2025 – Industry Investments and Unicorns! Data, AI, Security Are Winners – EP65

All right. You are watching, listening to, and enjoying the Move Sounds of episode 65 of Infrastructure Matters. It's just me and Kimberly.

This week, Diane is busy interviewing CEOs. He's released this huge CIO study on spanning and trends. com to find out more.

Kimberly, how's it going? I, I've had to straighten my antlers. They're kinda like bent all over the place here.

They're kind of So sorry. You, you can't disrespect the antlers. You, you, you can't have bad, you can't have bad antler head.

No. And on the way in this morning, so I, I, it's usually, usually what we're recording, this is eight o'clock and I come in about 7:00 AM but coming, coming down the, the main Broadway road in, um, in Boulder, um, the, uh, one of the deer families was tripping across the four lane road there. And, uh, so gonna make sure that they don't get is like screeching the brakes on as a little fa They're not, the fawns aren't little anymore.

They're pretty, they're pretty stout by this time of year. So, um, anyway, so I feel like I'm kind of channeling my dear friends that live in our backyard. Yeah, yeah.

You know what, we, uh, we didn't have it on the agenda, but you mentioning being in the office in Boulder, we got a big air conditioning upgrade, uh, what was it yesterday? Uh, it's been all weeks. So, yeah, they had to rerun a bunch of electricity as well.

Um, and, uh, they've got, you know, quite a bit of money that got dropped into some new air conditioning because there's certain companies that, we have certain pieces of equipment that are blowing up the data center and just the, um, air conditioning has not been able to keep up with it. And so, um, maybe someday we'll be a water cooled engine here. Um, because it's, uh, the good news is that the guys here are super busy at the lab.

The bad news is that they're super busy and they're working all over throughout Christmas. So, um, they're not happy about that right now. Yeah.

Cooling is one of those data center. Coolings is one of those thankless jobs. Like no one notices it.

The, the equipment stays cool and stays up and no one really notices the, the, those benefits other than it's more reliable and more performant. Well, when we had, we had a, fortunately we're not a high, high availability center and that kind of stuff, but we had some spikes that caused things to bet spikes. I mean, we're like pushing up to a hundred here.

And so we, I mean, that, that's, that I think gives an idea about the panic that's probably in some of the other data centers that are going on. And we didn't even have have the GPUs here yet. That's going into another data center that's in another location.

So that's a super excited. I can't talk about it yet. I don't think we can talk about it yet, but No, we get super excited about that.

No, yeah. But lots of, lots of big, big stuff. Big big lab.

Everything else, they, the boys are gonna have many, many, many, many toys. I'm sorrys Probably so many toys, The boys and their toys. So Let's start off the conversation.

Your, uh, data infrastructure industry in 2024 m and a. Let's talk about the year and kind of review what, what was the big news of the year? Well, you know, when people come and talked about data and data storage, they are often thinking that it's kind of one of the boring things of the world.

Um, especially when you, I've hired millennials over the years and have to tell them there's quite a bit of exciting things going on here. Some people love this world, you know, they, and so, so I was just reflecting on, especially as, um, Oxia, just IPO this last week. Um, 18 billion I think is what, you know, they are.

8 billion is kind of valuation or something along those lines that they took out. That's a hard drive in. Well, they're, there's, there's, um, solid state drives, et cetera, out of Korea.

Um, I mean big, big IPO for a company this last year, we've had a whole bunch of m and a and efforts. Um, we've had, and also investments. Veeam took a chunk of change this year, and they, from a, um, multiple, multiple investors coming in to them, and they're still privately held.

Um, and so their valuation, I think was came up to 15 billion. They're a data protection company, guys. I mean, how boring can you get around data protection?

I mean, it's just not as sexy as ai, but here they are. And when you don't have your data, you know what, it's bad. Um, the other thing that happened with the big stuff was the Veritas Cohesity relationship that caused Veritas to split into two.

So some of the stuff went over to Cohesity, and they're now this big huge data, another big huge data protection firm. And they spun off, um, another portion of it, which was, um, acter. Um, and they are, they're at 400 million A A RR run rate.

Um, so, and all their products are rule of 40 kind of environments. So they've created these three divisions that are offering governance, data protection and, and, um, high availability capability. So see what they do, they may sell off those pieces over the next year.

Um, we are seeing a wa we, Western digital WDC split up guard, we're gonna see Western Digital WDC become the hard dis guys, the hard disk were dead, but they're not. And then we have the other side, they're splitting off is the sand disk side, and that's gonna be this all solid state stuff. So we'll see some more competition with Samsung and, and those kind of folks.

5 billion, you know, uh, valuation for a global file system. I mean, that, that's, I'm, I'm amazed that they're still around and that they haven't been sub, you know, subsumed by all of the other global file systems, but Kudo. Exactly.

And then you got wca, wca, WCA as well. They raised an another 140 billion, million dollars, and they're valued at one. 6 billion.

So, I mean, it's kind of like if you, you're not valued a billion dollars. I mean, we used to say those were the unicorns, but these were all, I guess all of this is unicorns. How can they, unicorns are supposed to be something unique.

It's no longer unique to be a billion dollar valuation. So I guess the really unique one would be Databricks, right? And you were talking about what happened to Databricks.

Yeah. That, that, this, you talking about a popping, it would be amazing to value a company at $10 billion. They're not publicly traded.

They had a J round a, I didn't know J Rounds existed, $10 billion raise, which values them at $62 billion. And I think it shows you where, you know, AI is influencing companies like Vast and all of these companies that enable ai, the enable the, the concepts of data lakes and data warehouses and centralizing data and making data more accessible to ai. But that's a eye watering raise.

They're, they're now, they now have the, the record for the mo the largest private raise ever. That this is mind boggling. I mean, it's just, even when you talk about the open AI kind of stuff, you know, it Was, yeah.

Yeah. I think Open AI was the largest before that at six point something billion. So they beat open AI by, you know, three and a half billion, something like that.

And three, uh, three and a half billion dollars raise would be an amazing raise. So Uni, I think unicorns, I, I think we need a new name for unicorns, maybe Super unicorns and nothing. I, I have to consult my, I have to consult my five and 6-year-old nieces to find out what, what the, what's, what's more unique than a unicorn.

That's true. And then thinking about 2025, we're gonna, if the market continues as strong as it is, even though we've had a little bit of a leak, um, kind of expecting a bunch of IPOs, which I already mentioned, WDC SanDisk flipping out, um, thinking va and you mentioned Vast Data, it's good PO probability that Vasts go out. Uh, IPO, um, solid dime sometime in the next 12 months or so.

There're probably, there's another solid state drive player. Vast is a big file player for ai. And then maybe Veeam, um, maybe, maybe, maybe not.

I don't know if it would be this year, maybe 18 months. But they're gonna want look at it. You know, my big concern is as personally in my investments is thinking about, are we overheating again?

Um, and that's the AI's question, right? Are we overheating? Yeah, we, uh, we, we kind of thought the days of standalone storage companies was over with Dell, EMC, NetApp, HPE Pure, and those folks kind of consolidating those markets, buying up, uh, clever startups.

But the m and a from the big guys absorbing the Wicca and the vast datas of the world kind of dried up, right? And it's now, uh, these companies are showing that they're still innovation to be had, being is a backup company, you know, and Cohesity is making ways. Rubrik already went public earlier, uh, was it this year or last year?

They went public. They, it is been, I think about a year, about two years ago. So they're still, you know, I, I famously wrote, uh, did a podcast maybe about five years ago that said enterprise storage was boring.

It's far from boring. It's, uh, very active and maybe overheated, uh, part of the, uh, investment and it landscape space. Well, this gets back to Dion's CIO insights survey.

What's number one on a survey that's on top of mind Cybersecurity number two is ai, right? And when you look at that, what's going on with the data protection is also, if we did a drill down into the other data that we have on the data protection side, and the cybersecurity side is they, are, they, the enterprises are looking very strongly at how they're protecting their data, um, because of the ransomware threats, et cetera, which is continues to ramp up. Last time last week, we talked about quantum computing and where that was going and how fast that's gonna come and come and hit us in the butt for security reasons.

So why those guys are getting the valuation is, you know, they'll link to those two trending areas that can have maintained strength. Cybersecurity's been at the top of the list now for, since 2020, since we hit the fan with, uh, COVID. Um, and it's gone bunkers, and it's continued to be bunkers.

Um, there's no panacea to, you know, it's, Yeah. And AI seems to be, uh, AI and cyber security or a link at the hip. There's, uh, I think we could do probably a whole episode on how AI and cybersecurity are, uh, interesting bad partners moving on to, you know, know more investment news.

Broadcom now a $1 trillion company. They've joined the $1 trillion club. Maybe that's the ultra unicorn riding on, uh, uh, enterprise software, uh, assets.

So Broadcom, there's a lot going on with Broadcom these days. They're talking about their AI chips and AI strategy. Uh, there's plenty going around in their network bu around their network chip sets that drive everything that's not named Cisco.

And in some, uh, uh, instances, some lower end Cisco, uh, solutions, they're everywhere. But the probably most interesting news is VMware. It's more profitable than Broadcom thought it would be at this point.

And they are, uh, approximately, the estimates are about 50% of broad of, uh, Broadcom's $21 billion in software revenue this past quarter. Uh, somewhere around, I think VMware's less revenue numbers was, uh, $13 billion for the year. So, uh, and I think that's the annual number.

So year over year, uh, VMware is up 10%, I think in profitability. It is a really interesting story. Uh, it's, it is been, I think, a wonder story for investors.

Customers, uh, you know, well, the, the jury's still out. I think a lot of enterprise customers are still kind of not, uh, let's say all in in this transaction. And what it means for them, And the conversations that I've had, um, continue to have, and I'm sure you're having, is that they're not migrating necessarily off of VMware.

They're developing off of VMware, which means you have a very long tail just like you have with the other huge investments that he's made in mainframe. Right? So, you know, I think that his, the profitability and the valuation that they have, et cetera, is, is, you know, it's, it's probably super solid.

Um, yeah, no ton commented. He's super excited about the growth of VCF Beware Cloud Foundation, which he hopes and promises will be that landing zone for net new development. And, you know, the devil's in the details.

We don't know how much of that is because the VCF licensing is kind of the way to go. Mm-hmm. For large enterprises, it gives you license portability.

There's a lot of advantages. So how much of it is buying shelfware and not actually deploying the, the advanced features of VCF and enjoying some of the licensing benefits for traditional vSphere, and how much of it is people building new infrastructure for cloud enabled apps? It's a fascinating time to be a watcher and, uh, probably uncomfortable time to be a VMware customer.

AI continues to be top of mind. Uh, I don't know if it's a surprise offering, but Nvidia announced their Jetson Mouse just like the futuristic cartoon that promised us flying cars, but yet I have AI instead. I, I guess, uh, I'll, I'll get the talking robot made at any point thanks to Elon.

Hopefully Elon, I'm more, you can on my, uh, SEN powered, uh, made, but the SEN is a series of devices. The, the one that got the most attention is a small, I, I guess the best way to compare it to is a Intel nook size device that, uh, yeah. It's, uh, cam Lee's raising her fingers representing, you know, about a, a, uh, small edge device that has quite, that packs, packs quite the bit GPU capabilities in it.

Mm-hmm. And we talked about, uh, the possibilities of, on a call yesterday, we talked about the possibilities of stuff in a 61 terabyte or 122 terabyte drive in one of these things. And what would happen as a result of what CAP and what, what could you build at the edge?

Yeah, I mean, it's pretty, it's, and what they're, what they announced was 250 bucks, right? For it's an SDK, um, software developer kit. And, uh, so I'm not sure how much that's, I'm sure it's still deployable, even if it's software developer kit.

Um, and I, I asked, uh, Ryan Shroud, who runs our Signal six five, if he was getting that for all the kids, uh, in the lab. So they could, yeah, Every, everyone should, uh, for Christmas, everyone should get a just an SDK kit. Yep.

Build something, build something fun, this, this holiday season, Or, or, and then the other thing I was thinking about is like, okay, so could you use this for, would this be something you could use for, um, Bitcoin? I mean, I'm thinking, okay, so, you know, I, if I'm a Bitcoin guy, I'm gonna look at this and say, is this, there's something here that I can, you know, that's cheap, um, that I could stack and rack and, and create whatever I'm gonna do for Bitcoin and, and, and speed that up. You know, there's all kinds of possibilities when you think about that power at the edge.

Um, one of the companies we did a briefing with this week was Quantum, um, not as in quantum computers, but Quantum as the guys that have got, they're the data side of the house. So they well known for their tape, long-term tape and data protection, but also for store next. And, um, now a product called Myriad, which is a file system.

Very, very much so. They've very much so have been focused on the media entertainment. What they announced was, um, something called client GPU Direct with Myriad, right?

So think about it. If you're doing, you know, what I was trying to understand about how that would work is if you're looking at, you know, you've got client systems, et cetera, that have GPUs in there, um, can you speed that up of whichever you're doing? So one of the, you know, since they're with, very much so with the YE and e industry, um, which you're looking at is how can I do, do and improve, um, the work that the m and e industry, the media and entertainment industry is doing in terms of their rendering capabilities?

So that's what they're really focused on. You know, they've really turned their sight onto saying, often what they would have is store next would be there beside somebody else's file system, like an ISON or Adele's Power Scale, or a NetApp on tap box. And what they're trying to do is move into that market and take that market over, um, for the prime, for the really high speed environments that they have to have.

So kind of an interesting move for them. Yeah. And I remember talking to a company that did kind of data or SQL acceleration.

They presented at a tech field day, uh, storage Field day event, this is a couple of years ago. But the idea was to take GPU accelerators and put 'em in line with data storage devices like, uh, a Dell, uh, storage array or any one of the storage array vendors, and accelerate the processing of data for, and I think this is a little bit before the AI craze, it was for data analytics, but the math remains the same. If you can accelerate the injection of data and, and creating vector databases and all that good stuff, if you can accelerate that, that better, uh, that better aligns to your AI objectives and performance needs.

So I can absolutely see this Jetson being used for stuff other than raw ai. Interesting. At the Edge, Nvidia talked about plenty of services.

They really want to push their software services, why as A SDK, they want to push nims. But I think when you take a GPU shrink it down to the size and power envelope that these Jetsons are, it makes for some really interesting, uh, accelerators that we haven't really explored these off, you know, these off server accelerators that can add value in the data pipeline. The other thing that I think is really smart about this is in the pricing piece of it.

Um, so what we've seen historically is technology companies will seed the market of students to learn their technology, and in hopes that when they move into working for full-time and primary companies, they're gonna say, we want this, we want this. So if I'm seeding the comp, and one of the strengths that NVIDIA has is all their software. So if I have my guys, the, the guys coming outta college or the tech, you know, the tech schools, whatever are trained on the tools that Nvidia has and that runs on this thing, I'm gonna say, Hey, did this, I'm gonna move more towards that space and use that more.

So, um, very, very smart on his part. I could see, um, some requirements for colleges or the tech schools to, to acquire some of these for the classes. Um, or just be in part of your computer that you have, if, especially if you're, you know, a compsci compsci major in AI or a data AI major, something along those lines.

Yeah, my youngest son just finished a program maybe about five years ago. He's five years into his career. So this would be the Raspberry Pi requirement of, uh, computer side, uh, uh, students.

The, I think it's important to highlight that this isn't a new idea. This isn't, uh, something that Nvidia just a, a space that NVIDIA has developed all alternatives, Intel a MD. Other companies have talked about putting this much compute or similar compute at the edge, you know, maybe different form factors.

Uh, the A IPC is kind of an attempt at some of this capability. It's, I think Daniel, uh, uh, looked at a post from Daniel Newman, our CEO, uh, a couple of days ago, and he, I think he called the AI PC supercycle a flop. So it did not, that was kind of provocative.

I'm pretty, pretty surprised that was not picked up by some of the financial media. But yes, this is not a new thing. So whether or not it's successful or not is an interesting question.

I talked to some of our peers in the industry as part of the Gray Beers and Storage podcast yesterday. We're talking about the opportunity for infra in the enterprise. And I don't think, I personally, we haven't done enough research on this to, to validate, but I don't think we're going to see a huge bump and acquisition of hardware for infra in the enterprise.

I think most enterprises are going to use their existing VMware BPI clusters, their Intel and a MD uh, c, fourth generation CPUs. In the case of Intel, epic CPUs and the built in AI acceleration, uh, our former peer would tell you that IBM's mainframes have really great built in acceleration, that you'll, we'll see a lot of AI on CPU and smaller accelerators, and we won't see a, a, a hardware bump. So it's going to be an interesting impact on the market.

May not be a bubble in the sense that AI is disappointing, but we won't see the demand that we're seeing on the training side. com, have talked about is the coming refresh cycle for desktop PCs. Right?

Um, you're, those, those areas. So they're, they're, they have talked about, and I don't know that market, so I'm, I'm way out on my ski Here. No, we need to have Olivier on the, on the line.

And Olivia had talked about this in one of the meetings when I was sitting with him saying, really smart guy, um, and that there's, they, they, they've, they're aging. Um, there will be a cycle of refresh, um, not driven by Windows 11. Thank God they asked me again this morning.

Do you wanna upgrade? No, No, I'm fine. Thank you.

I'm, I'm okay. My, I, I finally have my system exactly like I want it, Don't muck with it anyways, but talking about, you know, you have PCs that are starting age. 'cause there was a big cycle of PCs, I guess 20, 20, 20, 20 21.

So we are three years in, you know, into that cycle. And so they're expecting this to cycle again this next year. Um, at least that's where some of the valuations were put or have been discussed in terms of where they're going, especially within Intel and those other companies are talking about.

So maybe with those, you know, when you upgrade, you automatically upgrade to an A IPC because it's, you know, why not? Because you need it for, I don't know, copilot or something along those lines. So we'll see.

Maybe we'll get a living on here to talk about that with everybody. So I it's not Infrastructure. Yeah, it's, it is not infrastructure, it's infrastructure A adjacent, but, uh, pulling the conversation back to infrastructure and kind of a tease of our predictions, you know, we all always have to do a predictions podcast for the new year.

We're coming up on the end of a cycle for data center, though. So the, after the huge shift in work from anywhere, we completely rejiggered our data center designs to accommodate work from home. We're coming on a mass refresh of that whole investment period.

Any off the cuff insights that you think, you know, will, any markets, any vendors you think we should be watching, watching in 2025 as part of, I think this, what I'm predicting is a hyper refresh cycle for the data center. You mean for people coming back into the data center? Or it Could be for people coming back.

There's a bunch of, there's a bunch of different drivers at play. We're coming out of kind of the rush to buy all the hardware to support the pandemic to support work from home. Now we're challenged with, with both, uh, the return to office and the return to data center.

I think I, I, I think I just coined a new acronym. RTDC, the return to data center, uh, RTDC, uh, yeah, from cloud providers. And there's, you know, this whole now again, rethinking of infrastructure to support these two movements of return to our own four walls.

I suspect we're gonna, if that is true and that that's curious, um, and I know the reevaluation is going on in terms of where do I place an application, um, do I leave it in the cloud, et cetera. Um, if that is true, then I would expect an uptick with klos. Um, I would Equinix, um, should be a, a play because I may not wanna build a data center or have that kind of environment, but I may be okay with cages that would be one, one or one kind of grouping that would say I would see that would benefit from it.

Um, I know that Citrix is doing extremely well, but I don't know if Citrix is doing extremely well because of the horizon uncertainty or if it's just because, you know, that's the vdi, those are the VDI guys, but those are still operating well. But, but you, the other piece that you have to overlay with that, you know, hides that kind of, that issue is what's going on with like, what we just experienced here. You know, the, the p weak power and cooling is not keeping up because we're bringing more highly powered, highly dense environments into these environments, and they're blowing our data centers up.

Um, you know, and you, you have to move. Uh, and that more than anything else is probably what's changing the data center. Um, less so moving back, um, with those applications is what I would think.

Well, this will, this is, I'm guess, you know, I am, I am talking in a very uninformed way. So In terms, yeah, I, I'm, I'm not, I don't think I'm, uh, I don't think I have my arms around this yet as well, but there's, I think this is a good teaser for we're going to do, uh, in rap and a predictions PO podcast, and I think that's enough to wet the audience's whistle on, uh, on what's happening. So Kaly, we can't end the podcast without asking what's your, you know, big tech takeaway from, and, and I can't say for the year because the year's been too big.

What's your le takeaway for the, the past quarter of like, I think this past quarter has equaled a year. What's, what's been your big takeaway? I'd say the last six months, or less than six months, is the learning that we're having on what it's gonna take to deploy generative ai.

And it is this super successful, oh, we, God, we failed, super successful. Oh, we, God, we failed. And, and so, you know, the numbers I've seen is 20 or 30% of the projects that people have initiated have gone out.

And those reasons are to do with everything from, of course, data classification of re NetApp process processes, not maybe, you know, there and also the security issues that are there. So that's the big issue. And in that classic, you know, what, the organizations are learning massively how to do this differently.

They're not gonna stop, they're just gonna take before we go release because we don't wanna screw up the brand. But that to me is, that's the big lesson right now with that, um, that I, you know, I've seen. And the other thing is that what I, we know now about AI is going to change next quarter, and it's gonna change the quarter after that.

It's moving so fast. It is moving incredibly fast. What makes it, uh, exceptionally difficult to plan for.

If you're sitting on the ITDM side and it, the, and I've talked to software and hardware vendors, they're in this same space. It's moving faster than what their teams can keep up with. I just as super compute 24, they were talking about one megawatt racks.

So mm-hmm. Uh, get ready for that, that liquid cooling coming into Boulder. This has been, I think, a really great snapshot of what's happening in the industry over the past week.

It has been a amazing 2024. We're going to have Dion, and maybe we'll find one more person to join us, uh, to have a end of the year wrap and predictions podcast if you, if you smoke 'em, have a great holiday. If you don't, you know, enjoy the, the downtime that the next week or so will bring for, uh, me.

I'm your host Keith Townsend and my co-host Kalee Bates. Have a wonderful season. Have a merry, merry Christmas, happy holidays, happy Hanukah, happy New Year, whatever.

Whatever you, Joy as, as my friends at Blues Clues would say, I mean, not blues clues that my, as my friends at o on Disney would say, happy everything.