Exploring AI and Quantum Computing with Mark O’Regan at SUSECON 2025

This is Textron tv. Hey everyone. We're back here in Suan Gate two in sunny Orlando.

I wanted to introduce you to our next guest. His name is Mark or Regan. Mark.

Well, mark, you're gonna tell him what your role is, but Mark's with Dell. First of all, welcome to Tech Strunk tv and thanks for being on with us. Oh, Thank you Alan.

Um, I'm absolutely delight to be here. Um, it's, uh, I dunno how many times I've been at now. Um, but yeah.

I'm Marco Regan. I'm the Chief Technology Officer for Europe, middle East, and Africa. Um, uh, so really I'm looking at the forward thinking strategy of the organization.

Uh, you know, how we disseminate that into, into our ecosystem and into the ecosystem of our partners and our customers. Um, so it's a, it's a cool job, right? It Is.

That is a really cool job. And it's, it's actually a, a place, I mean, there's a lot of cool things going on. Yeah.

Within ea there's a lot of regulation and, and, and for sure I'm stupid. The EU has the will to maybe enforce some of these things that make, make sense to be enforced. Mm-hmm.

But, um, I wanna stress that you CTO for Dell Technologies. Correct. And, you know, a lot of people look, Hey, it's a Dell dude, you know, that's what they know, but they don't really know the difference between Dell, Dell Technologies, et cetera.

Why don't you explain kind of the Dell Technologies mission, if you will. Yeah, well, absolutely. So, um, so I've been at Dell for, for 26 years and as well, no, really?

Yeah, absolutely. I started, uh, some say I was found or I was left by a stork underneath a rack in our, in, in our, in our, uh, support in our support department. But look, you know, um, yeah, look, I'm, I'm, I've, I've, I've always been in emerging technology.

I, I really started out with high performance computing. That's really my first relationship as a technologist with suse, where we were bringing in SUSE line's, hardening suse Linux down onto HPC clusters, You know, All over Europe, right? And, and building some of the biggest and most essential cluster technologies.

I, I remember those industries, they were using suse. Yeah, Absolutely. So look, you know, we have a 20 year relationship with SUSE from a Dell Technologies point of view.

I think where we really sit well with SUSE is that we both are very, very committed to open architectures and open open ecosystems. So I think that's the commonality that SUSE and Dell have. Um, suse obviously very well known for open source, um, Dell, very, very well known for, for open systems, um, and a little known for o for open source as well, right?

We do A becoming more and more, I mean, look whom he was used to be part of it, it was a lot of things going on. I, I wanna come back to Dell Technologies though, right? Let's be clear.

Dell Technologies are not the people selling you your in inspire Inspiron laptop, per se, right? You are sure for sure they are that, but that's de right. So you're right.

There's the Dell laptops and there's the Dell, and you know, in inspir range, and, and then there's the Dell, uh, enterprise architecture. So technology ecosystem. So we built some of the biggest and, you know, and, and most, we believe important platforms in the world.

Everything from our friends of Microsoft, helping them bring a, you know, Azure back in the day. That's correct. Yeah.

Amazon helping them kind of mount, uh, their landing zones all over the world with, with, with AWS but also with organizations like NASA and CERN and others, where we were doing, you know, really, really important research work and, and looking at sort of helping them fulfill their mission of, um, back in the day, landing the rover module onto the surface of Mars, that was Dell working with the jet propulsion team at NASA to help make that happen. So not just, you know, our technology, Right? This, this is just, yeah, we're not getting around this, this is the enterprise mission critical stuff.

Yeah. And that, that's really what I wanted to get across. Now, you mentioned 20 year plus history of partnership with Cuse and Dell here, uh, and Cuse, you, I remember those HPC clusters, right?

We were pushing, not we, I wasn't involved that, but, you know, pushing the envelope of like building supercomputers with Linux, in essence. Yes, Absolutely. And it, it's been, you know, some of the things that were accomplished were amazing.

Uh, I, I think, I can't help but think as we sit here today though, right now, all the thing is well, on how, how many GPUs can you fit in there, right? How much heat is that gonna throw off? And then, you know, and if it's not GPUs, what do you, everyone's talking quantum may be here sooner than we thought.

Like I always thought Quantum was 2035. Now they're saying 2028, who knows? But we're, we, we're certainly at a, a juncture, if you will, where we're seeing old and new kinda jumbling up.

No, we are. Absolutely. And that's, there, there're great points that you raised there, Alan.

So I think, um, for, we're here, we're at this week, uh, we look at what we've done in the past with all of those HBC clusters, what we've done within memory architectures and distributed architectures and what we're doing and what we want to do moving forward, right? And, um, part of what I'm doing from my office is looking at building a, um, a full stack reference architecture for artificial intelligence. And so this is looking at an AI framework that we can land into a customer or partners ecosystem and where they can consume AI and build AI on their terms.

Bringing the level of sovereignty now, and I talk about sovereignty. I'm not just talking about sovereign government on purpose, right? Keeping It within the nation's Unders one, the firewalls of the companies and organizations within those nations.

So looking at their obligations and regulation and compliance obligations and making sure that they adhere to that. But also, you mentioned earlier on, right? What's happening with the EU AI act?

This is coming, this is being launched in June of this year. Um, there are some pretty aspirational, um, pretty, pretty aspirational things around what, you know, the EU AI ACT wants and needs to accomplish. And that's cascading we see all across the world today, right?

It's a good thing. Yeah. You can't, you know, there are some people who live in caves who say, oh no, we don't do business with Europe.

We don't have to do that. But no, just matter if you're on the web on the net, you're touching everywhere and anywhere. And, and so it's very hard to say, oh no, I'm not subject to those, those regulations and what have you.

And but here's the flip side, good news of it is over the past 10 years, the EU has proven, I think, very practical in adjusting the regulations to, to not make them too onerous to not to, to use common sense. Yeah. I think they're, they're fundamentally being input to safeguard European citizens, not just European states themselves, right?

So it's the citizens and not Just European citizens. I think it winds up protecting all, All citizens, all cities, to my point is that, that around that cascading piece, absolutely. So where does this land suse?

Where does this land de Well, what we're doing is we built a full stack architecture, and we're now moving towards bringing suse AI and other SUSE technologies down into that architecture. We have built agentic workflows through that architecture. So looking at getting agents to do all of that heavy lifting, looking at those guardrails, whether it's PCI, whether it's GDPR, whether it's HIPAA compliance, whatever that might be, and maintaining that level of governance compliance controls and injecting policy into the AI event at the AI event.

Now, that sounds easier to do than it actually is. It's quite hard. I Was just, I was just gonna ask you that what, Well, what we get with large language models, which we've, you know, emerged over two years ago now, is extraordinarily apparent, extraordinarily region scope.

We also get some hallucination, so misinformation and potentially as time goes on, disinformation things that banks, healthcare organizations, manufacturers, you know, fintechs, anyone in finance really they can't tolerate. So the Dell and SUSE value proposition is that we can bring that capability under the firewalls, stamp it into an ecosystem, and then start driving artificial intelligence events across an ecosystem by not just landing those foundational models where they need to be under the firewalls, but pulling information from our enterprise databases, data repositories, CRM systems, ERP systems, to complement that model and enrich that model. So fundamentally what you're going to have on a, on a Del Su stamp is a technology that is powering a model that belongs to a bank or a hospital or a manufacturing organization that is more powerful today than it would've been yesterday morning, had a, had a foundational model just been published.

Mm-hmm. If you get what I mean. Yeah, No, I get it.

The other side of it that's really exciting to me is that when you look at, you know, um, SUSE's edge strategy and how we could then look at small language models or, you know, uh, LCMS things like this, um, we can then have those small language models or those numerical models, um, uh, or even those cognitive models, lcms create a relationship with a large language model sitting at the data core. So you have small language models at the edge, giving information and intelligence to the large language model at the data core, which is, you know, fully sovereign, fully secure. And finally the tokenization, Which fire point, Which it's down underneath your firewall.

Yeah. This is where the large language model providers earn all their money. Because when you're pulling in those guardrails, when you're making sure that you're within compliance, the tokenization goes sky high.

Yeah. And the bill with it. So it's a pretty compelling No, it is.

You know, we, we do this video series not here, you back your text on we with CloudFlare called the Last Great Cloud Transformation. And we talk about something that CloudFlare calls, um, the connectivity cloud. And it's exactly this, right?

When you look at organizations today, their, their infrastructure, data, applications, everything is everywhere. They've got some in the sovereign right? Within their sovereignty.

They've got others at the hyperscaler public cloud, they've got other stuff on the edge, they've got other stuff still in closets, right? Or, or you know, stuff like that. Absolutely.

And the future is multi, the future is all of the above, but it doesn't work unless it's all connected. It's an ecosystem conversation. Yeah.

It's, and so it's not a Dell conversation. No, sir. It's not a SUSE conversation.

No one could own all that system conversation. You're absolutely right. You know, so figuring out, and it's Latency involved there.

They're in security, correct. Obviously is involved. They're in access control and, but this is modern computing and I don't know if you were at it is, what was it, height?

Uh, HPC Con or something was in Atlanta this year. Okay. Yeah.

HPC con super Sc Yeah. Right. SC 25 5.

Were you there? No, I wasn't, I couldn't make it this year. This Was the biggest one I've seen in 10, 15 years.

Yes. I mean, because we are on the cus of just crazy. I mean, it, it's like, like, well, we Really are right?

And I mean, um, and we are in the age of misinformation and disinformation. Um, unfortunately, unfortunately. Yeah.

Yeah. Um, but artificial intelligence can do extraordinarily powerful stuff if we maintain the level of control that we need to maintain over ai. Yep.

Now, you know, two years ago, a lot of talk about, you know, um, you know, the terminator, The, the hundred people signing that these will come Exactly. These will come in the shape of agents. If agents are not managed, they are not coming on, you know, with shades on motorbikes with guns, these are agents.

And what agents do is what AI does really, really well. It looks at the best possible way to achieve or execute against something. And if that means taking shortcuts or creating aphorisms during their chat to each other, that's what They'll do.

That's what they'll do. And if they do that without governance, we could end up very quickly in a situation where we have agents actually creating their own language that we don't understand. That's the danger zone.

So once we can keep outta that and make the AI contextual, I see a world, and you brought this up earlier, I see a world where today we have infrastructure in the shape of compute network and storage supporting applications and cloud native applications and even edge native applications. Mm-hmm. Going forward into the, in, into the future, we can see a fraction of that very much being the accelerator engine for this being quantum driven and the application engine or cognitive engine being AI driven.

Um, 2036. Lemme take you back 26 years though. Did you ever think we'd be talking about this here?

Um, no. Not to this either. Not to, not to this left.

I mean, well, so it's like, You know, Isaac ov kind of science fiction, doesn't it? Yeah, It does. It does.

Maybe we need some robotic laws for the agents because the next thing is what we call like this physical ai Yes. Where we're gonna embed agents into robots, robots and so forth. Or AI in the robots.

Yeah. It's a, it's a great time to be doing all this. It's, look, there's a lot of uncertainty, but there's so much potential, so much Again for suse Andel if there's a, a couple of places where suse Andel have done so much business together, it's exactly in that space where with sensors and with actuators and so on and so forth.

I think, you know, we hear a lot in the, in the media around a GI and stuff like this. We're a bit off that. I think what's exciting is, you're right, um, robots, they don't have the dexterity to be able to compete with humans.

Certainly not yet. Not for a long, long, long, long time off, but they do have the capability to take those embedded devices and become much more intelligent and much more effective and functional. Yep.

And I think that's a really, really, really exciting place to be when you're a technology company like suse or a technology company like, or technology company like Dell. Crazy stuff. I mean, I look my view on Agen ai, I was talking about this with some people the other day.

Yeah. You're not gonna have some Hobbes English butler as your agent. Right.

A lot of these agents are gonna be sort of very single use, single purpose, almost ephemeral. Correct. Right.

Yep. And they'll come and go, we may need some orchestrator for agent Yep. Like some Kubernetes kind of thing for agents.

But I do think it's coming sooner than later. But we're not quite, as they sit here today, it's not here, but it's coming. So The architecture that we have installed and what we're, we're, we're, we're pushing towards, you know, um, injecting into the, into ecosystems, into our customer's ecosystems.

It's kind of exactly as you just explained, right? It's getting the right level of compute. And this is very, very important, right?

So we're, it's searching for efficiencies and looking for those efficiencies in the new layer so we can figure out, okay, are we pushing everything to the GPU or is there some of this we can hold back and pushing, commit to the memory because It's, the GPUs are expensive. They're expensive to buy it, very expensive. It's to cool Their hearts come Electricity and they're scarce.

Yeah. They're scarce. Exactly.

So is there a way of looking at taking some of that workload tuning the technology architecture, it's what I call the downstairs in order to optimize the energy efficiency as much as we possibly can. Look, Intel was saying this right? And they got bang for it.

Ab Ab, absolutely. And this is the space that we're playing in right now with suse, with suse AI sitting on top of that where all of our engineering is looking at committing to tasking in an event as I just described. And then being able to push that out to a much cheaper, much smaller compute architecture that would sit at a, And that's the key to this thing by the, and by the way, that's the same thing we're gonna run into with Quantum, right?

What do we separate off for a quantum that needs quantum, right? Because it's a whole different level versus what do we do on traditional computing? What needs a GPU?

What can, you know, standard CPUs? And, and So before I came to speak with you here today, uh, I'm, I've just been asked for the fourth time, uh, in a row to go to Saudi to speak with one of our major customers, major HPC customers out in Saudi. And for the third year of those four years to speak exclusively around what we're doing with Quantum.

Now Quantum's an extraordinarily interesting space, particularly around the infrastructure layer. So what we're doing, what Dell have done at Quantum, with Quantum is we've looked at, um, how you interface through a classical infrastructure, you know, zeros and ones, Right? Binary By binary, how you, how you interface into qubits.

Okay. So we, so we've written a number of SDKs that we put up onto different open source repositories, really, and these SDKs comprise of tools that engineers, data engineers and data architects understand. So Jupyter Notebook, Kubernetes, as you rightly mentioned there, uh, PyTorch frameworks.

Sure. And we wrap that into a nice, uh, software easy Yeah. Deployment kit that they can now interface into the quantum system through a virtual QPU, not G-P-U-A-Q-P-U-A Quantum processing unit.

So in short, what that allows us to do is use the skills and the horsepower of HHPC in order to interface into and interpret what the qubits and the quantum system are actually telling us. So we can unravel that and decipher them. That's that's crazy too.

That's really exciting. That is exciting. Yeah.

I'm sure you probably, you saw the mi the recent Microsoft announced. I Did indeed. Yeah.

Yeah. Interesting that, that's that old Irish pro proverb, right? Mean you live an interesting time.

We were saying this about Google right? As well with Willow, I mean Yes. Willow a couple months before.

Yeah. Look, I, I really suspect we're gonna see real quantum before 2028. I think you're right.

And um, so look, we have our global CTO, John Rose who, you know, has put a lot of money and a lot of resources into the research around quantum computing around five or six years ago. Um, I'm very involved in that, obviously, and in the artificial intelligence end of things. And I remember sitting at a round table with John and we were kind of saying, how far out is this?

Two or three years ago we were saying, how far out is this? I used to think it was 20 35, 20 36 Was the figure. Was the figure.

And I have to say my, myself and two of the researchers were saying, I think 2026, here's the thing. If we look, if we look, if you look through the cracks, you'll find the light. Yeah.

And, and, and the light is telling us that China are very, very hard. They're pushed, if I may say so. Yeah.

From a research perspective, They put a lot of dollars or again, or Wong, whatever, they're, well, If SA gets cracked, we need something that can respond. That's robust. Well, So that's where NIST and some of the post quantum algorithm Correct.

Post Quantum applications. Exactly. You know, they, they're ready.

I don't wanna say they're ready, but they've made noises and they've got plans in place for that. Correct. Right.

But anyone who thinks that quantum is just about post cracking encryption No, you're missing the boat. No, No, No, no, no. This is gonna be game changing stuff.

No. Anyway, we could do on all day mark, but we got people waiting. I got a joke.

Hey man, you better, I'd love to continue this conversation. Oh, the pleasure as well. Right?

We're gonna grab your information, we'll get you remotely or something. 'cause this, I'm looking to do something around a quantum, uh, I'm just telling you like no one's watching, but we're gonna do a quantum virtual event next fall. Excellent.

I'd love to have you talk at that. That would Well, I did a fantastic me. I did, I did a keynote out in, uh, Dubai just before Christmas at the World Knowledge Summit.

Did you? By the United Nations? Um, and it, yeah, on, it's on YouTube.

You can Google. I'll check it out. Don't reach out to you for that, but we'll be in touch, Alan.

Yeah, I'd love to do that. Love to do that. Marco Reagan, uh, CT O EMEA for Dell.

We're here at Scon. We're gonna be back with more.