondary market I can go eBay, "I need more tokens," and get them added to my account instead of yours right now?

I would love that. Right? We could buy some discounted tokens, that would be amazing.

Yeah. And yes, it is against the rules if you look at the terms, so before you think about it. Um- So does that mean tokens are going to be the next digital currency?

Right? Yeah. Ooh.

That's exactly where I was going with that. Yeah. Well, they have to rethink it, right?

Because the market finds a way. True. And we will fix it ourselves if they don't.

So what should enterprises ask vendors to disclose? Should they ask them to disclose token economics, model mix, throttling policies, or worst case, cost per workflow scenarios? Is that- I think it's going to be incumbent upon the providers to outline each of those.

Because I think, as everyone else has said, and Tracy said it specifically, the market will find a way. So whatever constraint you think you're putting in, whatever guardrails that you think are going to work, I've been in security 50 years, people will find a way around controls every single time, I guarantee it. So it's going to be up to the provider to say not only terms and conditions, but "Here's what we can provide for you.

Here's why you want to use us instead of someone else," and make it worth the company's investment. Yeah. I just wonder if we're going to end up with this AI for dev price like cloud or metered pricing rather than SaaS per seat pricing, and how fast that transition happens.

I don't know, what do you think about that, Tracy? I don't know, to be honest. Most of this stuff is so hard to keep up with, especially when you're trying to manage a team who is using tools like Copilot.

I just see that we have more and more of a dependence on tools like Copilot, and we want it to do more for us. Mm-hmm. And we don't want to pay more.

We're already paying enough. And now they're complaining that their subscription's not covering enough. Something's got to be addressed.

So are we going to see the rise of FinOps for AI usage and for token maximizing or...? I think we're going to see development teams starting to move away from some of these foundational models and start building their own environments. Mm-hmm.

There's tools out there to do that- Mm ... that's not going to cost so much. Yeah.

And not have a token restriction. Cost Why shouldn't we? Right.

Yeah. Why shouldn't we? And there's better- Starting to make the- There's good models out there for generating code that's open source, and you can put it on your local machine.

You can put it on your mini Mac. So you're predicting the AI equivalent of cloud repatriation. I guess that, or just a revolt against- ...

being told you don't have enough tokens to keep working. Tokens. Hey, sounds like it's time to call it quits, early day.

We're all out of tokens, right? Yeah. So token maxing, is that it?

That just that in check? Yeah. Token maxing.

Okay, we got to do a story on that. Okay, um- And instead of buying more tokens to increase their subscriptions, right, we just sit around and wait for the 30 minutes or whatever the token is, and go do something else. So we're revolting against having to buy new tokens just by not doing it.

Mm. But that eventually is going to cause developers to go figure out a way to do it on their own with some of these- Mm-hmm ... smaller free models, that maybe actually have less hallucinations.

At least you can get an AI bomb on them, so you know what you're using. And potentially more security, less data loss. There's a whole bunch of goodness in doing it that way.

Agreed. Yeah, there is. We have talked about it.

We've talked about building our own and getting out of Copilot and GitHub for that. Wow. Because GitHub's a place where you're supposed to get stuff for free, right?

It's hard to pay for things when you're-When you've gotten used to how you interact with a company- Yeah ... and it's part of it, it's part of the relationship Just wait until you have to go invest in GPUs and power and your power bill for those things. Yeah.

We shall see. We're going to shift gears a little bit and go back to what we referenced earlier, tech conferences, and there was yet another AI conference, so to speak, in Vegas this week, and it was the Google Cloud Next conference. And, I was going to start with the great Tracy Reagan about this.

I guess the theme, and I talked to a couple people who were there, is we're moving from training models to running inference at scale, or at least that was their takeaway from this show, and that Google's positioning the TPUs as purpose-built for the shift toward inference-heavy agent workloads. With separate ship architectures for training versus inference. They made a ton of announcements.

It was hard to keep track. They've among with NVIDIA, for instance, for a GPU alternative market form. I guess, Tracy, are we watching a real GPU alternative market form, or is NVIDIA still default for the hardest workloads?

And maybe in addition to that, maybe we can talk about just what your takeaway, what you gleaned from this show. Well, I didn't get to go to the show, but I did read the article. Thank you.

And I felt like at Google's announcement around these next gen TPUs and this kind of expanded collaboration with NVIDIA, which I thought was interesting, kind of shows how these hyperscalers are now starting to compete on not just models, but silicon and networking, orchestration, and inference. Like you said, it's not just better models. So the real battleground, I think, what this tells us, is becoming the runtime platform where enterprise AI agents execute.

Now, we talked about this last Friday a bit and how these platforms are starting to become-- This is the direction companies are moving, which is probably why Apple is thinking about they've got to get in the game, and now they have a different CEO who's more of a platform person than he is a salesperson. The one thing that I was excited to see, though, in that article was in Google's kind of new AI security agents, that they're kind of beginning to signal this move beyond just alerting and dashboard towards systems that can actively investigate and respond to threats, which is something I'm interested in. So instead of humans triaging alerts after the fact, their agents can do some threat hunting and analyze telemetry and prioritize exposures.

That would be amazing information to be able to pull when trying to address vulnerabilities, because we are still struggling with SBOMs, and we haven't even started down the road of AI security. So these changes, this is kind of a real platform shift happening that is really going to support an AI infrastructure, that it can become distributed and secure and more policy-controlled by design. And I think that's exciting news.

Tracy, I think this is a thread that's going to run through every Friday for a while. I think so, too. I hope security becomes part of the conversation, though.

Mm-hmm. It's so critical for these large infrastructure, these large platforms, that's going to be supporting us in our AI efforts, build security into it by design. It has to be there.

We can't have memory poisoning, okay? We just can't have it. We can't have jailbreaks, and we have to know how to version prompts and track prompts.

That basic stuff, those basic guardrails, is what's going to allow us to push AI to where it needs to be. And without it, we are-- The government just reauthorized this CBR program, and I was digging through all the different offerings, and so many of them were around using AI. Mm-hmm.

And I'm thinking to myself, man, the government, if they really do pursue this right now, they're opening themselves up to more danger than I think that the AI can solve. Because most of these vulnerabilities, 50% or more, are targeting the US government. And China is not sitting back on their heels.

They're figuring out how to leverage this and weaponize it. So I was happy that Google talked about security. I was very happy.

" It has to at this point. We don't have an option anymore. We really don't.

But I will say, Google's problem is that Google doesn't know what type of company it wants to be, right? We talked about Apple, right? Think about, go back to the Apple discussion.

Apple turned itself into a consumer electronics, a B2C company, right? They really got out of the business-to-business bit world. They're a hardware plus entertainment and media.

Google's an advertising company, a search engine company, a car company, a self-driving car company. They're an AI company. They're a hardware developer.

They're a cloud provider. They're so many different things. A sovereign cloud company.

Right. " And can you continue to be successful on that? And more fundamentally for Google is the ad world is changing.

We've already moved away from search engine optimization to AI engine optimization. AI, and I can't-- There's all sorts of acronyms that are coming up, right? And very quickly we're going to see a change in the advertising dollars, which is the- We already, yeah, we have in a sense, right?

dollars thing that drives them, right? Right. I think Meta overtook them, or is overtaking Google in terms of advertising.

So Google's got to think. You're right, Jack. Go ahead.

I'm sorry. I was just going to say, at some point, that cash cow is going to not be able to sustain their spend. Does Google want to be a chip company?

It takes a lot of volume to produce chips in an economically profitable way. Apple can do it because they sell millions of iPhones, and millions of iPads, and millions of Macs, and earbuds, and all the other stuff. But Google's volume isn't in the tens of millions, it's in the tens of thousands.

Those chips are very, very expensive that they're producing. Can they charge enough to cover the cost? Like GitHub, can it charge, right?

It's the same question. It's an interesting parallel you made between Apple and Google, because in a sense, they both have their cash cows, as you mentioned with Google. And then with Apple, it was the iPhone.

And these things kind of... Eventually, you can't count on the consumer continuing to buy this. They're going to have other options, especially as fast as things are moving.

So, it brings up an interesting point about Google and where they are at this point, and it's something that's kind of under the radar. It doesn't get a lot of coverage, actually. I wonder why.

But it begs that question. I was going to ask, maybe, I don't know, whoever feels free to... The infrastructure numbers are getting pretty extreme, so what's the practical enterprise scale footprint for agents and who pays for it, given what Google's doing?

Man, you got some loaded questions today, John. I- You thought about this. I'm sorry.

I'm sorry. Yeah. John prepped more than we did, Tracy.

No, I'm sorry. No, I didn't. No.

No, but these things just flash through my mind sometimes, and I'm just wondering where it all is headed because it's all moving so fast. But let me get back to something that Tracy and Jack mentioned, which I think is very perceptive on your part. You always are perceptive, but especially on this.

I'm noticing, and tell me if I'm crazy, that the security is becoming more of a topic. It's not topic A or B, but it is mentioned. It is nestled within all of these announcements.

Even open AI. But this is starting to gain more traction among these companies when they make these announcements. They're trying to kind of foretell or foresee where this is going, and that must be refreshing to you all.

So refreshing. So refreshing. I'm so happy that they are bringing the word security up.

If anybody's watching right now what China is up to, they passed a law recently that basically says... And they have a lot of engineers with a lot of education. If you find a vulnerability, you got to report it now to this national vulnerability database that only China gets to see.

And they had two other public databases that we could see, and those two other databases have flat-lined. According to, I think it was MITRE, they said that they believe China has about 10,000 more vulnerabilities reported in their private vulnerability database than we know about. So they're weaponizing this.

They're using this as a potential cyber weapon. And we have to be serious about it, and our government's not being totally serious about it, even though NIST has started working on it. But companies like Google, and AI, and Apple, and Microsoft, and Meta, anyone who's trying to support these platforms have to start talking about this.

And they have to do it, they have to be motivated to do it, all on their own out of the decency of their hearts, and that's really hard. Tracy, I'll agree with you with all but one thing. I don't think the goodness of their hearts are ever going to come into it.

I think they need to do it. For those of us who have been doing this for decades and saying security by design, security has to come first, we're all there. The reason I think it's come up the list and it's more of a topic now is the speed and scale of technology and computing and infrastructure.

So every problem that you had that you could cover up or recover from, now you don't have time to do that because it scales so much and it goes so fast. So I think security, it ends up becoming a necessity. It is refreshing to see it.

I'm glad that organizations are walking, but they need to be running. Yeah. And that's really where it's going to be.

And the government isn't going to do it for us. We don't even have someone in charge of CISA right now. Yeah.

I know. I know. Jeff, don't get her started on that, and I'm sure Alan's not happy.

Yeah, I know. This is a touchy point for Tracy. That's Al's sore point topic.

It's a sore point. Yeah, Al likes bringing that up. Yeah, Al likes to push me.

It annoys me every time. Yeah. Yes.

But NIST is still working. Mm-hmm. NIST is still out there.

They are still doing work. They're doing work around the AI topic. They're doing really good work.

And companies like Google are starting to present more of the new- ... attack surface around AI. So they're at least talking about it.

But John, to answer your earlier, more difficult question- Sorry ... who's going to pay for it? Who's going to pay for it?

Eventually, all of it always trickles down to- Mm-hmm ... we, the consumer. Right?

Well, look- It's just going to be more and more expensive. And to Jack's point, I can't even tell you how glad I am I don't have to worry about a-Google clicks anymore and trying to bid on clicks. I think that AI has solved that problem for us, and Google's got to figure out a better way to do it.

Well, and I will just say from a Google, from a security point of view, my heart was warmed that Google bought Mandiant, they're buying or bought Wiz, right? So they have a security division. It is self-centered because they needed to focus on themselves because they had a lot of problems they had to cover.

But this is good. Microsoft is very, very big. If you look at Google as competitors, Microsoft has a huge security division.

They have a lot of investment in SIEM and in other different tools in the identity space. Jeff's familiar with all the stuff they're doing on the identity space. Amazon is not necessarily in the same spot right now, and so when you think about the big three, that's one of the things I think of people who are looking at their cloud environment and as they dive into their AI environments, where does security come to the picture and what does your cloud AI provider provide you in terms of security?

There you go. And Azure has so many more complexities over AWS. Developers are tending to like AWS better than Azure, and I think that's becoming a problem for all those reasons.

Hey, we're running out of time. This was great you guys. You guys were fantastic as always.

Thank you, Tracy, as always. Jack, Jeff, and Andrew, welcome aboard the train. We hope to see you again soon.

I'm sure we will. I really appreciate it. And that ends another crazy week.

This is not sustainable, but it just keeps going. AI is driving us to heights that we haven't seen and fears that we haven't believed before. But in any event, thank you for watching "Techstrong" gang.

We have plenty more programming on "Techstrong" TV. Alan and Mike should be back next week, although I'm not sure about Alan. He might be in Europe for a couple more weeks.

He might be a little bit on vacation, rest and relaxation, but Mike will be back and so will I. Thanks again for watching everyone, and we're out of here.