67. Re-Imagining the Mainframe for the AI Era at SHARE – Tech Field Day Podcast

As Tech Field Day heads to share in Cleveland, we're considering the many ways that the mainframe has been reimagined and rebuilt for the AI era. This episode of the Tech Field Day podcast features Cynthia Overby of Rocket Software and share Derek Brit and Jeffrey Powers discussing the modern mainframe with me, Steven Foskett, in anticipation of Tech Field Day at Share. Welcome to the Tech Field Day podcast, where we bring together a group of IT technical experts to discuss a single idea about key concepts in the industry.

This podcast features a variety of perspectives from members of the Tech Field Day community, and is often recorded in association with one of our events. Tech Field Day is part of the Futurum Group, and this podcast is also published on our sister company Site Techstrong tv. In this episode, as we prepare for the share conference here in beautiful Cleveland, Ohio, we're talking about the new IBM mainframe.

That's right. There's a new generation this year and it was engineered for the AI world. But before we dive in, let's meet who's on the panel today.

Hi, I'm Cynthia Overby and I am, uh, the director of Strategic Security Solutions at Rockford Software. Hi, I'm Derek Britton. I am a mainframe industry commentator and advisor.

I've been in the industry for over 30 years now, and I was lucky enough to be at the Last Tech Field Day and share, uh, meeting, uh, in Kansas City this time last year. So I'm looking forward to going to the next one. My name is Jeffrey Powers.

I run a couple different websites, uh, primarily gee zine, which is, uh, more, uh, consumer tech. But I also run a website called, uh, build Day Live, which where we dive deep into enterprise technology and building the perfect enterprise. And I am Steven FoST organizer of the Tech Field Day events.

And of course, uh, president of the Tech Field Day business unit here at the RUM Group. I am very excited to be returning to share, uh, not least of which, because it is in our home city of Cleveland, Ohio, but also because I loved Share last year. The thing that I, that I really enjoyed about it, number one, was that it was such a cool group of, a diverse group of people.

It was not just, you know, sort of the, what you might think would be at a mainframe conference. It was a very exciting group of people who are excited about technology. And number two, the thing that I love about it is seeing, it's kind of like visiting another culture, you know, where you go and you see how they do everything differently.

'cause I come from the Open Systems side. I was a Unix systems administrator, and, um, and I talk a lot about open systems, but I also run an AI podcast and a security event and networking and storage, and all of those topics are obviously incredibly relevant in the mainframe world as well. And it was very eye-opening to learn just how all of those things are being implemented and the many ways that the mainframe is way ahead of the open systems world and in, and, and in other ways where the mainframe is sort of catching up in terms of DevOps and AI and so on.

Um, Derek, I wanna start with you because I feel like you've, well, number one, you were there for the tech field day, uh, presentations last year. Uh, you were sort of one of my guides at Cher. Um, what is your interpretation of how the mainframe is advancing and, and where it fits in the modern world?

Well, that's a nice simple question to start with, Steven. So thank you for that. Um, I think, um, I've always observed it this way that, um, there's, there are basically two very well plowed parallel tracks just going off into the distance at, at, at breakneck speed.

Um, one is the mainframe world, and the other one, which is very, very adjacent to it is, um, let's call it the open world. Everything that isn't on a mainframe. And actually if you look, uh, if you look at it through a blood lens, you can't, you can't see the difference.

There's so many points of intersection, uh, uh, potential join points. And I, and so, although the mainframe space is unique, it has its own vocabulary, obviously, its own systems. They're solving very, very similar problems to how everything's happening, you know, away from the mainframe.

And in fact, I think the term, the connected mainframe, talking about the synergy between that platform and, and anything not on that platform, that was coined a long time ago. And we've been talking about the interconnectedness of enterprise systems where there's mainframe or non mainframe for some time now. And I think with every passing year you see greater convergence as those two worlds meld together.

Um, and I think none more so now with the advent of ai, which I think has as a catalyst caused the mainframe innovation to accelerate even faster. And you only have to look at the announcement that IBM has made, but also as many other vendors have made in the space, um, leveraging and, and trying to leap ahead using AI technology, um, that is fresh on the mainframe and, you know, is, is very likely to cause that platform to accelerate even faster. I, I totally agree.

What, what, with what Derek is saying specifically, um, there's, you know, that the, the interconnectivity between the mainframe and open systems has become very sophisticated. Um, and I I truly believe that, um, organizations are beginning to really clarify the best of both worlds. You know, where does, where does, where should your data live?

Um, where is it most resilient? Um, where do you get the, the, the, the best overall security in many cases that's on your mainframe. Um, where is data fluid?

Where is it easy to be able to manipulate? Um, where are the best app tools to be able to, to do what you need to do, um, with the data that's on open systems? So, uh, most organizations that I talk to today are looking at a, at a really cohesive enterprise, um, picture.

Um, and whether you call that modernization or not, you know, that's, that's up to you. But, um, but you know, the mainframe, uh, definitely, um, from a, you know, from a processing perspective is, is really, um, the place as, as Derek said, you know, the 17 and, and, and what they've done to be able to process AI transactions, um, and to do the manipulations and the evaluations that you can do with AI and get the responses, the 17, uh, was engineered to be able to do that specifically. So I'm really excited to see, you know, some of the presentations at, at Share in Cleveland, specifically around, um, around the 17 and, and, uh, the seventeen's gonna be on the floor so people can take a look at it and ask questions.

Yeah. And, and as Cynthia's saying, the the 17 that you're referring to, that would be the Z 17, which is the, the, the new generation of the IBM mainframe just introduced, uh, released, uh, in April of 2025. Um, you know, the headline from IBM is the first mainframe fully engineered for the AI age.

And Jeff, I wanna throw that to you because what do you think about IBM choosing that specifically as the headline for the next generation mainframe? You know, as opposed to all the other things, and I mean, that, that they could have chosen, they specifically said it's engineered for the AI age. Well, of course, AI being a power word, a key word that, uh, that perks people's ears up.

I could see this, uh, just basically as, you know, hey, we've got a new mainframe here, but for the most part, it's also about saying the mainframe is not something that you see in Mad Men TV episodes where, uh, tapes are flying around and, and, uh, and people are trying to use punch cards or anything like that. Uh, the bigger, bigger question though is will people start to migrate to that? Because, uh, there's a lot of people, uh, at least in the area that I am in, and, and the people that I talk to mainframe their mainframes are, are longstanding devices and they've probably, you know, they've probably switched out servers in their server room, uh, faster than they've, uh, switched out their mainframe.

And so being able to say AI in the mainframe might be that, uh, push over the edge for people to say, okay, I, it's time to actually do an upgrade on this system, and I am going to stick with mainframe because now I have all these new abilities to it. And I think what might be happening also here, Steven, is the, the reality check that, I mean, certainly some of the mainframe clients that, that, that I've been lucky enough to speak to that the primary challenge with how ai, uh, the promise of ai, how it would need to be actually implemented, is that most of the data that they'd want to train their models against is on the mainframe. You know, most of the value of, you know, what AI might promise requires them to spend some time examining mainframe data.

And so we're, we're better to get the value of artificial intelligence than, you know, as close to the data as possible. So I think to, to a certain extent, this has been, um, you know, it, it's been something that the market has been clamoring for, for some time because I think now in, in a very practical sense, the promise of AI can actually look, look to be delivered, um, where it is most necessary. And, you know, whether or not that's true, and of course, you know, that's, that's just a supposition on my part.

Um, I think only time will tell, but I think the fact that IBM has invested so much and you know, and, and the chip has been designed with it in mind, I've got a funny feeling that that will accelerate a lot of, a lot of investment. As, as Jeff said, I think it might just be the catalyst for people to really examine it seriously. Well, you, you, you take a look at data manipulation and, um, and, you know, some of the queries that people want want to, uh, to put towards their massive amounts of data, um, um, the, the being able to do that, um, in an efficient manner, um, really lends itself to the mainframe specifically when you're looking at, you know, a hundred million records, those types of things, to, to evaluate different, different queries, that type of thing.

You know, how how many servers would you have to string together for that? And, and that's exactly right. I mean, you know, it has to run, it makes sense to bring the, the AI processing inferencing to the data rather than try to exactly let the data out or, you know, especially because, you know, the thing can do it.

I mean, I, I'm not sure, um, I haven't seen the benchmarks and so on, but IBM is claiming, I mean, they, they, they actually built an AI accelerator into the previous generation as well, but this one is, is much, much faster. They're also promising to have an accelerator available later this year that would, uh, further accelerate, um, machine learning operations. I think that what they're trying to say, and, and, and it's funny 'cause I hear this as well from many of the companies in the server space, um, in the, the cloud world as well, is that, you know, you hear about these AI supercomputers for training, and that's all well and good, and obviously Nvidia rules that roost, uh, hardware wise.

But the next challenge is going to be how do we actually do something with this technology? How do we make some productive use of ai? And the answer is, it's going to be in inferencing operations, uh, maybe agentic running locally, running locally on a variety of systems.

And, and it's very obvious that that's IBM's goal with the mainframe as well, that those processes would run locally because not only have they added the hardware, but of course they've also modified, uh, the ZOS to enable the use of these hardware AI accelerators across applications. Right? Yeah.

The bigger thing is, uh, not as much of the fact that we did, 'cause the mainframe is very capable of doing a lot of things. And in fact, I remember when cameras and AI were starting to come out and they said, we're gonna have millions of points of data, but we only use like 1% of that data. And now we're starting to explore in cameras on how are the other points are going to be used.

And so when we're talking mainframe data collection happens that exact same way and bringing in AI to kind of understand all of that information. So if somebody's using their mainframe as a customer, uh, face database where, uh, where somebody comes in like a storefront or something like that, and then they're, they're trying to find the right color paint or anything like that, and AI can come in and say, oh, this is Joe. He was here last week and he was looking for something like a different type of paint and, uh, and this is what he's working on, and all that information can be stored and then brought in light when it's needed for any type of situation.

Uh, that's where I see AI starting to really help with the mainframe. The problem and, and maybe the solution with that is the problem is the coders. We're trying to find people to code mainframe, and of course, COBOL being the big language out of it.

And then maybe the solution is what I'm calling AI kitties, kinda like script kitties, uh, where you can kind of put in a chat GTP prompt and say, build me a program that will do this for the mainframe. And then starting to implement those types of scripts or at least, uh, wireframes to an actual script that will work that you can edit. And actually, it's a fair, it's great that you said that, Jeff, because even at last year's Tech field day and, and, and even the, the wider share conference, actually there were, um, notable vendor offerings, notable pieces of innovation that were being showcased.

And Cynthia, you'll know at least one of the vendors in question, um, that, that were saying we're using ai, uh, machine learning or AI based tech to help with the, um, and I've just, I've just talked about the data side, but this is the application side. Of course, the, the other thing that happens on the mainframe is the, the gazillion applications that, you know, and the hundreds of billions of lines of production code that still run the global economy that need to continue to be understood, maintained, and then upgraded to support, you know, the new digital age, whatever, whatever that dictates. And of course, most of those systems were built quite some time ago because, you know, the, the, the, the languages that they were used in the, the, you know, it's far out, far exceeded any realistic expectation of its lifespan.

So, you know, the, the knowledge of those systems is, you know, a bit dusty, no one's quite sure how they're constructed anymore. And AI is coming in to act as the, you know, the, the, the very keen, uh, intern to say, well, I'll go do some digging and, and give you some useful information. It's the, it's the buddy that you've got sat next to you that you've always wanted to show you what a program is doing.

Exactly. On the DevOps side, uh, you know, it's, there's a, the, a lot of things that can be done on the DevOps site specifically, you know, to take a look at those programs that were written 40, 50 years ago, we run into it from a, on the security side, when, you know, you're going in and they're saying, we don't have, the resources are gone, we have no idea, you know, why these particular, uh, rules of the road were put in place or where these exits were done or, or why. And, um, you can do a lot with utilizing some of the ai, uh, chat bots now to just go in and, and ask query questions about things.

And they'll, it's amazing what what they find in, in the, you know, in the bowels of the applications. Uh, that was a standout last year at Tech Field. Exactly.

That. Um, you know, the ways that AI is being used to expose how these systems work, um, code commenting, uh, code modernization, um, you know, and, and those are really good uses of, of this, uh, large language model technology. But of course, um, machine learning goes way beyond that.

I mean, security is a huge area as well that can leverage, uh, machine learning technology now, and I'm not talking about having a chat bot be your security sidekick. Uh, I'm talking about actually using it to collect more data, process more data and identify more risks. And, and Cynthia, I think that that's, uh, definitely the direction that, uh, we're looking at in the mainframe space as well as open systems.

Yeah. You know, using AI to, um, to look at patterns, um, uh, pattern usages of, of, of particular, you know, your users, um, to be able to, to effectively alert, um, there's all types of, of ways that you can use AI to help, you know, to help, um, just monitor, um, usage, um, of, um, especially now with phishing attacks being what they are. Um, and as sophisticated as they are, I was looking at, at facial, you know, facial patterns and those types of things the other day where it was like, well, this person, this is really an AI bot, this isn't a real person.

It was pretty scary to say the least. So yeah, I'm gonna bring in a very touchy subject. Uh, and that is airlines and of course, the problems that we saw at the Newark Air Airport and the air control system that is pretty antiquated in itself.

And of course, there's a lot of mainframe that's used in airports and in airline security in, in the airline, everything. So, uh, updating those are going to be key moments and bringing AI into it. They've got a, that's a fine line they have to walk because you can use AI to kind of watch the planes and see if there's a potential problem track and, you know, understand the patterns.

And then of course, you don't want people getting in so they can put, bring in their own AI or bring in their own scripts to go from there. So there's a big level of securities involved in there, and that could be used in, in a lot of different, you know, not just airlines, but train systems, but bus systems, uh, financial systems, all that. Well, you know, most people don't realize that the SRE system, uh, runs on exclusively runs on mainframes and, and, you know, anything happens to the Sabre system and there are no planes, uh, in the air.

So, um, but you're, you're, you're right on Jeff about that. Um, it's a fine line and it'll be interesting how it evolves over time, And I think that's another great example of the sort of data that lives on the mainframe side that, um, just the world wouldn't be able to get on without. Mm-hmm.

Um, you know, transportation, uh, finance, of course, I, I, I don't know what the percentage I, you, you guys probably know that, what's the percentage of financial transactions that run through the mainframe space still, um, and, um, and of course other applications in, you know, military and and science and so on. But, um, yeah, I mean, in all these cases it doesn't really, well, it wouldn't be a good idea to try to bust that data out. It would make sense to, to be doing that processing natively.

Right, Right. The, the mainframe, you know, if most people don't e realize the architecture is based upon IBM's sit statement of integrity and, and, um, there's a separation of function between applications and at the operating system layer. And, and even a lot of security professionals today don't understand that, uh, the architecture as designed originally with, um, MVT, uh, is, is very sophisticated in terms of how it keeps transactions from, from trumping on each other and keeping memory from trumping on each on, on each particular, um, function that's going on.

And, and you don't get that in distributed systems. And, um, and so it is a very sophisticated operating system in terms of security. Um, can it be breached?

No. Yeah, it can be breached. Has it been breached?

Yes, it's been breached, but, um, it's, uh, it, it is, it is much more sophisticated. And the more, and the more you can keep the data, um, from my perspective on the mainframe, um, and do the manipulation you need and just, just send the answers down, um, the better off we are. So everything we've discussed is really what's going to be discussed at Cher Cleveland, because this is of course, top of mind for everyone, um, as expert visitors, as, as folks who've been to a lot of share conferences.

I'm not sure Cynthia or Derek, which of you has been to more, um, certainly more than me. Uh, what, what is there to expect at share? Uh, well, it's a great question.

It's always a great question to anticipate what might be coming up, Steven. Um, I, I think, um, I, I don't think the book makers will be taking any bets on AI being the most, uh, topical discussion. Um, but you know, you, you were in the room with me last year at Tech Field Day, and we all then were able to get out and, and, and go around the show floor share and see some of the sessions.

We were lucky enough to see BMC, uh, broadcast and, and, uh, and my friends at Popup Mainframe, they were all, they all presented to tech, uh, field Day, but also we saw them all on the show floor as well, as well as many others. Of course, um, throughout that session, even last year, there was, there was a lot of talk about artificial intelligence and, you know, the, of course, the, the mainframe security was a big topic as well. Um, I don't see any change there.

I see that, if anything that just being layered up one notch, I think those conversations will be probably a lot more, um, targeted now, a lot more practically focused, because I think, you know, we've had 12 months of innovation, including the hardware itself, where I think now it's not a case of imagining possibilities. It's a case of actually talking, you know, practical approaches towards, you know, fixing real business challenges. That, and, and where the mainframe and the mainframe tech stack is now has evolved to support that.

So I'm pretty excited to see what, you know, what will unfold both at the Tech Field Day sessions where we, you know, we get usually to see previews as well as, um, as well as GA technology, um, but also in the wider show floor, because I think it might be one of the busiest yet. And certainly it might be one of the most interesting ones. The number of, of sponsors, um, um, that will be, uh, will be in the tech exchange, um, is is up.

It's probably going to be the, the most that we've seen in, in, um, since the inception of tech Exchange. So yeah, I'm, I'm expecting to see some, some really interesting, uh, interesting, um, demos on different products and, and, uh, there's a lot in the DevOps area. Um, so it'll be, it'll be interesting to see, uh, what, uh, what it, what it actually, what what we actually see on the floor specifically.

But, um, in terms of sessions, um, uh, just looking through the sessions, uh, a while ago, um, there's a, a lot of diverse, really diverse sessions, um, this year. Um, so I'm, I'm, I'm excited. I, I'd like to see more user sessions, to be perfectly honest.

Um, but there's a lot of partner user presentations this time that we haven't seen for a while. Um, so, um, and, uh, a lot of panel discussions also. So I know on the sec on the security side, um, there's a couple of really good panel discussions where some of the, you know, some of the top, uh, security professionals on mainframe security professionals are gonna be sitting on panel discussions, so that, that'll be exciting.

You know, Jeff, uh, it's, it's, I think what you would find is that it's pretty much what you would see topically if you went around at AWS Reinvent or CubeCon or any of these other conferences, HPE Discover, except all the hardware is different and the software is different, but everybody's working on the same things. Um, you know, does it shock you to hear that that DevOps and has come to the mainframe space? No, I don't think it, no.

It's, it had, shock is not the word surprise a little bit, but I suppose, uh, it's if if they want to be relevant, they have to have what's coming in there. So I don't know. I, since this would be my first share conference, uh, for me, I, I think I would see a lot, I would gravitate towards a lot of things that the regular people would gravitate towards.

Like, for instance, IBM coming out with the news E 17, so I'm guessing a lot of people will have a lot of interest in it and around that, how to migrate into, uh, newer systems. So you'd have, and I know, uh, Broadcom also, uh, is a big name when it comes to mainframe. So a lot of Broadcom in there, uh, being able to do the software, like with, with, uh, something like Rocket Software, um, uh, associating with the clouds, so we'll see a lot of A AWS and, and, uh, and Microsoft, uh, in there.

And then of course, companies like CDW that'll basically say, Hey, we'll just do it all for you. And, uh, and, and, uh, these, these are the things that we see that I'll see in almost any conference. And of course, they'll all have good explanations as to how to do the integration, how to do the migration.

Yeah. And, and actually I'm glad you mentioned AWS because they're one of the companies, um, that I was most surprised to see, um, at Share, but yet they have a lot to do with this. They're doing a lot with the mainframe space.

Um, Zoe, um, big presence with Zoe, um, open, open Source. Um, last year there were, uh, I maybe five or six specific sessions on, on Open source, um, different vendors giving presentations on what they can do with open source. Um, so that was, uh, that was real, very interesting for a lot of people too.

So Yeah, the Open Mainframe Project, they've done a, a, a great job of building out a range, a range of DevOps tools, including Zoe, and I think their sessions on quite a lot of that stuff. So, um, um, I'm looking forward to seeing some of those guys too. Well, I am, I'm very happy to give our audience a little bit of a sneak pre, uh, about what's gonna be, what, what's gonna be at share.

Also to share, um, if you'll pardon the pun, a little bit of the modern mainframe with them. Um, if you're listening to this and you didn't realize that, well, there was a new, uh, mainframe announced in 2025 that there was, that it's engineered for ai, that open source and DevOps and AI are everywhere in the mainframe space. Well, you know, maybe check out, share, um, you know, Cleveland's easy to get to, uh, come join us.

You'll see me there. Uh, hopefully you'll see the rest of us there as well. Um, and hopefully you'll learn, learn a thing or two about the state of the Modern Mainframe.

Thank you very much, uh, for joining us, all of you. Um, before we go, um, where can we connect with you and continue this conversation? I'm, uh, I'm very active on LinkedIn, so, um, feel free to query me on LinkedIn.

Yeah, do the same for me. Um, hit us all with the same stone. Yeah, check me out on LinkedIn.

Um, and, uh, yeah, I'm happy to, happy to take any dms. com. But we'll have, uh, LinkedIn's on both Gee Cuisine and Build Day Live.

com on our YouTube channel, as well as, uh, tech Strong tv, our sister site, where the sessions will be live streamed and, uh, posted afterward. Thank you so much for listening to this episode of the Tech Field Day podcast. If you enjoyed the discussion, please do subscribe on YouTube or in your favorite podcast application so you don't mi miss an episode.

And do consider giving us a rating in a review. We would love to hear from you. The podcast was brought to you by Tech Field Day, home of IT experts from across the enterprise, which is part of the Futureum Group.

com/podcast or visit us on Techstrong tv. Thanks for listening and we will catch you next week.