Thinking About AI – Dell, IBM and Nutanix Round Up – Infrastructure Matters, Episode 42

Hello, welcome to episode 42 of Infrastructure Matters. It's just me and ca, I say just me and Kimberly today. It's obviously the Kimberly show that I get to co-host.

Hey, Kimberly. How are you? I'm going good.

Actually, I'm doing pretty good. I got some sleep last night, so, oh, That's good. That's good.

We like that. We like sleep. Yes.

I, I, I feel your pain. So we've been on the road this week. I'm gonna go to you first.

Kimberly, you've been added Dale. We, our team was scattered to the four wins this week. I think there was Twilio, there was Dell, there was IBM I'm gonna cover the IBM piece.

There was earnings, the backend of earnings sort of, and coverage as well. So we'll get into that. But I'll give Nut Nutanix and Nutanix.

We had team out at Nutanix. I think Informatica was this week as well. I dunno whether we had the, we were joking about it with this week that there's just, this is the week.

I think there was eight events running this week. So let's, without further ado, let's go to you and Dell. Lots going on.

So Dell big show, it was all about multi-cloud, hybrid Cloud. Just kidding. So they didn't talk about AI at all?

No, no. They just skipped right over it. Didn't even talk about it.

Whatever. It was all Server feature, server and storage speeds and feeds and no ai. Well, the, the curious thing is they did kick off with talking about Power Store, but everything was really about ai.

Um, power Store had a major release. That's what, for those who don't remember, this is their follow on to Unity and Clarion and all their, their famous stuff. You know, the, the core, one of their major systems that they have.

Um, and then they released A-A-Q-L-C version. They released a lifetime, you know, trade up to the cer to the controller kind of capabilities. Um, they did a big software release on it that boosted up in terms of performance.

Um, and that's a freebie to everybody. So you download your software and off you go and you get a whole bunch more performance. So it was, um, a really good announcement for that, but the really big bulk of it, um, but congratulations to the Power Store people for getting that one out.

It's a great system. Um, is, uh, it was all the AI and this concept called the AI Factory. And I'll have a little bit more detail.

I know Bob Suiter is writing up on the AI factory and what it is and et cetera. But what, you know, I was pre announced on this. We were pre announced and pre briefed on it before we go to the show.

And I was kinda like, okay, thank you very much. It's a nice architecture. Thank you very much.

Here we go. Yeah, you're building boxes and you're integrating systems, but this is what the real important thought process was. And that is that we are gonna be standing up a large environment that will eventually be a very large environment that will be driving and creating this data that is coming out of these, these analysis systems.

And those systems are gonna be, yes, they're gonna be some of the training piece of it, but the big bulk of the business is gonna be long range, is gonna be the inference engines. Mm-Hmm. And that is where we're going.

Um, I'm gonna give you some stats here, um, that they cited is what that's coming from their research. Um, I sat down and talked to Mindy Can, who's talking about how they've been, you know, developing this research, but they see that 83% of the data is on premises. 53 is coming from the edge.

Um, what that means is that if you're gonna develop an AI system, you are gonna move the compute to where the data is. Mm-Hmm. And the data today is still on premise or most of it.

Um, and, uh, that's what they found. That's what they know they have to do. You know, they've been implementing some generative AI systems, some of it, for instance, um, customer service, they, or a website chat bot kind of thing where I'm in.

You have to move that processor to where the data is in there and they've gotta be close together because you can't, you cannot stand the latency that's gonna go over Mm-Hmm. You know, you know, big, big line on that. So with that, the other thing that they did, and this is their fu the future between 20 27, 20 30, they expect the, um, compute to exceed, um, 27 with 30 zeros following it.

If you can imagine how big that is of that flops, that, and this is going to be for the AI piece of it, and to look at the amount that it's going to be, is that this is gonna be 75% of the data center requirements is going to be for ai. And they expect that 90% of that is going to be inferencing, 10% will be training. But this is, you know, we're gonna train.

And then you gotta get to the execution of it. So they're looking at where is this going and what do you have to do in order to set up this AI factory, if you will. And when they think about factory, they're thinking about, I'm inserting data and I'm outputting the inference engine information.

And that's why they, that's why they're calling this an AI factory. And the fact that it's so much technology that has to go in there in order to drive that. And this is, you know, the entire company is behind this and wa and walk step, I mean, it was very clear you didn't have this group over here doing this, this group over here doing this.

You've got their entire, It was of all on the same page. Right. And they're all on the same page from a product management starting at product, actually starting with their own systems, internally, working through product management requirements, r and d work, et cetera.

Um, you know, it's some coming from the top. Um, but with major leadership from Jeff Boudreaux who runs the, the entire AI solutions piece of general manager. So really, really, um, significant, um, effort on their part.

I'm, I'm gonna give a shout out to one of my co um, analysts out there, gentleman by the name of William Fellows, who works for 4, 4 51 nowadays. And we were in the meeting with, um, Jeff Boudreaux. And at first I kind of giggled at it and then I said, God, he is right.

Um, but it, and he's right about, not just about Dell, but about the entire tech industry, as he said, and I make not quote exactly, he says it, it's like, AI has rescued Dell from Multicloud. And I said, oh, it, it, the entire tech industry has been, because, you know, last year we were kind of talking ai the year before it was all multi cloud. We were still talking multi-cloud.

But you know, this, this wave that's going on right now and over, you know, many, many years, if you buy off that this is gonna play out. Yes. It's a huge, massive investment, um, that all the industries are gonna be making.

Yes, there's gonna be bumps in the road, yes, there's gonna be problems, but, you know, how do you stand up not only the, the systems such as Dell is gonna deliver, but also the capabilities that IBM is gonna be delivering. You know, that which is really the software layers and all those pieces, because there's some big sticky wickets in terms of like the data pipeline problems that have to have, have to be addressed, you know, in terms of the training and the modeling and everything else. So with that, I'm gonna mention one other item and then I'll pass it over to you because I know I've been talking straight for like 10 minutes almost.

It feels like I I'm hanging on every word, so I'm sure the listeners are. Oh, good. Let's, let's go.

So the, the one of the items that they announced was with hugging face. Mm-Hmm. And this gets into the modeling piece of it.

Um, with hugging face, there is now a, um, there's hugging face Dell Enterprise hub there. What they've done is they're putting up models up there, um, that you can go download, go, go download. There's specific, some of the vertical or use case models, you can download it.

I mean, it's super easy. Um, or at least when they demoed it, it was super easy. You download that, you can select the processor, you wanna drop this thing on, and off you go.

I mean, this is the kind of things that innovative kind of things that they're looking at doing to say, you know, what can we do to streamline this? Even though primarily the company is a infrastructure matters company. Um, they're an infrastructure company.

Um, but they're looking at all the different things that they can do because they have felt the pain of trying and successfully in some bumps in the road, successfully setting up some major use cases that they know are having huge impact into the company. Things like developer velocity, which is the coding kind of capability, content, everything, content, you know, cutting down and getting accuracy up to 90% of their content is a whole lot better in terms and also with the people that are dealing with the content, customer service website, um, assistance, et cetera. So all of those things that they're, they're looking at and they're having to go through this process of saying, do we have repeatable, stable, systematic, simplistic processes that we can actually train something on?

And then, do we have the data to do that? You know, We're in the deployment phase of ai, we're past the hype, we're in the deployment phase. Yes.

And, and the pain is gonna be in, in, in not only just the pain is gonna be in setting this, setting those environments up, which they're heavily addressing how to set up all the networking, the servers, the, you know, and create some flexibility within their, and then they're also moving up a bit of the ladder to say, can we help folks on those other pieces which have to do with what's gonna happen with the data management piece of it? What's gonna happen with the model model recommendations. So really great show.

Um, I, I usually can come out, you know, poking at a few things and that kind of stuff, but this time around I was, um, this time around coming out of this particular show or whatever, um, yes, it's was extremely, you can see the energy in the company and the enthusiasm in the company. So, Well, I mean, I think there's a whole bunch of research notes from what I see in our channels that are gonna be coming out. So I'm gonna be looking for those.

It was kind of a shame that Dell and IBM run their shows at the same time. 'cause I think you would've had as much of a good time Mm-Hmm. At IBM as you obviously did at Dell and vice versa.

But, you know, pivoting to I-B-M-I-B-M had think this week, smaller show than some of these big Uber sort of 40,000 people shows intentional by IBMI think they're on a, I think two or three versions of this format of think there's not the big sort of multi-vendor expo. It's all IBM. It's all consistent.

It's all very well themed. And I think that's enabled them to tell a really good story and a really consistent story. IBM is now laser focused on two things, hybrid cloud ai, and we're starting to see that come through in all of their discussions.

And, you know, there was some really, really good presentation by Rob Thomas that talked about what they're doing in the hybrid cloud space. VIN took the, a more of the AI piece, you know, d very good sessions, um, a MA sessions with Vind and Dario Gil who leads research and, and Rob and a whole bunch of the team. But it's really clear that IBM's getting very, very focused now and when starting to see real depth come behind Watson X.

So, and that's consistently coming through the organization. I spent time with Ross Morey this week who runs the mainframe part of the business. And you can tell there's a connection point between what he's doing with research and Dario's team around an a, a Watson X assistant for the mainframe.

And that's, he's on message, he's consistent and he's connected to the broader message that Rob and Dario are running for assistance, for instance, across the whole business. So going back to what you're saying, this is where good leadership is, this is where this holistic sort of company view of we're all on the same page. You know, just to give an example of what this, um, assistance is for the mainframe, there's obviously a huge skills gap.

There's, there's challenges there. Mm-Hmm. What they've done is ingested all the red books, ingested the support database.

Yeah. So a huge corpus of data of all the support calls that have been logged, been able to pull that together, put it behind a large language model, and then be able to present that. And as an assistant, I conservatively think that it takes you five years to grow a system program for the mainframe.

And I think what they're doing is probably taken two or three years off the back end of that. So the ability to kick off tasks, be able to reorder tables, be able to actually not just query a knowledge base, but then actually do things and kick things off. Mm-Hmm.

In context of the assistant, they've launched one for bi. You know, there's a whole bunch of these assistants that are now coming from IBM all business focused, all sort of technology focused. So there's a real consistency there.

The other key takeaway from me, from think this week, as I say Rob Thomas did a really good job. I'm starting to see a pattern emerging around what they're doing in the automation space. I think IBM had some assets there.

I think obviously with the Ansible assets from Red Hat, what's going to be coming with HashiCorp. But then when you bring that through to what they've got from a finops portfolio with Instar, a turbonomic easy for me to say. And, um, the Cloudability stuff from aptio starting to see sort of those five assets come together in a really consistent automation type portfolio that's gonna be, I think, saw that vision for the first time from IBM this week.

So that's gonna be really interesting of how that pans out. Obviously couldn't say too much about Hashi, but you're starting to see that layer cake and that we're gonna see AI and automation in that sort of software portfolio. Mm-Hmm.

I think another, I mean kind of throwaway comment, but really helped me with the QRadar divestiture. Mm-Hmm. One of the analysts, I'll give another shout out.

Um, Ali Mellon from Forrester asked the question to Rob Thomas, how should we think about the IBM security portfolio, post QRadar obviously move into Palo Alto. And Rob gave me a really good answer of, we're gonna get outta some of that threat detection kind of, I don't track this base, but sort of XDR SEA type space. Mm-Hmm.

And we're gonna focus more in on data security. So he talked about Guardian and that sort of data flow, obviously foundational from an AI perspective. So IBM's gonna retain its focus there.

So that was the first time I'd kind of heard the Guardian still important to us. Data protect data and securing that data layer was foundational. So that's fascinating for me.

I mean, I think another key thing that came out of it, and I met with AWS at the event, a really strong focus on getting IBM software onto the hyperscale providers. If you wind back five or six years mm-Hmm. IBM was trying to be a hyperscaler and compete on IAS complete shift.

IBM's now partnering with those, what they see as IAS providers and saying, running Watson X on AWS So Is that mostly for development though? I mean, is it mostly that they're considering that? Are they, are they looking say, because I know there's a bunch of mainframe stuff initiatives that are going on in that place, but that seemed to be mostly a developer kind of space to play less of a running your entire operational kind of environment on the cloud.

Yeah, so I mean, from a Watson X point of view, I think it's, we've got really good models with Granite. We've really got that. Um, and 17 models with some really interesting performance data.

And I know from previous history, trying to get performance claims out of IBM and through legal is hard. So I, I kind of do believe the claim some really interesting graph granite performance claims. But to go back to your point, I think from that, what X layer?

Yeah. IBM's not really caring where you run it. So if you've got a AWS relationship for infrastructure with EC2 and you want to run it there, they don't mind.

Whereas they would've minded before. So I think, you know, interesting kind of pragmatism on the public cloud and some of those hyperscalers. So fascinating week chance to connect with a whole bunch of IBMers from my past, but really, really fascinating to see how laser focused VINs got them on hybrid cloud and ai.

Wow. So back to you for Nutanix. I know we were, Paul was out at Nutanix, Actually no, he wasn't.

He was at something else. Guy Courier was at Nutanix. Oh, Was God.

Our team was everywhere this week. So how, how did, how, what, what's your, I've not dug into the nut Nutanix stuff. How did that Go?

Well, so, so the, I haven't dug into Nutanix at all except, um, guy, guy was slacking me about something about Nutanix and DA's Powerflex and we went, huh. And, um, so I said, well, let me see if I can find out more. So he is over in Barcelona, I'm in Vegas.

So, you know, I, I, um, I, I raised, I was at the partner event. They have a, they had a great partner, they've got some great partners there, but they had a partner event and um, Denise Miller, they had a separate session after there was q and a with the press, and if it was press and analyst or just analyst, I can't remember. Anyway, so I raised my hand and I asked the question about Powerflex and Nutanix and, and how they were gonna position this with, um, their other products.

And, um, she had this look at her face at me, which was, I'm really glad you asked that question Either that, well, well, I complete surprise. So, so evidently Nutanix had pushed out the press release or announced it at their conference, but I don't think totally Dell knew because then late, late that night or some, somewhere around nine o'clock, I get the press release from Nutanix, actually from Dell that Nutton has put out to take a look at what was going on. It's like, oops, this happens.

Um, we've seen it happen before. But anyway, so it was, it was, it's a really interesting announcement. So here's what's going on.

Um, and suffice it also, I wanna say that at um, Dell, they really didn't talk about virtualization that the, you know, at all. Okay. Except on this, you know, when I, when I got into this, and this is when we kind of started talking the virtualization strategy, um, start first by Dell saying, you know, they're working on, you know, quite a bit of different areas, but what they're trying to do is make sure customers have, there's an ecosystem that they have that customers have choice.

Mm-Hmm. And they understand, they've clearly understand some of the, the agony that's going on with, um, the, the vSphere pricing or the VCF pricing, et cetera. Um, so they're working through there to say, okay, so what do the customers wanna do?

You wanna stay on VxRail, you know, because it's a well loved HCI, you wanna do A, B, C, you wanna do something else? We're, we're gonna help you, we'll help you wherever you wanna go. So the Nutanix one here is they have a, a product called A HV, which is their virtualization offering.

And, um, it's highly integrated. Um, mm-Hmm. When you, when you, you know, grow it, you really grow compute and storage.

And they're, and this actually started a conversation a year ago, so this is before all this stuff. This is not something that happened since January. And the numbers rolled out where there were customers and, and Dell had customers that had nut tonics, that had smaller pods, if you will, or clusters.

They couldn't scale, couldn't bring them out. So they started having conversation about using Powerflex, which is a, you know, distributed storage product. Um, also known originally as scale io.

Um, but anyway, so how, how do you do this with powerflex providing a block block capability with Nutanix? So this is what, what they announced is the ability to use, you know, take out the storage that Nutanix has put in Powerflex, be able to scale this thing to a much larger environment. And to give you an idea about Powerflex, Powerflex has, it's, it's a block device.

It's not, it's big. I mean the, the average systems are in the petabyte sizes. Um, you know, with 10 nodes going on that, you know, this, it's not small.

It's, you know, it's a big, it's usually a big, big environment that's going in, being deployed. So this is going and targeting to the very, their larger customers, you know, the global 2000 mm-Hmm. Area that have got those environments and they wanna bring those offerings to market.

So, okay. Nicely done. So yeah, Interesting and need to dig into that one.

Maybe I need to spend some time with Guy and understand implementations of that. As I say, we can't be in, we can't be everywhere all at once. We can't be at IBM m Nutanix and Dell all at the same time.

Yeah, yeah. Yeah. So the final one on the docket here, and I'm just catching up today, got a research note, kind of half written collaborating with Olivier Blanchard on it around Lenovo earnings.

So they announced Q4 for year and four year results. Um, I think yesterday some highlights up 10% year on year, which is impressive, I think for those guys. Double digits mostly From AI stuff, or is that Yeah.

Yeah. So I think the, and this stat stood out for me, non PC revenue mix, we get briefed on this. So obviously in Lenovo's got a huge sort of PC business, a historic high of 45%.

Um, so, so the non PC revenue closing in on being half of their business now, um, I think that's interesting for me because that says the revenue's coming from the data center and their services play, um, all of the business units showed growth. So they split their business into the infrastructure business, all their data center, the devices side, and then they've got a services play. So those are the kind of three big chunks of lenova, all of those demonstrated growth.

Obviously AI starting to come through. Did they mention any of the, I'll take a look at the transcript, but did they mention anything on the storage because, you know, they, they've done really super well on the entry level and they're trying to push up into the mid-range. Mm-Hmm.

And gain, um, market position there, um, with their, their offerings that they have. Um, most of their offerings are, you know, they, they OEM them, um, but you know, they're still growing that data center stuff. So From a data center point of view, up 15% in Q4.

Okay. I mean, as you say, the storage piece, they're coming up those price in tiers. Um, what are they number three now overall in revenue in storage?

Mm-Hmm. The continual conversation we keep having with them is how are they gonna get to those? Yep.

Maybe some of the traditional sort of power store type places you were seeing from an EMC perspective, you know, evolved. How do they get into that tier zero, tier one type storage space? You know, I think they're doing a fantastic job of executing in those kind lower price bands and the server connect type stuff.

I mean, true scale starting to come through services up across the board, you know, so double digit again, on services, true scale's a key part of that. A lot of the stuff they're doing around some of the AI sort of, um, packaging pieces like you were talking about with Dell's AI factory Mm-Hmm. That fits into the services, uh, solutions and services group.

So solid numbers across the board. I think, again, I've got my sort of quarter post earnings quarterly briefing today, so I'll get more detail on what's driving that. But I think what I think what came outta my comments from you from Dell, and what I'm gonna see here from just my first blush reading of Lenovo is we're now into the deployment phase of ai.

It's not some crazy POC where, you know, it's kind of like, let's throw some stuff up on the cloud and really sort of get, uh, trials and POC stood up and lab environments. I think we are now past that. We're at the point where we're starting to see production deployment that's starting to show through into some of the quarterly results in the strategy from what you saw from Dell.

So I think we're probably now at the point where AI is starting to manifest itself in some of the quarterly earnings. Be interesting to see whether that growth continues quarter on quarter. Yeah.

So one of the things I didn't mention, um, was Dell's a i pc MM-Hmm. And um, in addition to all the other big numbers that I threw out there, um, in terms of growth or where this market is going, they're expecting p all PCs to be revved, which makes sense between now and 2030, which is about 28 million, 28 billion, you said. How many was it?

2 billion PCs? 2 billion PCs? Um, the entire PC inventory, which would make sense, but mo many of those are gonna be, or maybe all of them will be, have some level, they already have some level of GPU in there because of the pixels and everything else, but we're looking at more powerful GPUs or NPUs in there, um, to, to roll out.

And in fact, uh, you know, we're seeing that, you know, huge impact, um, on those systems as they're coming out in the market. So very, I mean, I don't cover the intelligent devices group stuff for Lenovo. That's where I'm collaborating with Olivia.

Mm-Hmm. I think Lenovo seeing that same trend in its devices business. Okay.

And obviously they've got a sort, what they call pocket to cloud portfolio. Mm-Hmm. So we're seeing that also from their Motorola business from, you know, so all the stuff that the Qualcomms and the those guys and the Broadcoms and those guys are looking at the absolute far edge IE your phone that you gotta remember whilst in the US it's a sort of 50 50 split between sort of iPhone and Samsung, maybe more even iPhone in the us, other parts of the world.

Motorola is a massive player. Mm-Hmm. So that kind of, I think they're number three behind iPhones and Samsung.

So, you know, there're a massive player in this market, so that kind of pocket to cloud play for Lenovo, so interesting. Uh, I'm gonna be looking, I, I've literally got my Google doc open and there's a Olivier as good as do his bit on intelligent devices. So we'll see how that pans out, but, okay.

Fantastic week as always. I, how we managed to get through that in 29 minutes is kind of beyond, I don't know. I, I have pages and pages and pages of notes and I was, you know, trying to get it down to, I, you know, the, I I touched on page one, I touched on, I, I talked page one and that's it.

So you guys there so much more that's, that's gonna be rolling out here. So yeah, You, you did a good job on the summary of Dell. I'm interested to chat to you about what's going on there, but we always try and keep this to half an hour every week.

So at that point I'll wrap us up. You've been watching another episode of Infrastructure Matters. Please click and subscribe and do all those things to and share it with your friends and we'll see you next week.

Thank you very much for watching. Bye.