AI Training Shifts, Electric Bill Scrutiny, and Oracle’s Data Center Setback | TSG Ep. 991

Hey everyone. Welcome. You know, you need two jacks or better to start Textron Gang today, but we've got 'em.

You're watching Textron Gang. Hi everyone. Happy Monday.

It's the Monday before Christmas. What are we doing here? Well, we still got today and tomorrow we won't be doing a show Wednesday, and then we won't have live shows probably till after the first of the year after that.

But nevertheless, savor these last couple shows of 2025 because 2025 is going to go down as a really a boundary layer year. And if you'd want check out my shimmy says from last Friday where I talked about this. Um, I do believe it's the first year of the 21st century.

I know you think my math's crazy, but it's true. Um, let me introduce you to our gang today. We got some hard cuts.

I mentioned a pair of Jackson. Better to to go over stuff. Got our friends Jack Poller and Jack Gold as, as well as Mike Ard and myself.

And we've got some interesting stuff to go over today, gentlemen, welcome. As we get ready here for Christmas and New Year's and the holiday break, the news doesn't stop 20, as I said earlier, 2025 is gonna go down in the record books as a, an interesting year to say the least. Mike, what do we got for today?

Well, at a time when everybody's trying to figure out how to gain AI skills, there's a lot going on. 5 billion. IBM is partnering up again with Pearson to kinda revamp their entire online learning category log using AI agents.

Sorry about that. And um, we also have some senators probing some other stuff. We'll get into that later.

But the issue seems to be the following. It's clear that we need to re-skill a huge part of the workforce, but it's not clear to me that organizations are willing to invest in that just yet. At the same time, we don't have time to go sit in formal programs anymore for, I'm gonna sit around for three weeks to learn something.

So we're trying to use AI to deliver the right content at the right time. So the AI agent's gonna observe what you're doing and notice that you might be struggling and then pop up with some things that say, Hey, um, you might be interested in learning this now because it seems like you're trying to figure out how to do this particular thing, ma'am, in some quarters that might be creepy and others might say it's useful, but the whole way we train people seems to be in need of a revamping Jack. What you take on what's going on here.

I sometimes feel like organizations are kinda like the Stein branders. All they want to do is hire the best and they don't wanna invest in training. But do we need a farm team for ai?

Yeah, we do. We certainly need more than we've got, uh, you know, various statistics out there showing it's what 80%, 90% of, of AI projects are failing at at large enterprises to, to achieve the, uh, TCO or RROI goals that they've set. There are a number of reasons for that.

Some of it is because, you know, there, there are not, you know, they're, they're not being able to pull in the corporate data into the AI models to make it work properly. Um, and uh, the other issue of course is that they're not able to redirect or, or, or, or redo their workflows to make best use of ai. But a lot of it also has to do with the fact that we're not training people to use AI effectively.

You know, when Office first came out as an example, we just kind of threw it at people and they said, here, you know, here's word, here's Excel, here's PowerPoint, here's all this great, these great tools. Go to town, go use them. But it took several years before people really got good at using those tools.

We're, if we don't train people with new methods of, of work with new methodologies, with new applications solutions, we shouldn't expect them to be good at what they're doing in using those things. And so training is going to become one of those issues for large organizations and small that will be a deterministic, uh, or determinator of how well AI is gonna work within my organization. You can't just throw these things at people.

Um, AI is good if you know which questions to ask it and how to ask them. That's really going to be providing the results that, that you want. As we move to Ag agentic ai, it gets even worse because now we're expecting AI to be able to replace Mike or Jack or other Jack or Alan, uh, with actual actions.

If you aren't able to tell it what actions you want it to take precisely what you want it to do, then it's gonna take actions that aren't gonna be very good. Uh, and I I think you're going to see a real effort by large organizations to start doing a lot more training. You're seeing it of course with the, uh, the issues that you just mentioned, Mike, but I think you're gonna see a lot more of that.

I think you're gonna see a lot more money spent there, which is also, by the way, why Coursera and IBM and others are beefing up their AI training capabilities. 'cause people are gonna actually have to buy those technologies, those capabilities from them. Mm-hmm.

A what's your take on what's the right balance here? Because there's certainly a certain amount of personal responsibility here, right? I need to be relevant and I need to get my own skills and it's part of my career, but at the same time, organizations need to invest some level of training.

So is there a balance to be struck here or what's your sense of what's going on? Well, sure, you need to take some personal initiative for sure, but the question that I would have for large organizations is do you let that kind of go out there on its own, uh, initiatives, uh, you know, let people go do their training on their own. Some will do it very quickly, some will do it over six months, nine months, 12 months a year, two years, or do you try to impact that training?

Most companies forget about AI for the moment, but most companies have some kind of training re-skilling capability in-house anyway. And so, uh, they realize that training people to do their jobs better presents really good ROI and I think you're going to see that come about with, with AI as well. I think it's not gonna be just about personal training.

Um, me going out and trying to figure out how to make this work, I think it's gonna be about kick-starting that training within organizations. And maybe it's driven by hr, maybe it's driven by other groups that are going to implement training sessions to try to get the efficiency of people up. That's, that's how you get corporate efficiency to go up.

It's, it's on a personal level, not just on the application level. Mm-hmm. Alan, do you think that this is a national priority issue?

Is this something we should be addressing at a country level? Because ultimately it comes down to our ability to compete, Frankly, gentlemen, I think you're looking at this all wrong, right? I I have a decent amount of experience here.

Having founded the DevOps Institute and we got that up to 65,000 people becoming CER certified in DevOps in about six, seven years before we sold it to people. Cert one of the largest training companies in the world, right? Um, you gotta know something about the education market.

Jack, to your point, there's a difference between reskilling and upskilling. What you are talking about is upskilling. In other words, I got a developer and I'm gonna make him an AI savvy developer.

I upskilled him. That's very different than having a marketing product, marketing manager whose job has been eliminated by AI and they gotta go become a coder or a, or a car mechanic or something that's reskilling, not upskilling. And, and the real crux of what we're gonna face in this world over the next year or two or three years, is it an upskill or a reskill?

Are people, is that QA person gonna stay a QA person or just be harness AI to become an upskilled QA person? Or is he gonna throw the towel in and say, QA is the, is the land of AI will rule that I'm gonna go become a developer or a, or a platform engineer or something else that, that's number one. Number two, prior, even prior to ai, over the last 10 years, we've seen watershed moment in, in technical skill education and training.

We went first from doing these things in person, right? There was a time Jack and Jack where if you wanted to go get a cert, you went to a class, there were 20 people in the class. It was usually two days a class, 16 hours, 15 hours, and you got a piece of paper that suitable for framing that said you were certified.

Well, 10, 15 years ago that changed. We went to virtual training, right? Where instead of being in person, you could do a virtual and then at your own pace it's prerecorded and you just take tests at the end of every chapter and you gotta score an 80% or whatever to move to the next chapter.

That became dominant. What we're on the verge of seeing here now is AI monitored, managed instruction AI training where it's no longer an instructor doing the curriculum. It's gonna be ai.

Who's gonna train you to use ai? It sounds a little crazy, but that, that's AI becomes the trainer, right? Is that letting the fox into the head house?

I don't know. But AI is the trainer, not the person, not the individual company. So I look at this Coursera and Udemi deal, and they're both big traditional trainers who now do a lot of stuff virtually.

What, what the, the other change was like people cert, for instance, DevOps Institute, we at DevOps Institute, I had 175 partners around the world who delivered that training both in person and virtually. Now with the web Coursera Udemi, they deliver that training directly. They killed that whole channel.

A whole business economics bus, you know, economy was killed off there in the training channel because people go direct to these people. But there's a reason why this is an all stock deal training as we know it is dying. If it's not already dead, this model of Coursera and Udemi, my AI will gimme that training without them.

Why the hell would I pay them? Because I'm going to get the cert from them. There was a time where to go get a job as an ops person, you needed an idle certification, right?

A lot of jobs said IL certification, mandatory security, Jack, right? You needed that C-I-S-S-P certification. Those days are done, those days are done.

Ai, you just go to AI and say, Hey, make me smart. I want to take a a security class draw, draw me up a class and it's gonna draw you up as class as good as udemi or Coursera's people cap, they're dead men walk in. I wouldn't take that stock and use it for wallpaper in my bathroom.

Now see Holland, I thought I was gonna be the only one who's gonna be contrary and think this is the insanity. This is an insane deal. It's Over, it's over, it's over.

It's, it's over. It was if, look, there's, there's, there's two parts of this. One is, Alan, you're a hundred percent right in that we went from in-person training to online training.

And that's about to change again. Um, if you look at what's motivating Udemy and Coursera to sell is that they had a big spike in online learning during COVID when nobody had to go do to work. They didn't go into their office, they went and did a whole bunch of training.

The big problem has been all along is, who pays for this? And how did they, that, you know, how does it happen? Right?

And the reality is, those companies now realize that people aren't going to pay for this because they have alternative means, as you said, AI in the web and companies will start paying for an AI enabled course and then quickly realize that none of their employees take it because none of their employees ever do this. And it becomes, uh, it's a people process problem as we like to talk about. Is that a company that goes and says, I'm gonna pay a hundred thousand dollars to Coursera to give my, all my employees access to all these courses will do that, but they won't go the next step and say, I'm going to give all of my employees a month of time that they can go and spend on this instead of their own personal time.

They can spend work time to go take their courses. And unless they do that, nobody's ever gonna take the courses. And then next year the company's gonna say, we spent all this money on this and we have three people take this course and we got nothing out of it because they took this course upskilled and left.

Because now they have to, and, And LinkedIn really killed their market too, Jack, Right? And, and LinkedIn bought Linda, that was, I dunno what, 10 years ago, something that was a long time ago. And all those courses are available.

If you buy LinkedIn premium, then they're all available and They're all you, you need. So let me just add, wait, Mike, let me just add one thing. This is not uniform across the world, right?

I learned another lesson I learned at DevOps Institute. So in the US less companies pay for training, right? A lot of, in the US a lot of people, individuals pay for their own training.

'cause they want to be more valuable in the job market, not at their present job, but for their next job. In Europe, the the model is that companies do plan for training because in Europe, those employees often stay for their entire career and only one or two companies. And so they are truly uplifting their, their employees, right?

And, and oftentimes in Europe, you've gotta sign that. If you're taking the cla the course through the employer, you've gotta stay at least three years afterwards. So they in essence get their money back.

Then you've got India. India is a whole world unto itself, right? It's a gravity well in the amount of people who are there working in it.

And for whatever reason, the folks in India love their certifications. They love taking these courses, they love getting certified. They don't like paying for them.

So we had, we had different pricing. We had worldwide pricing and then India pricing. India pricing was 60% less than Europe and, and North America off the bat, off the bat.

And then trying to get the money was, was a lot of fun. So, you know, if you have the India market, which I say is a gravity well, and they're saying, should I pay for it or use the AI to make it? It's, I I think There's a couple of things at play here that we didn't mention.

One is, I think a lot of people are getting training on YouTube now, and it's free and it's a video and somebody watches that. And that's part of the thing. I'm not sure I agree with the stock not having value though.

'cause these companies at least theoretically, should have content that's behind a firewall somewhere that is of value that people will pay to get access to. And it will not necessarily be something that chat GPT has seen. And if they don't, they should, Should.

Well, Mike, what are they gonna have that chat? GPT doesn't have? Well, I'll give you an example with, uh, AWS who's talking about the following thing.

They're revamping their certification programs as we speak, and all of that stuff sits behind our firewall, But there's nothing proprietary about it other than the certification itself. I've been down AWS offered DevOps training eight years ago, nine years ago. We looked, we had people take the course.

It wasn't, it was, it was a, it was very AWS specific, but nothing that you couldn't learn from just being on AWS What you couldn't get was a certificate from AWS unless you paid them. And that's what, it's the same thing with the C-I-S-S-P. You've got plenty of security qualified people.

They don't got the sheep skin as the old commercial used to say. Right? So the, the sheepskin as you described it, is also changing though.

And what AWS is talking about now is, um, smaller certs for specific skill sets that you can demonstrate. And you're not gonna take some, you know, six month certification program. They're gonna give you, you know, a a, a micro cert essentially that says, you know, yes, you've proven that you have the ability to provision this and here's your gold star or whatever you want to call it.

And I think that's gonna be more of that in the age of ai because a lot Only, but Mike, only if employers demand it. If employers don't demand it, it's useless. How many of you raise your hand if you go through LinkedIn and you see your friend's post, I just obtained a new certification.

I, I I presented at the Linux Foundation event, or I took a Linda course on, on oasp top 10 B-F-D-B-F-D. It's a nice LinkedIn post. I'm sorry.

There is one issue though, Alan, and, and, and I think we need to, to, to look at this as well. AI training isn't free either. So, you know, Coursera is making money, obviously by offering you a course, you can go to chat GPT, but you're not gonna get a good course if you're not a subscriber to chat GPT 20, 20 bucks a month.

A lot cheaper than that Coursera cost course. Well, and, and, and I will, I will argue that as well, which is that there are plenty of people we know who are experts in these things who will gladly put up a YouTube thing for free and take the YouTube ad money that whatever money they get, because that's the way the, the information wants to be free. I mean, this goes back to the days of the early days of the internet and information wants to be free.

And it is. And how many of you subscribe to 20 different online news sites? We don't.

Right? Because if you, if you go and you look at it and there's a New York Times article on something and you're not a New York Times subscriber, you say, okay, let me Google that and see what other news article I can find that has the same info for free. And that's the way the world works.

And unless I agree with Alan, unless somebody demands a cert for a job, and we see this, we still see this a huge amount in the security environment. Um, but you know, I, I consult with the Sands Institute, and this is an issue for organizations that's existence, is to certify people. It's how do you convince an employer to allocate the time and money and d for the cert so that they can demand it and have somebody qualify and need it?

And do they really need it to do the job? And that's the big questions everybody's struggling with. Do you need a cert from chat GPT to be able to use chat GPT?

Well, I can guarantee you that every, all of our parents here are using it without getting certs and, you know, Right? And then also, and you know, when I talk to DevOps folks, you know, they're of two minds. One is they get the cert, although they hate it because they had to pay for it, and they're only doing it so they can get past some sort of search engine on an HR system somewhere that's looking for keywords.

But you know, when you go talk to people who hire folks, you know, they'll tell you every time, if if it comes down to somebody they know or met at a conference that they know has the skills and can demonstrate it, they'll pick them over no matter how many certs they got. It. This hence why we sold, oh, hey, we're 22 minutes into this one, man, we, we could talk about it all day, but we can't.

We've got other stuff to cover. Let's take a quick break. You're watching text on gang, You've earned it.

The spotlight, the responsibility, the weight of teams, companies, and entire industries fall on your shoulders, lives depend on your decisions, your home life included that work you are protected physically and digitally. Nothing gets through your team without a fight. But in a globally connected world, everyone sees you, including those who mean to cause you and your organization harm.

And now home your sanctuary attackers see an opportunity, your digital front door is wide open. And what compromises your home can breach your boardroom. Because the devil's greatest trick isn't targeting your workplace firewall.

It's convincing you that your personal life isn't at risk. Black cloak, digital executive protection, defending the new attack surface your personal life. Hey folks, we're back and we've been talking on this show about AI and power for a while, and now some senators are starting to ask some difficult questions about, well, are all these AI data centers increasing the actual cost of electricity for the average household?

I think we've been kicking this around for a little bit, and I'm not sure that the evidence is solid yet, Jack, but what's your take on what's going on here? Well, I think, you know, first off, there's politics going on and then there's the rest of the world. In the real world on the political side, you have, this is an issue that Senator Elizabeth Warren can get ownership box and say, Hey, look, I wanna hold inquiries and say, you know, the big bag corporations are doing in the common man.

So that's one part of it. The other part of it is, and and still politically related, it's how much power is AI using and how much power do we have and how much power do we need? If you look over, and I did a quick search and research for this, uh, segment in the last nuclear power plant the United States approved was in 2012, so that's what, 15, 14, 15, 16 years ago, something like that.

In that time since then, we've added, uh, about 500 gigawatts of power. Now you look at what AI has committed for and, and ai, sorry, open ai, open AI itself wants something like 30 gigawatts in the next year. So that's, um, almost 10% of what we've done in the last 14, 15 years.

They wanna consume in the next one to three years. They wanna build data centers and do that from a data center perspective. So, and, and by The way, that's just open ai, right?

Jack? That's Just open ai, right? So if you look at, and, and that was just an easy one because they're very public about what they've committed to do.

And that's, you know, um, if you look across all of that, you can figure that that's probably not four and a half or five gigawatts, but probably, or sorry, 30 gigawatts is what they wanted. It's probably more closer to a hundred to 150 or 200 gigawatts across all of the major, uh, AI data centers that's close to a quarter to getting, you know, getting up there in the amount of power we've added over the last few years. At the same time, our government has made it very, very hard to open power plants, and in particular the one type of power plant that we really need the cleanest and most efficient power plants that we can do, which is nuclear.

So if we're not gonna do nuclear and we are going to expand our power requirements, then we have to build coal and natural gas, which are dirty and noisy and et cetera, et cetera. And regulations. And getting new power plants built is very costly and time consuming, getting through the government regulation process.

So you have more demand chasing after a restrictive supply. Prices are going to have to go up. It's not necessarily evil power companies that are trying to do in the consumer.

It's basic economics of supply chasing demand or sorry, demand chasing supply that isn't caught up yet and the government not allowing us to build quickly enough. There's another piece to this as well, and that is not the, not in my backyard phenomena. Yes.

And I, and I, and I've talked about that before. When you go outside of Manassas, Virginia, where all the big data centers are outside the nation's capital, and the noise from these things is truly horrendous. Uh, data centers are very, very large.

They consume a lot of land and they produce a lot of noise both from the fans, the backup generators, and the humming of the electricity lines and the transformers. And they employ very, very few people. So it's can be, um, have a very large impact on local communities, particularly in They Property tax.

Yep. Yeah, which is another piece. I gotta step up on my soapbox here a second.

I'll step down, Alan, so you can step up, step up. So Jack, you're right, there is political theater and then there's the reality. But every once in a while, political theater matches the reality at the heart of the political theater.

Here is this basic principle. You wanna build AI data centers, you wanna power AI data centers. You are gonna make a lot of money off of those AI data centers.

Why the hell should my electric bill go up? That's real. That's real, number one.

Number two, you know what, I'm all for nuclear power. And I agree we should, we should look at authorizing and investigating. But let's not rush to bring nuclear power on without proper safeguards, without proper inspections, without proper guidelines.

Otherwise, let's build it in Chernobyl or next to Love Canal so we can get ready the next generation of irradiated cancer survivors, right? I'm not saying nuclear is not safe, I'm saying you don't cut corners for safety, right? This is out of the movie.

What was the movie, Karen? Uh, well, China syndrome was one to three mile out. No, but Silkwood Silkwood wasn't it, wasn't it Silkwood the movie in the book True Story.

We, we don't want that. And Jack, when we talk about, hey, let's do some clean efficient energy, well, why the hell is our government putting the kibosh on wind and solar and renewable clean energy that you talk about our capacity growing, our capacity has grown from those energy sources, right? So if you, if, if the political theater is, you've got a com, a country that just, the president went online and said, Venezuela stole our oil, right?

When this isn't about oil. Yes. Next gen nuclear maybe.

Yes. To solar, yes to wind truck media themselves. Yes.

They announced that they're merging with a company to go build fusion energy plants five years from now, 10 years from now. It's always five to 10 years from now. That's the real issue here.

I don't, Elizabeth Warren isn't wrong here. I let open AI pay for their energy, let Meta and Google and Microsoft and Amazon pay for the energy that they wanna power these AI plans for. I'm already paying enough for my electric.

It's wrong. And I'll actually give you a real world example here in Massachusetts. Every month I get a bill from the electric company.

It's also true on gas where they have added the state has mandated an additional cost for upgrading the infrastructure. Now, who's gonna benefit from that infrastructure upgrade? It's probably not going to be me.

You know, the lines on my street are 50 years old and then I'm not getting up updated anytime soon. But the data center that they wanna build, you know, down the street or, uh, you know, uh, open AI putting in something nearby, those are the folks that are gonna benefit from that. It's the new construction Absolutely.

That they benefit From. So on my bill, so we have FPL far to power and light. Here's an interesting thing on my bill.

Every month they show me how much of my energy consumption was, was done using solar, right? And it, and it goes up a little bit every month. This, this is a fundamental issue we're gonna have to face as a nation here.

And I'd make it an election time issue. Why should we pay for the tech bros data centers? I don't understand why.

If they want their data center to be approved, they just don't show up and tell everybody in the town that they're gonna pay their electric bill as part of the deal for billing The data. 'cause they're too busy telling them all the jobs it's gonna create. Well, that, and they're not making any money to begin with.

Where's the profitability for these guys? Vin Krishna, the CEO of IBMI, I wrote about this about two, three weeks ago. He had a dead on guys.

Jack, to your point, about $8 trillion isn't going to need to be spent to build and power these data centers worldwide. That's a worldwide number. And when you look at the economics of what the return will be on that $8 trillion, understanding that every five years, you need to, you need to regen those data centers, right?

Re re you know, upgrade the, the, uh, infrastructure. The emperor has no clothes here, guys, unless something fundamentally changes. This is a losing proposition.

Yeah. They need 800, $800 billion a year just to pay the debt. Can I tell you something that happened here in Westchester last week?

So there was a meaning about the powering down of the Indian Point reactor, and one of the people there stood up and said, well, you know, we're powering this down and maybe you guys might wanna use this land for data centers. Well, the whole room flipped out. Oh, I'm sure.

But, but Mike, this is a perfect ex. I lived in New York. I remember when Indian Point came online and, and, and everything, we knew that it was end of life thing.

Why didn't we build a replacement for Indian Point that went through the regulatory process and is ready to come online instead of saying, oh my God, we gotta do something right now. Cut all the regulations, flip the switch. It's just so, I it's p**s poor planning well Be, because the answer is because people like you, Alan, who say, I don't want nuclear because it's scary.

No, no. And the reality is, Jack, hear me out. I didn't say I don't want nuclear because Yeah, Yeah.

You know, if you're gonna Silkwood and Meryl Streep and Nuclear and all of this stuff, that's, I think next Gen nuclear is very different than what's an Indian point right now, or what was in Three Mile Island, right? And, and all I'm saying is don't cut corners on regulatory practices. And I don't, I don't think anything anybody is arguing for that.

I just think I, and, and I'm not, I Absolutely are arguing no mean, that's just what the Senate, there are bills in the Senate and the, the administration is pushing to cut and they're, and they're preventing states from putting in regulations on this. They wanna cut, turn, drill, baby drill, turn it on right now. And, and the answer is because, you know, look, I, my, my former home was in the, the wonderful state of California where nothing can ever be built because of the regulations they Still have.

You can't build, they still, you can't build power plants. They, the biggest economy in the world, they're doing pretty good compared to the red states Because they're buying their power from their importing power rather than generating it themselves. And at some point when the, I mean the power's gotta come from somewhere to, to drive Silicon Valley data centers.

Jack, you, you're say, you're saying, okay, let me show you the extreme case. I'm telling you, bring next gen nuclear power online. Make sure it's, it's done right.

Don't cut corners. There is another piece to this guy. So let me, let me kind of throw something in here.

One of the problems with nuclear is it generally takes five to 10 years to get a plant built and up and running for five to 10 years. The power companies are spending billions of dollars to get that place up and running with no revenue to show for it. And so from a financial perspective, if it really takes and forget about regulations, regulations are not, and, and, and Alan, I agree with you, it has to be safe and all of that stuff.

But if it takes five to 10 years to build a new plant, get it up and running and generate no revenue, but still have debt that I gotta pay down, who's gonna pay for that? We do, but we always have. We always have.

And that's okay. No, that's the pushback though. People don't want to pay that, that, that shows up in your bill.

Well, the, The and, and the, the, the alternative is to pay forward dirty energy like, uh, uh, gas or coal or, or solar. Solar is not clean. So here in Florida, Jack FPL has invested a ton in these solar farms or whatever you wanna call 'em.

Um, guys, we need, you know, the shimmy says that I did on Friday is that 2025 was the first year of the 21st century. We're done for the first 25 years of this century. We've lived off of and worked in stuff that was done in the 20th century.

We need energy for the 21st century. Obviously, if we want to do this AI stuff, if we want to live without glowing at night, if we want to, you know, realize the promise of what's before us, we need a plan for the 21st century. And it's not drill, baby drill, that's the issue.

We, and we gotta do it right. I don't, We could spend any time, I don't suppose we could spend any time making the existing grid system more efficient than it is 'cause that don't think That that has to be done too. And better batteries and all of these things.

We need a 21st century energy plant, not, not John Rockefeller's plant. And, and, and people overlook the fact that the current energy grid in this country loses about 20 to 25% of the electricity just in transmission. I think it's more jagged pioneer, be More sure.

But that's a significant chunk. You know, I don't wanna get emotionally involved here. Let's take a break and come back to our next thing.

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

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

Contact us today and tell your story to the world in the most powerful way with Textron Group. Hey folks, we're back and we're talking about the, well, the AI bubble again, apparently Oracle has had a setback on funding for a $10 billion data center deal. And it's raising questions.

And at the same time, you know, I also happen to notice that Morgan Stanley is funding some sort of data center initiative using junk bonds. So I guess, Alan, are we running outta money here or are people saying that, you know, we don't wanna fund these 'cause maybe we're over provisioned. What's your take on what's happening here?

Well, you know what, this is Mike. That's the world's smallest violin playing hearts and flowers for Larry Ellison. Um, look, the man increased his net worth by about 40% when, when a, a bunch of these, uh, Oracle Cloud AI deals were, were announced with Nvidia and open AI and, and Blue Owl and et cetera.

And now, you know, what goes up must come down. Gravity is no one defies gravity in spite of the song in the movies. You can't defy gravity.

I I, I wrote about this a couple weeks ago as well. The real issue is, and it goes back to our last segment, no, not the nuclear stuff, but the, but the stuff about, uh, what Arvin Krishna wrote about the economics of these things, of these data center factor of these AI factories. They're, they're spacious, if not downright fraudulent.

I mean, it just, it, the number not fraudulent in terms of a crime, but the numbers just don't add up it from an arithmetic point of view on given current, you know, the economics of this currently. It just, you know, whether you're talking about tokens and what tokens are really gonna cost, whether you're talking about where the energy comes from, whether you're talking about what you can charge for, for this stuff, I, I think we, we got a little ahead of our skis and, and now, you know, gravity set in All Jack Gold. If I take away all the investments that are being made in ai, what is your sense of what's the real state of the economy and what are people kind of actually spending money on?

Because from what I could see, at least the numbers that suggest that AI accounts for half of the GDP growth. So maybe we're not growing as much as we should be because, well, everybody else is not in the AI Business. If we had real numbers, we'd know, wouldn't we?

Yeah, I was gonna say the same thing, Alan, if you look at it, I mean, it's all circle financing, right? Uh, Nvidia in invests in my cloud company for a billion dollars, but makes me buy a billion dollars of your chips. Essentially.

You're giving me your chips for free for my company. Uh, so it, it's really, really, really hard to tell, uh, until we get to a point where the investment is more transparent and less circular, then we'll really know. By the way, there is another issue around all of this that, that, that people tend to gloss over.

If we're spending all this investment, all this money, and there's a lot of money out there, let, let's, let's be clear. But if we're spending all this money on ai, what technology, what future technologies are we not spending on? What's it taking away from?

And there's always the risk that if we're spending so much, you know, it's, it's like the, the the, you know, the, the, the web, uh, craziness, right? Of the, of 20, 25 years ago. Uh, everyone is spending all kinds of money, you know, a OL and Time Warner.

Great example, right? W where's a OL today? Uh, but there are new technologies that are potentially coming in line that we're not investing in, in, um, in healthcare, in, you know, like the last segment in, in nuclear power that we can actually deploy quickly and cheaply.

There's a lot of stuff that's not happening. So it's not clear how all of this money that's being spent, uh, is going to, uh, help us. And by, and by the way, Alan, you mentioned earlier that these data centers have to be upgraded every five years.

It's more like two years. If you're an AI data center, you need the latest NVIDIA chips, or you're not a data center anymore. And they're only, they're only good for a year or two.

In fact, they, I saw a statistic that said about 30% of them are failing per year because they're driven so hard. Yeah, That's a lot of, Lemme piggyback on what you were talking about there though, Jack, here. Here's the situation.

The fact of the matter is, this first phase of AI expansion that we've lived through, let's say over this last two, three years, was primarily funded by the cash reserves of the Mag seven. Well, minus Apple. I don't think they've spent a dime on AI to speak of, of being, yeah, I'm being face, I'm being face facetious.

But you know, it was primarily spent by the, by the cash reserves of the Mag seven. And you know, you could chalk this up to fools in their money, right? It's their money.

They could do what they want with it. But what we're seeing now is a lot of this AI spend or plan spend is coming from VC money funds, money, debt financing, you know, more traditional ways of, of, uh, capital expenditure here rather than just, you know, because the fact is coming outta COVID, the Mag seven, were sitting on hordes hordes of cash, you know, collectively in the trillions of dollars. But as that pile goes down and we start financing this stuff with Wall Street money, you know, pension fund money and stuff, do they have pensions?

They're still pension funds, public employees and Middle East money. But when that, you know, when things don't work out like this deal here and, and the stuff hits the fan, well, then it becomes a little bit more painful, doesn't it? Right?

Because you're not just saying, yeah, Google will make it back, right? Ellen, I think, I think there's, there's another part that's very important to layer on top of that, which is financially all of the, the Mag seven, or at least the, sorry, let's look at this way. The big three, the cloud providers, Google, Amazon, Microsoft, all have cash engines, cashflow engines that are just throwing off cash left and right.

That's something that Google does not have at the same scale. So from a, sorry, not Google. Um, Oracle does not, sorry, Oracle.

Oracle does not have, Well, I wouldn't take up any collections or you know what they, but No, no, but not, not, they don't have the ability to replenish the coffers in the same way that, you know, Google has that, that Google, Microsoft, and Amazon have. Now, I bring that up and I got tongue tied over Google because Google's going to be the next one hit by this because the entire search engine business is changing completely. And I think we're going to see an interesting impact on Google's cash flow in the next one, two years as people are massively changing from going to Google for searches to going to AI for searches.

And what is that gonna do for Google's ability for, to get Wall Street money to finance this stuff? Well, Well, what they're doing, who's Google's competition? Open ai, perplexity, anthropic, let's call it generative search.

And Google is betting that they're gonna become a generative search company. But you know, Jack, it's funny, I I did a, I did a shimmy says last Thursday, aren't sure we break up big tech. Now, I will tell you historically, I'm not a big breaker upper, I, I don't believe in that.

I, you know, the only one I've seen in my lifetime of sides was the at and t breakup. And you know, it, it worked out okay, but I watched a debate over at Johns Hopkins University, uh, I for the Starrs, n Carlos, Argo, whatever you could check out my article and text strong it. And they, this wasn't your presidential debate where you have one guy who can't put two words together and another one that lies all the time.

It was smart people arguing both sides of this. Take Google for instance, if you took the Google search business without the AI business, it's not long term. I'm not putting a lot of money into it, but if you, but if you give it ai, it, it, it will compete with the best of them.

But does it need YouTube? 'cause YouTube's the second biggest search engine, right? If you spun YouTube out of Google, you spun Gmail out of Google, you, you know, you spun some of the, and then Google Cloud out of the search business.

And the same thing. If you spun AWS out of the Amazon retail business, if you spun Instagram and WhatsApp, WhatsApp out of Facebook, those are gonna be all viable companies. You're gonna have a lot more competition and maybe we'd all be better off.

No. Hey, I don't, we've seen that. I mean, Oracle's gonna now have TikTok.

There's gonna be a lot of money that we made there as well. I know everybody's passionate about this, but I think I'm just gonna have the last word here 'cause we're running outta time. So go ahead.

Hey, AI brother, can you spare Adox or Megawatt? Yeah. It's crazy stuff.

Megawatt. There you go guys, thank you very much. I appreciate you coming on today.

Thank you for watching. I hope you get as jazzed and juiced about these things as we do. Um, listen, it's Monday Christmas is coming this week we have one more show tomorrow.

It's gonna be a special event where we're gonna have an extra large gang panel with a lot of our favorite gang folks. But we're only, it's gonna be a different panel for each segment. One segment will be on ai, one will be on it, and one will be on cyber.

So stay tuned for that. Stay tuned for Techstrong It do go check out those. Shimmy says episodes I spoke about.

They're pretty cool. Read it. You know, we, Mike and his team do an amazing job.

We publish a ton of stuff on our text Drunk Sites, as well as on text drunk tv. If you wanna stay in the know, that's your way to do it. Don't have to go pay for some bogus education.

Um, until then, until tomorrow though, there's Alan Shimel for Text Drunk Gang. Thanks for watching.