Techstrong Gang – August 27, 2024

Hello everybody. I'm Mike Biard, and today we're talking about IT budgets for 2025. Then we're gonna have a little chat about, well, just how much AI is there in the enterprise these days.

And finally, we're gonna be talking about why a lot of organizations want their data back from those SaaS applications. You're watching Textron again. Hello everybody.

I'm Mike Ard, and today I am in the IT conference capital of the universe, otherwise known as Las Vegas. And joining me today is Guy Courier who is at home, I think Guy. Is that correct?

Uh, no. I'm in the office. You're in the office.

All right. But the office is in Austin, Texas, correct? That's right.

Awesome. And then of course we have John Willis, who is joining us once again from I think, your home as well. Is this a rare moment?

Auburn, Alabama. All right. And then finally we have Mr.

Steven Foskett, who joins us as well from his perch over at the Futurum Group. Steven, how are you? Excellent.

It's good to be here in beautiful Hudson, Ohio. All right, let's get started. There is, um, the first in what will undoubtedly be a series of reports about projected spending for it in 2025.

This first one, I, at least that I've seen, comes out Forrester, and it seems to suggest that things will be steady because there'll be an increase in spending, but it's on par with inflation. So people don't seem to be expecting a huge uptake. But Guy, what's your sense of what's going on with budgets for 2025?

It's, uh, at the moment there's a lot of people trying to figure out this conversation. I think CIOs are having these meetings right now. Yeah, this is a fascinating report, Mike.

Um, you and I have been watching these kinds of reports for many years, and it's, I think we're seeing a real shift, um, away from the cost control conversation. I mean, if you remember, cloud was originally pitched as a way to save money. Mm-Hmm.

Um, if you think about all of the, the sort of waves of, uh, you know, technological innovation, these big waves, internet, cloud, you know, et cetera, uh, what this reminds me of most actually is the PC revolution of the eighties, because, um, uh, ai, just like the PCs back then, when people look at it, just average, ordinary people look at it and look at its effect. They are immediately struck by its potential for value. And so the support towards greater spending, really, to me, parallel parallels what happened in PCs, if you remember, for years and years, PCs were bought and used and acted out at the, uh, economists were sitting there going, well, we're not really seeing any productivity change.

We don't really understand, you know, but people kept buying them and buying. Right? And I think in the same way right now, the potential benefits of AI are so obvious that, um, there is a lot of tailwinds for, uh, CIOs, CTOs to increase their budgets and spend at the same time they are, uh, uh, modernizing.

That was something highlighted in the report. Um, budgets for modernizing and talent to take advantage of AI to modernize better, to take advantage of cloud to transform applications talent. Uh, cost is going up.

I'm actually a little, I think Forrester and maybe some of the other research firms just don't want to go out and talk about increasing IT budgets because they've been talking about talking about IT budget control for so long. But I think there's a real potential upside next year simply because there is such business priority towards adopting ai, even when it's uncertain what the actual immediate return on that investment's gonna be. Steven, let me ask you this.

'cause part of my soul says, well, this is interesting, but the CIO doesn't control all the IT spending these days. There's just a boatload of it that exists outside of their control and the various business units. And I think sometimes if you add up all the spending in the business units, probably more than what the CIO is spending on IT budgets these days.

But, um, is the actual amount of money being spent almost, you know, nearly double maybe what the analysts are thinking when they only ask CIOs? Oh, I absolutely think that the cat's outta the bag in terms of, uh, spending on ai, um, you know, that, uh, dark spending that, uh, IT organizations have been dealing with forever. You know, it, it does remind me of, as Guy pointed out, the PC revolution, where essentially the, uh, uh, IT organization was trying to get their hands around it and figure out how to make it productive and, and how to make it corporate and everything.

Meanwhile, everybody, uh, in the sales department, uh, was going out and buying a PC on their own, putting it on their own desk, buying their own software, running things. Uh, you know, it's the same thing with SaaS, uh, as, uh, of course John saw for the last 20 years here. And it's the same thing, uh, with AI now.

I mean, you know, the cat's outta the bag on, uh, open AI and, uh, chat GPT subscriptions. Uh, you know, the, the only thing that that, that the CIO really has visibility and control over at this point is any kind of enterprise applications that are being developed using, uh, ai. And, um, and, and for that matter, I think that most CIOs are pretty positive about it.

They're, uh, hoping that they can find the right applications to enable chatbots, uh, to, for example, interact with customers on their website or to search their, uh, enterprise data sets with rag. And, and I think that a, a lot of those projects are getting, uh, up to speed, and maybe that's really what's being talked about here. But yeah, I think that, uh, organizations are spending a lot more on AI than they know.

John, are you seeing the same thing in your travels out there? Because, um, it's not clear to me who's driving these AI projects. Is it small little groups within a business unit, or is it the IT team?

I think it's maturing. I mean, it's, um, I mean, a couple of things. I, you know, I I I'm not that smart.

I get to hang out with smart people, but, um, I know a fair amount of quants to my relationship with Pacific Crest and, uh, key Bank, but you know, they, they seem to tell me that there's gonna be a tsunami of spend in the fourth quarter this shoot, just because basically what you've had is centralized budget control for quite a while now. I mean, you know, we talked about the drunken decade we had where we were just spending insanely, right? And then all of a sudden within the last, uh, sort of poster mid covid years, we got into like centralized budgeting.

Even CIOs for most things that they had complete freedom had to go through sort of CFO. And, and I hear that there's gonna be sort of a tsunami. Again, I don't know this, I'm just listening to what other people are telling me, but I, you know, I think there is, there is a lot of interesting things going.

I I love the PC example because I think it's very similar to the cloud. I agree that cloud was a smaller footprint in an organization, but like the shadow thing that we've been dealing with ever since I've been involved in this industry, which is a long time, right? I think the PC one is a better analogy than even the sort of shadow it of of cloud, but they're all similar, right?

Like, people have accessibility to technology. They can't wait on centralized control. You know, centralized control is trying to figure out what's the best way to do this.

So while they're like, I need to do it now, the only other thing I'll add is we, we recently wrote a paper, uh, called Autonomous AI in the enterprise. It was originally gonna be called Dear CIO, which was a bunch of DevOps people, original DevOps folk that would like, sort of a letter to the CIO, you know, Hey, be careful here. You know, proxying out a Chief AI officer and then proxying out all the infrastructure and technology that comes with all this stuff.

'cause at the end of the day, AI is actually really a small footprint of all the technology it takes to run ai. I mean, if you look at every, every massive AI technology stack that I've been seeing, or you see in a blueprint, you can count on your fingers, the things that are really directly AI related, most of its Kubernetes, Kafka, Redis relational database. It's a massive infrastructure.

Our biggest fear is that you have a shadow, uh, IT infrastructure problem where chief AI officer are gonna go out and build out things. And, and we're already seeing some of the vulnerabilities with the SAS is already like the hugging face vulnerabilities with, uh, uh, infected models where I think they're, they're throwing up infrastructure and not doing their ABCs in terms of things we all know, and this Panel need to be done to run infrastructure. This is a really interesting point because, um, uh, the, the truly massive investments, um, almost have to happen on, you know, an industry or an institutional basis from service providers.

I mean, one of my little hobby horse just lately is there is no such thing as an AI application. They are AI services or whatever you want to call them, but an ai, you know, what gets called an AI application, even like a chat, is really a chat application with AI, as, you know, an essential element of it. And so, uh, when you're thinking about an enterprise context to begin with, you can incorporate services that are externally hosted AI services that are externally hosted.

You have this data sovereignty and protection issue. So the rag or tuning side of it, um, you know, and the inference side of it may be something that you host yourself, but you're not talking about this, the, the, you know, infrastructure footprint for that that you're talking about for trading. So, um, as pervasive as it is, I think John is really right, that you don't necessarily have to provide that high level of investment to take advantage.

But then on the third hand, what is really what's triggering my thinking about the parallel with, uh, the PC era is back in the PC revolution area, PE era, people were buying these systems that they didn't even know how to use. And I think we've already seen a lot of that as AI spending ramps up, people doing things like buying $10,000 GPUs that then sit, you know, uh, in storage for a year because they actually didn't know how to use it. They just wanted to make sure to have, Yeah, it's, it, we just recorded two podcasts on the, uh, tech field, a podcast, uh, two in a row that kind of go together.

The second one just published this morning, um, the first one was called AI is just a Fad. And the second one called AI is not a Fad. Uh, the AI is just a fad goes to your point right there, guy, where essentially the thing that we're hearing about AI right now is, uh, these building out these massive GPU clusters and AI training and as much electricity as Indiana and things like that.

You know, I mean, that's the fad. You know, it's not a fad in that it's completely useless. It's a fad in that that's what everybody was talking about.

But the second half of that AI is not a fad. Well, it's the rest of what you're talking about. Essentially, we're gonna get real with this, and we are actually starting to get real with this.

Right now. We're seeing real productive applications coming to market. We're seeing applications that use, uh, local inferencing on lower powered hardware, uh, that are integrated with applications.

And, and again, to your point, um, AI is not the application unless maybe you are anthropic or open ai, I guess it is the end application. But to basically everyone listening to this, AI is not the application. AI is the tool, and AI is a feature of the tool that allows you to do things that you couldn't do until we had large language models or, you know, uh, generative, uh, you know, image creation or whatever it is that you're using that tool to be.

And, and I think that's where AI gets real, and that's where AI is not a fad, because truly there are a lot of really productive applications for this that don't burn down the rainforest and require a supercomputer that costs a billion dollars. And Steven, you know, it's funny. I, I keep thinking like, where is this debate?

I think you just gave me some clarity. I mean, the fad, you know, when, when people started looking like, is it working, the fad is, and people conflate training and inference enterprises are doing way more inference than they are training in my experience. Yeah, more and more.

That's right. And they, the, the, and even sort of the, the train, the, the commodity parts of training. I mean, hell, when I can train a model like that's without a GPU or, or really it's not training, it's an improved model, right?

But with, with without a GPU, right? Like that's telling you this has gone complete commodity, but the big guys are gonna still, still do massive training. The the anthropics, the Gemini, the, the open ai, right?

The Microsoft what list them all out, right? And there's gonna be massive spend to keep up with the massive inference. So I think, I think there's something in that the Fed might be the misunderstanding of world as GPUs.

How come people aren't buying as many GPUs into certain segments that we think they are? When the truth of the matter is it's equal, right? We've got enterprises and large corporations, and even startups doing massive amount of inference if you're not in the business of providing models.

And, but the, the, the, the infrastructure somewhere has to keep up. But you're not seeing, you're not seeing modern day chat bots or, uh, audit products or, uh, email campaigns or just list on and on all the things people are doing. Interesting with Gen ai though, you're not seeing those as being massive GPU farms.

I wonder though, I think there's a, I think there's a game afoot here. And, and, and I've seen this game before, the CFO is telling the CIO to reprioritize spending to drive these AI projects, but not to fund other things. So certain other projects may starve because of ai.

And then secondarily, the CIO is sitting there going, well, what do I need to fund versus what can I push off on those business units? And, and essentially treat their budget as, as, as their budget by kind of just saying, all right, well, you know, if you want this Mr. Business unit leader, you have to fund it.

And, and that becomes this ongoing game that's been going on in IT budgets for years, and it's now just playing out in ai. I know, guy, I think you've seen this story before. Well, yeah.

So I, I was, uh, just, you know, reflecting on, um, whether in the data that that Forrester and these, the other firms are starting to use to, uh, forecast for next year. I mean, they're using survey, survey data there. They're speaking to and surveying, uh, these budget holders and IT leaders and so forth.

And it in implicit in this conversation is whether or not those folks are, um, sort of have matured in their thinking and in their understanding of where these I AI budgets and other budgets would come from and what they would be used for. They may still be in this sort of hypey mode, many of them, of, we need to grab these resources now before, like the GPUs in particular, before they are no longer available. And to your point, Mike, they may also be saying, my budget may not be changing, but the overall spend may be changing because of the way we are pushing them out into, um, uh, other business units.

The overall supervision of all this, I mean, this is where finops came from, more or less, was started with cloud and sort of expanded out to try and get a hold of, and an understanding of where all the investment is, no matter where the investment is coming from, how it's being used and managing all of it. The striking thing about the Forrester report is, unlike a lot of these reports, it really doesn't talk about trade-offs so much. It says budgets are gonna be flat on a real basis, you know, rise at the same space, pace of inflation, and everybody's gonna have to invest in talent modernization and ai, the only thing they don't mention is increasing investment in security.

Otherwise, it's like so many of these reports of the past, everything has to go up, and yet somehow it's gonna wind up stable. I, I don't know how else you can explain it, Mike, other than you know what you're talking about, a kinda shell gate approach. And there's another lesson learned here that we need to get in, right?

We, everybody thought when everybody went to the cloud, we, we just reduced our technical debt, right? And it, it was equal to the on-prem technical debt. Mm-Hmm, that's right.

And so it's the same thing is happening in, in, you know, I gave a presentation and I showed Bedrock nothing bad or nothing good or positive against Bedrock. I happen to like it, but I showed a picture of the infrastructure there. There's a lot of moving parts, and not all of them are ai.

I mean, so e so again, just because you're moving your AI to the cloud, right? And, and like, okay, now training and all the sort of GPU is all their problem. You still have a large technical debt.

And so I think we keep falling into these same traps. Oh, when we go to cloud, we're not gonna need operations or infrastructure people anymore. And then we find out, oh, we did.

It's actually equally as expensive from a technical debt perspective. Um, now it's the same thing. If you look at the infrastructures that people are building in the cloud, these are complex componentries and infrastructure that just don't go away because they're in somebody else's infrastructure.

All right, folks, well, I'm sure we're gonna be talking about this topic again, especially with our folks over at Tuum. There will be doing their research for 2025. But, um, right now it's a scramble out there.

As usual. It's that time of year. And if you want funding for something in 2025, now's the time to speak up.

We'll be back in a minute. All right, folks, we're back and we're gonna continue this conversation about AI a little bit because, well, John Willis has been out in the field talking to some folks about what the actual level of adoption is. Uh, 'cause right now there's confusion.

Wall Street seems to be banging up on companies who are not generating enough AI revenue. And yet, when you go talk to folks out there, it seems like that's all they're walk working on. John, walk us through what you thought.

Yeah, I'm gonna give you a little history lesson here. Around 2014, there was this interesting debate about whether the enterprise will would use DevOps and not so much Wall Street, because there wasn't really a Wall Street topic yet, but certainly the large, uh, providers, the big three, big five, whatever they were at that time, were all saying, this will never work in the enterprise. It can't happen.

New kids are cute, but, you know, but, you know, patent uss on the head, you know, it will only work in Silicon Valley. And Gene came, ran a conference called the DevOps Enterprise Summit. And I was on the selection committee.

And you know, I, and, and I, you know, some of my friends would argue online with all these people that work for large, you know, the large church Protestant, and I saw Target, I saw Nordstrom, Disney, I saw, uh, Homeland Security, and we gave that conference and we put that to bed. I mean, we had Target, we had, uh, ING, the list is Disney. We had, um, uk, one of the UK sort of government agencies, which a thousand year old agency was doing DevOps, right?

Um, gene Widow, gene k widow. Well, you know, I've been tracking this sort of Wall Street and, and, and some analysts. Um, we, we'll be saying that like the enterprise's failing.

It's the $600 billion mistake. It's the, you know, all this stuff. And I've been in the field, I spent a lot of time with large clients, and I'm, that's not what I've been seeing.

1, um, is that a lot of these companies are not telling Wall Street what they're doing until they have it. It's a competitive event. If you're in workspace automation and your number one competitor's Workday, you don't want to tell them what you're up to.

So there's not a lot. In fact, I've been given like special NDAs not to talk about their AI work, right? Anyway, so, um, Jean Kim renamed this conference to the Enterprise Technology Leadership, uh, summit, which is the old DevOps Enterprise Summit.

And I told Jean, I, I thought that we, this, it was in Vegas, um, last week. And on Wednesday, um, was the AI day. And I thought, you know, for those of you know, the DevOps history, right?

It was the 10 deploys a day at Flickr, which sort of changed everything, right? Like that. Like we can like, do stuff.

And, and then we had our conference where we showed that Target ING banks healthcare board, uh, what a protection could do. Dev, uh, we're doing DevOps. Well, this is what Wednesday was Wednesday, um, you know, first Source Company up is, uh, Vanguard.

Vanguard just starts showing unbelievable. They, they built something called Audit GPT, they're turning over their auditors. I mean, it, it, this is real stuff because I'll tell you what, if you've ever been to Gene's conference, these are real people.

Gene says there's no Velvet Road. People are real honest and like what you see on stage of real people working. And so Vanguard goes through their whole thing about productivity.

But the one that blew me away was Adidas. Fernando Cargo has been, if you follow him, he's been doing digital, he's VP of digital transformation at, at Adidas, and he's been doing it five, six years, revolutionized him. He gets up on stage and he's a no crap guy.

And he starts talking about 81% less time searching for, for technology. He talks about 65% complete repetitive tests just across the board, 79 to 80% more pull requests, just numbers. These are real numbers that are happening now.

You're not seeing these on Wall Street and in Cisco, John Rouser gets up and he makes a statement that a 1% increase of productivity in a company like Cisco is huge. So when Wall Street's looking for a NVIDIA's, like a run rate and calculating it and seeing what the stock price and then panicking when it dips, right? What they're not seeing is these companies are talking across the board.

Adidas, John Deere, you know, John Deere Tractor Company is telling me how they're, and, and it's, it's the same thing we heard back in 2014. It wasn't like, Hey, this is glorious. Everything's a win.

There's no problem. This is beautiful. It's like, you know, one step forward, two steps back, one step forward, three step forward.

You know, it's, there are lots of hard problems, but the thing that you have to take away, whether it's 80% or 1%, these are in massively scalable companies. Um, Adobe, Adobe has gone, has been in this for a couple years now. They've got, they've got, they've built a, um, a structure of data classification so that you were seeing a maturity, right?

The same thing we saw in cloud. I remember when I'd go into a company and ask you cloud, the first question I asked them, do you have a data classification strategy? And if they, if they looked at me like a Deere in the head mark, I knew they had no clue what they were doing was, that's the first thing you had to do if you were gonna implement cloud in a place like in Nike or where, you know, where like the data is incredibly important, and Michael Jordan's contract is really important, right?

Um, and, and, and Adobe's got a very mature, um, data classification. They, they've got a, a, a matrix to include, uh, privacy, legal security. So, um, again, you know, um, of course Google, um, you know, talking about their technology advances, um, just across the board, the company is like, you know, IBM doing full, again, I IBM's IBM, but they're doing full company Geniah hackathons.

Adobe's doing a hackathon once a week. Should it, like this idea that it isn't happening, it might might not be happening the way Wall Street wants it to happen or views it to happen, but it's nonsense to say, you know, you know, my footprint is huge. Like most of you guys, you guys spend much more time with large, larger, but my footprint is pretty strong.

'cause when I'm, I sit and have conversations with real people that are usually three from the CIO or two from the CIO that tell me the real things they're doing. And so, you know, I, I kind of want to, like the DevOps thing, whether it was a debate, whether DevOps was gonna happen or not, the enterprise, I feel like what Gene did, and when the videos come out and people can watch these, I think it's gonna be that kind of moment where it's happening now, whether it's gonna be a train wreck or it's gonna break, or we got security concerns and there's a lot, a lot of dragons out there, but to, to say that it's not happening, it's just nonsense. Well, what do you think about, guy, let me go into you because you've been tracking research for a long time, but some who told Wall Street that they were gonna be driving AI revenues in this quarter or that quarter, when it does seem to me it's a long haul and, um, we may not see it as some sort of, uh, you know, overwhelming tsunami as much as it is just a steady flood over time.

Well, I lived in New York for many years. I even worked in financial services way back, and I can't tell you who tells Wall Street all of this kind of stuff, honestly. So let me just start with that.

Um, I, I think though that, um, you know, follow the money is, is a, is, well, usually you're talking about doing investigation work and auditing and stuff, but that's really, um, how, uh, the Wall Street mentality and, uh, um, culture and ways of thinking come into being. And there is a huge push towards investment in any form of AI or AI as an industry. And so I think that's the origin of that kind of a story, really.

It's folks that, um, the street trusts, uh, either because of track record or be because of background who are not practitioners, because that's what John's talking about. I mean, how many times have we gone into these shops and looked at what they're doing and asked about what they're doing and seen a disconnect between what the market is saying is happening and marketers and vendors and especially, um, the, you know, uh, folks Hawking stocks, hawking, you know, uh, um, investment opportunities. There's a huge disconnect between that and the reality.

The threat of truth that runs through it is really the, the main difference is time window. The main difference is time window. This is, this is a real market problem in general, which is how do you get investment for something now that's really not gonna pay off for five years when you're up against, um, uh, other investment possibilities that are starting to pay off in a year?

Well, one way you do it is by ACY results in a year. That's my best analysis of where this would come from in research, you call it response bias, that people's background and their agendas affect how they report what's happening and what they're doing. And I think we just have a, a, a, a disconnect in background and agendas fi financial people trying to, um, do good work to bring in investment and support for fast development of ai because the benefits are obvious, have to, you know, kind of explain things in a certain way to other non-technical people.

And then that gives the practitioners the ability and the space to actually get stuff done. So when it comes to these questions, though, I think there's another point too. So certainly Guy, I agree with you that, uh, and, and with John, that there's a timeframe question here.

And, and in some ways, uh, the AI revolution is not really compatible with Wall Street's timeframe constraints because Wall Street really wants to see next quarter or the quarter after that as the timeframe when things pay off. Well, they won't. And there's been a, an essay that's been running around in Silicon Valley, the, uh, AI's $200 billion dollar question by David Kahn just updated, um, by Sequoia to the $600 billion question.

Because the truth is that essentially at the rate that we've purchased already, uh, GPUs and other AI training related infrastructure that needs to have that kind of revenue in order to pay it back. But where are we gonna add $600 billion to the economy directly attributable to those, uh, GPUs? Well, we aren't, that's the point.

We are just not, and, and theoretically, theoretically, um, you know, these companies could come back and say, well, hang on, we really are gonna add $600 billion to the economy once we use these models that we've created for practical purposes. And, and honestly, I think that's underestimating it to, in, in my mind, long term, the value of these models is going to be trillions of dollars. It really is gonna have a payback.

The problem is the dollars spent on that are already sunk by companies who probably won't see the, those trillion dollars coming back because the old models are already passe. Um, you've got companies like Meta out there making open source models. Uh, frankly, the companies that are making really raking in the bucks on AI right now are the ones that are using open source models, uh, to deliver value to their customers.

Uh, they're paying customers and those companies are just cleaning up. Those companies are making this money, and those companies probably will answer the $600 billion question. The problem is, it's a longer timeframe, it's a different company, and that's just not what Wall Street wants to see when it sees this kind of dollar investment going into capital equipment.

Well, Steven, you know, uh, you, you were, you've been doing, I know you've all been doing this a long time, but like these questions, I mean, I, that, that sort of 200 billion and $600 billion AI question stuff, right? Um, I mean, they're looking at the cost, the Nvidia, right? They're, they're literally doing run rate calculations against Nvidia.

I wonder if you went back, and I'm certain you could, and I'm not going to, you went back to early Day Clouds and trying to predict the infrastructure cost that Amazon was building going into this massive cloud run, right? You know, you, somebody could have done that same article like, Hey, look at all the servers that are getting bought by, you know, from whoever they were buying, you know, mostly white box, but a lot, you know, Dell and, and, and, and all the Amazon spend on, and there was, I remember a lot of these like, well, is Amazon really gonna make any money in this cloud thing? Right?

And, and, but oh my goodness, the, the, the, um, the net effect of what happened. And, and so I, I think it's the same thing. I think it's a miscalculation, a misunderstanding.

And the other thing I think is really hilarious, right? You got the Sequoia paper basically just tracking run rates of Nvidia, right? Um, and, and sort of, and there's, again, I'm not a financial person, but I look at the run rates.

So I went back and I looked at run rates from like 23, 24. They're all about the same. So I don't, I don't know exactly where they're taking their calculation from, but, but the other thing that sort of like drives me nuts is at a course, all this $600 billion mistake thing, you have this, uh, Jim Covelo at Goldman Sachs talking about a project they did that didn't pan out well, Goldman Sachs, like, that's sort of speaking outta two sides of your mouth, because I think, like Goldman Sachs probably eight, nine years ago probably had a thousand day traders, 3000 day traders, and guess what?

They used to take that down to like 50 or 30. It wasn't called JI, but it was ai, right? Like, you know, like today, you don't have, these houses have like thousands of day traders, right?

They do Al Gore trading. They do, I mean, it was all AI based stuff, right? And so for them to sort of like say that the, that, you know, well, this is a failure because we tried to do some human investment with AI and it didn't work, and we found a six to one failure.

Look at your history, your own history of the productivity gains. And, and, and by the way, you know, who's buying the most GPUs out there right now? It's the big three, and it's the big, you know, it's the Goldman Sachs, it's the JP Morgan, it's the financial institutions, right?

So, um, all Right. You know, it seems to me though that, uh, that that, um, a lot of these companies can do something like resize a workforce and claim AI victory almost in advance. We expect AI to increase our productivity by X amount.

So we actually only, we we're, we're, we're cutting our marketing staff by 600 or something like that. They're being real Cautious on that though. They were like that.

They like, they're, they, they're they lessons learned on that, right? Like, anyway, so they're being reasonably cautious about they are doing that, but they're being very, Yeah, I'm just saying like this, it, it could be a self-fulfilling prophecy in that sense. What our, our view, or at least my view is very much from the practitioner standpoint, which is to say, you know, it'd be to begin with, are we coding faster or are we coding more thanks to these kinds of accelerative tools like ai, but also that can extend out to, um, all the productivity measures we've been using since the, the birth of MIS back whenever, which is, um, how much more are we accomplishing with, you know, we are, we're, we're our staff, uh, across the entire organization, not IT staff.

You understand, our entire employee base increased by 5% over the last three years, but our productivity increased by 10%, as, you know, whatever. And then we can go and say, well, you know, some of that's because of our IT investment. And, and in this case, if you just replace it with ai, it's kind of like we can do the same thing.

We can go and say, well, our employee base was stable, or we cut 10,000 jobs and, but thanks to ai, we're making more money than ever. So it, so it becomes self-fulfilling without that kind of chain of causation that as from a practitioner point of view, we wanna see, we're just gonna claim victory and keep going. All right, folks, we gotta end this conversation here.

But, um, one thing we can say, even though I am not a Wall Street person, is pretty sure that anytime somebody is promoting some sort of stock somewhere in about six months, they're gonna short. It just seems being a natural game of thanks. We'll, benic, 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. Alright folks, and we're back with our final block of the day and we're talking about this movement to poll SA data out of these SaaS applications and put it somewhere else.

'cause it turns out well, the bad guys are targeting all the SaaS application providers. ak what we saw with CDK Global and paralyzed a large segment of the automotive industry. And now people are waking up saying, Hey, maybe I don't want my data shared in the same spot with everybody else.

Steven, are we gonna see a lot of this data migration stuff? I don't think so. Um, I think it's just too compelling.

Um, you know, we, uh, certainly we heard the same thing about, um, well, the migration from mainframe to, to open systems. We heard the same thing when it, when, when about, uh, oh, cloud, we're spending too much on cloud and, you know, are we spending less on cloud today than we were 20 years ago? abs you know, I mean, it's ridiculous.

No, no, we're gonna use SaaS. Um, I I think that there is a real story here, and it came out in some of these interviews that, uh, tech Strong has published over the last few, uh, few weeks here. Um, you know, talking about, uh, the third party risks and of, of SaaS with, uh, Brad Hier, um, securing SaaS platforms.

And then finally the, uh, discussion with, uh, haiku that Alan did. Um, in all of those cases, uh, what you heard was that SaaS is number one, extremely compelling to companies. And that number two, they really jumped in with both feet, uh, in terms of moving and adopting SaaS applications during the pandemic and afterward.

And number three, most companies are using way more SaaS than they, than they know. Uh, you know, specifically one of the things that, uh, my good friend Sebi said to Alan in that Haiku interview is, um, you know, he talks to A-A-C-I-O and says, you know, how many SaaS applications are you using? The guy says like, ah, four or five, whatever, in five minutes.

He's like, oh, well we're using this and we're using that. We're using this other one, we're using this other one. Pretty soon they're up to 15 or 20.

And those are just the ones that he can think of in five minutes. Um, we recently, uh, had a situation like that with, with our company where, um, you know, the company's trying to get an understanding, uh, not to reduce data usage, but just to sort of, um, standardize and streamline so that we only have one Zoom account instead of 50. And, um, and, and they went out and asked people, um, just within the RUM group, uh, what are you using for sas?

Um, the answer was over 50 different SAEs. Um, and I'm not shocked at all by that because at least 30 of those were mine. And so, um, you know, that's just the world that we live in.

I don't think we're gonna be able to re repatriate. I think we gotta take a realistic view of it. Uh, we gotta look at things that can do data protection, um, of SaaS applications.

We've gotta get real, which is again, one of the points that came up in, uh, in these Textron videos as well about securing, uh, SaaS data. We gotta get real about keeping pass keys and two-factor authentication and not sharing passwords and, and basically approaching this as another enterprise tool instead of saying, well, the SaaS company's gonna take care of it 'cause they ain't gonna take care of it. And we have a great deal of risk if we're, uh, basically opening all these things to anyone to access.

Well, the subtle thing that's going on there in some of these cases is people aren't throwing out the baby with the bath water. They're saying, I'm gonna keep the SAS application, but I just wanna store the data on my own S3 bucket somewhere and I don't want to use that shared resource. That's the target for the bad guys.

I don't know how heavy a lift that is because, uh, and I'll ask John this, but you know, early in my career if you were moving data and nothing good happened, right? It was you were spending money, you created security issues, and you probably were introducing errors. So, um, is data migration easier these days?

You asking me that? Yeah, no, yeah, I mean, that is always the, you know, the under, I mean, it is the debate of what I, whether I do on-prem or SaaS or do I take the risk, the risk mitigation, all that. So yeah, now our data, you know, uh, you know, where your data's located has been, you know, probably the biggest problem or discussion point.

The only other thing I want to talk about, and, you know, not that every conversation has to go back to CrowdStrike, but I, I think we're, you know, I love when you're talking about Steven, about like how CIOs, you ask him, you know, how many Saturdays they say four, and then like next thing you know, what about that one? Oh yeah, that one too. That one too.

Yeah. Like, they don't have no clue, right? And this goes back to the shadow it, but the biggest thing is there's too big to fail.

Problem is something somebody in the industry needs to figure out, right? Like, what are these SaaS components or infrastructures that when they fail, they, they fail, you know, like across the board, right? And, and I don't think anybody's really taking a hard look at what are the SAEs that are so pervasive in organizations and how well are they managed?

And you know, and again, not that we need another governing board, but, but I mean, you know, we saw, you know, what we saw at CrowdStrike was a great example, you know, of these, these technologies like everybody, like can we at least identify the ones that everyone has and, and put a better, um, you know, sort of microscope on those, those companies? Guy, what do you think? I mean, is it feasible?

Will somebody show up from the audit side and say, regardless of how hard it is, we need to make sure that our data is, uh, available? Well, I think there's three issues. Uh, the, the availability or let's say performance side of it, because we're, you always want the data next to the, next to the application if, if, if you possibly can.

And so you've already created a networking connectivity bandwidth problem by how's it somewhere else, wherever it's housed under your desk or in a data center or in S3, like you say. So that's one issue. Another issue is continuity, um, or, you know, disaster recovery, backup, all that sort of thing that's inherent in SaaS.

Now. You gotta manage it yourself or someone else's managing it for you. That's creating an operational problem.

And then, uh, the, the third issue, uh, is, uh, security, uh, you either handling security or, or you're trusting someone else to handle security and access. If all you're doing, for example, is tiering it so that you can go and hug your data when you need to, but it's pro it's being, you know, duplicated, its proximal, it's next to the application and all that stuff. Is it being protected by the SaaS vendor?

And this is where my ignorance comes in. I would ask that the, the, you know, uh, Steven and John or you, um, to what degree are their standards, um, that uh, the SaaS companies can adopt and attest to and be compliant with such that you can say to yourself, well, maybe it's housed over here, but, um, there are, you know, uh, legal attestations to get back to your original, you know, framing of this. Uh, Mike, there are legal ways in which we're protected anyway because, uh, they've, the, the vendor, the SaaS vendor has adhered to a common standard.

I don't know that such standards exist. Maybe there's a need for them. Yeah.

Mike, it's interesting 'cause you talked about that in a couple of these videos, specifically the question of standards and also you brought up, uh, one of the things that that, that you didn't really dive into and, and is, um, the effects of the Chevron decision. You brought that up, Mike. I'm curious what you think of those two questions from guy.

Well, when I look at it, I would say that, uh, a there's too much data in all these SaaS apps everywhere, so I need to consolidate that at some level. A lot of these SaaS apps have now grown in capabilities. So things that I bought for a different function are now a features in a different SaaS app.

So I think there's an opportunity to consolidate all that stuff. And then it's a question of trust in my mind around the provider. So I might have more confidence in say, one SaaS application provider than I do in another because they have the resources to invest the level of security that I need.

If I don't have confidence in that, then maybe I will migrate that data to somewhere else, or at least have a backup copy of it somewhere that I can get to in case they do get attacked. 'cause if they're encrypted, I don't want be completely out of business. And I don't think it's a, a good excuse to say as a CIO, oh, I'm sorry we couldn't process this application anymore 'cause our SaaS application provider was hit by ransomware.

You're supposed to know that. You're supposed to be aware of that and make provisions for that, and that's part of the job. And when you signed on to take it, that was the responsibility that went with it.

So I think, uh, everybody needs to kind of come back and say, what is our methodology gonna be to ensure that we're still able to process transactions? And I feel like post covid, everybody kind of went out and bought these SaaS apps. A lot of business unit leaders did it, as we talked about before.

And nobody seems to have a handle on how we're managing all that data and bad things are gonna happen. 'cause as noted in the article, cyber criminals now look at those SAS apps as a giant honeypot. They're like, these things are great.

I mean, you know, if you look at a bank, right? Like, like how they deliver software is highly regulated, right? I mean, the OCI, you got a bit there, you get shut down.

Like, you know, it's one of the few industries where you literally, no matter what, ha you know, like if you, if somehow you are irresponsible with somebody's system or record data in a bank, you're no longer a bank, right? Um, these SAEs, like I go back to the two big to fail. There's nobody putting, you know, like you look at healthcare, you look at insurance, they all have regulatory control on how they deliver software.

I've been in a lot of these SAEs, I mean, most of 'em don't even have, not that CICD is the answer to everything, but they don't even have, they don't have minimal maturity in terms of how they deliver software. So when we see these massive breaches with these companies that they're on nobody's radar. Again, if you, you a certain asset holding in a bank, you are gonna be controlled, you know, like the audit, the OCC, uh, FDIC, I mean the, the regulatory controls at you.

But there are these companies now that literally can take out the whole country and there's nobody that even looks at 'em from a, like a, a governing body. Like how are they delivering software? You know, the things that would be like you go to jail if you're a CIO in a bank for some of the things that like the, the, the practitioners or the owners of these SaaS companies are doing.

Um, and they are too big to fail. They they do. Again, I hate that everything has to go back to a recent CrowdStrike, but, but I mean, that was a massive infrastructure failure.

So maybe guy, does the government need to come around and look at various SaaS application providers and deem them critical infrastructure and apply a different set of rules and methodology to that? Oh heck no. Please, no, please, no.

Uh, uh, I I, it's an interesting question Puts PCI in mind to me. 'cause what's, what's hipaa? HIPAA is PCI tailored for, you know, healthcare?

Is there something that already exists? Um, that, you know, so I, I guess what your question is, Mike, is like, you know, it did, can the market correct itself with, uh, some kind of movement so that enterprises know that, um, they're not liable or less liable, um, if, uh, they're using compliant, whether it's PCI or some other kind of compliant, um, uh, uh, you know, badged at attest SaaS. Um, and I think where, where the, your question we could take it seriously.

Does, does, is there government intervention required is, oh my God, like, doesn't just start with how many SaaS applications and the fact that a lot of them aren't even known, but how they interact with each other. They're not all used on their own. Many of them are private networks.

They're connected with, with Zapier or other, you know, similar integration tools. The data can travel. It it, that's why, that's why the reason for my strong reaction, it's that sounds like an impossible task and just let's stick on the page.

I don't know, gentlemen, we're running outta time here. But I would say that, uh, the threat of that would be good enough to maybe get some people to behave better. And the fact that somebody's gonna come and kind of take some sort of analysis of that and apply a different set of rules seems to work in other industries.

And we have regulated industries. And to John's point in banking, they behave better because there are rules and regs. So, Well, the question, are you Critical?

It needs to be a real, a regulated function. Maybe it's a regulated function. Sorry.

No, the question is, are you critical infrastructure? And that's the thing I think nobody looks at, young people, look at what is supply they look at are all these other industries. They're not looking at what is the new age of critical infrastructure.

And I, I think it's just not being addressed. And, and I'm not a big fan of regulatory control, but I think it's, when it's critical infrastructure for the common good, it has to be All right, I'm in Las Vegas. I got 20 bucks that says we'll be talking about this conversation again real soon.

All right folks. Hey, thanks everybody for participating and I want to thank you all for watching the latest episode. And please stay tuned.

We have some awesome content on Textron tv followed up Until then, we'll see you tomorrow.