Hello, I'm Rob Hirschfeld, CEO and co founder of RackN, and your host for the cloud 23 podcast. This episode is a bit of a path, and I hope you want to wander it with us, because we followed a tangent from our discussion down the rabbit hole, and it was really rewarding. What we started talking about was refurbishing gear, so recapturing gear from the data center back and how it can be used where it's being used. What the market is interesting ramifications for that market. We really followed the path of what happens to gear in a secondary market, and I know you're going to enjoy this conversation.
We partner with eclipsium Part of our decommissioning work that we're doing, we've done, which is incredible bargain, actually, because we can automate decommissioning. It normally costs companies 1000s of dollars to decommission a server, which is more than we cost a lifetime in the server. It's insane. And one of the things that we can add isn't, is eclipsium will do a certification of the system is clean. So we decommission and clean it, and then eclipsium will produce a certification that it's been cleaned. So when you ship out the machine, you can, you can prove that you've cleaned it, you've got a certification, and it's immediately marketable, sailor, and then we'll store that back with the machine. And so you can actually go and historically verify all the it's pretty cool.
So you have nice provenance Exactly. And right. So, so here's something that you may not be aware of, that I was not overly aware of until recently. Do you know that the reefer business for hardware is now a $5 trillion a year business
hardware includes what?
Wow, servers. No. I mean, speaking,
we're just talking about certain data center servers. We're not talking about reselling servers to new users. Wow, yeah, yeah, I believe it, yeah. That doesn't surprise me at all. I mean, think about the the continued value of even, you know, older, older boxes with, with, you know, one generation or two generation out BP, use, those are still, I mean, highly functional, highly usable, highly useful in a lot of lot of situations. And it would make a lot of sense that the market is that that big, it would also mean that the kind of certification, you know, provenance, that that you're supplying there it was be like saying, Okay, I'm buying my, My refurbished MacBook from Apple, you know, as opposed to, you know, Jake's, Jake's, yeah, but yeah, corner, corner.
But what I find is remarkable is with all the ESG initiatives and all the sustainability pushing, why aren't people talking about this like, do you really need to have the latest and greatest to repatriate from the cloud and put up a data center or expand your data center?
Well, here's a question. Who's buying if you if you went back and took a look at that number, who's buying that? Who's Who's shelling out a trillion dollars, you know, for refurb, refurb data center stuff. My guess is it's not as likely that it's an enterprise with their own, you know, in house data center. It might be for placement in a managed service, you know, situation, or it could be, you know, kind of the upscale MSPs. But I'd be interested to know who's buying who's buying it and and for what purpose, and
I can tell you on the OEM side, Toshiba. Toshiba. Samsung. IBM is doing business in it. Equinox is doing business in it,
buying or selling, buying or selling, buying
up. Yeah, and then reselling. Okay?
My question is, who's the ultimate who's the ultimate
customer, buyer? What? Yeah, the ultimate buyer is mid tier companies in every sector, if it's the same guarantee and it's got this and it's certified and clean and everything else they want it.
And are they placing it in, truly in house data centers, or are they using MSPs and and, you know, kind of colos the the colos, you know?
No, they're rebuilding data. They're building data centers that they didn't have before, like you have a whole group of mid tier companies that went to SAS model software so they didn't have to buy anything, or went to cloud now, realizing that their cloud bills are out of control and that they can afford now to buy some higher end boxes and hire a very small or outsource the running of their data center, but keeping it under their own control.
This is, this is, there's an interesting market. I mean, for from my perspective, there's an interesting market opportunity, but a lot of that right now is very below the radar.
I know, unfortunately, that's what I don't understand. Why people are not talking about this. Because if you're a mid tier, and you don't have a large IT budget, and you may have one or two people in it, or maybe three or four, whatever the number is, and you see that you're spending how you know, like, anywhere between 102 180,000 a month on Cloud, because you've grown and you need all the capability, it makes sense for you to buy
upcycle the equipment. And you just listed out, you just listed off names of the, the major suppliers of the refurb equipment. And they're not going to spend a lot of money messaging, you know, come on down and and buy your, buy your refurb servers. They're going to try to sell you their brand new stuff. And, you know, it's a, it's a sales it's a sales pitch. Hey, I'll throw you, throw in these, these additional ones. Or, you know, give me a, give me a license, you know, quote, a license to sell you the the entire population of of your data center. And I'll, I can complete your requirements with you know, good as new, refurb, refurb equipment. They're, they're not going to, they're not going to push it, push the message,
yeah,
what they're doing is they're selling you. They're more advanced. I'll buy your old stuff, right? I will for a bit. I will rev share, the the use of stuff, and I'm going to put in brand new right for a discounted cost, because I can do all of the internal refurb.
That's exactly what the new parts. That's exactly what the bank refresh that we we did was right. The the refresh, which included great discounts for the buyer, the new gear, included them giving selling back the old gear as part of the discount. So Joanna, entirely, entirely, right? The thing that's not as clear to me, though, is a lot of you wouldn't put you wouldn't virtualize that old gear. So, right? It doesn't have aI processors that are worth anything in there. So it's, are you sure about that? Yeah. I mean, take, take a bunch of, you know, second gen, you know, two generation out GPU cards, yeah, well, okay, if they're recovering gear from like a NEO cloud, where the systems or or a bitcoin miner, where the systems are not worth it for Bitcoin mining, they might actually be a fine, you know, fine use for that. It's, yeah, the thing I maybe the one of the reasons we don't see it is because the people who do that are very are going to be cost conscious, and they are going to go cheap and open for that. But do everything? Do everything by hand and open source? No, this is what's crazy. I'm trying to work on our value descriptions, because we could take we could let you take second generation gear that hasn't been patched and updated, right? Get it all back in order and get it running, and you could put those discounts on the gear to even better work. Yeah, I just, I haven't seen there's stuff that we just maybe don't have the reach. For the customers we're talking to, they'll see we they're usually. I want much more. Who would end up making up making that decision? Rob, are you? Are you in front of the right the right decision maker? It's a great question. I The people who I think would do this are ones that have very, very small hierarchies, where you could be right, where the CEO is, or you know, or at least the CIO are directly hands on enough to want to know this their technology stack. Because, you know, the organizations we see typically don't want to take even though it would make a ton of sense, they don't want to take on the risk of this. You know, I might have a higher failure rate in my gear because I'm buying retired gear, or I might need to buy more more gear because the performance is less, but the cost of running that asset is, yeah, I just don't work it. I'll give you a funny story this. You might have been there. It was. It's from the glucon when they'd already changed the name to whatever X something, yeah, right, they had, they let this guy who had done this storage company, low budget, s3 competitor, oh yeah, yeah, just rebranded like they're, they've Got this red, very red, like, you know, new storage or something. But the guy, the guy, the guy got on, he had, he bought a keynote for glue con, and he was doing, he wasn't the CEO. It was his, like, you know, one of his chiefs, architect or something, the guy spent of this 30 minute keynote, he spent at least 20 minutes of it telling the story about how, in this company, they were selling storage, and they were doing it with commodity drives, and they couldn't get it was the time when there When the drives were all when there was a flood, and they nobody could get drives, especially not enterprise grade drives. And so what they did was they would go to Best Buys and buy all of the USB attached backup drives which had the same thing in it and in because they had this devoted following of people using the service. They had people buying them drives, getting reimbursed in service, free service, and shipping them the drive, their local Best Buy. And he was bragging about how everybody in the company would go like Friday night, they would get a keg, and they would basically shell these drives to plug them into the bread racks of the home built machines. Oh man, no, I, I wasn't, I wasn't there for that one. Oh, my God. And he was telling it like this, you know, this wonderful story, and it is, I mean, there's, there's a, you know, garage entrepreneurial component for it, but I'm like, and that, you know, you know, you have to be so very, you know, sort of cost conscious, You know, right? But even there, their customers are off loading, the management, the ongoing ops and management of their infrastructure to, you know, a specialty company, even, you know, even if they are bargain basement, the question becomes and you were asking before, why? Why don't we hear more about the idea of enterprises buying, buying for their own use, the refurb equipment? And the question you characterize them as as kind of, do it yourself, you know, go, go open and so forth. They're still sitting there. And what is it that they're using for their for their ops? What are they? Are they hiring a bunch of, you know, displaced you know, displaced persons from, you know, from the, from the, from the AI, you know, wave coming in and wiping them out. Or are they train, training up new people, ops, craftsmen who value the hand, the hand, hand tooling, yeah, yeah. Like, you know the guy comes in with, you know. A, you know, a hammer and only one saw and one one screwdriver. You know, it's just, it's, what are, what are they doing there? Or are they hoping that they can get budget services like the ones that you were just describing, the the s3 competitor, or, you know, somebody that will, you know, basically best a fi, your data center management, this, this is, I think, an interesting question. And you know what, where that gear is ending up.
Not only interesting, where it's ending up, it's also where it's coming from. To your point rich about Bitcoin mining, 1000s upon 1000s, upon 1000s, oh, yeah, of facilities that were designed for Bitcoin that are no longer, yeah,
they're, they're, they're, you know, they're using their servers for boat anchors, right? Too much paper weights. Oh, my god,
yeah, but, but there are, there's a whole group of companies that is trying to not only support the green initiative, but also to lower their overall IT spend that are going after some of those and try to pull the Asics out and put new machines together, or alternatively, Just use the servers as the racks, you know, as is, as is to get done, what they need to get done because, and there a lot of them are like in the Midwest area, and
I was going to ask, Are these? Are a lot of these folks net new, you know, kind of first time data center, data center
owners? Yes, okay, they've outgrown what they had. You know, Excel doesn't work for them anymore. They're actually going to real software, not that Excel is not software. But you know what I mean, enterprise quality, and they're looking at ways to do it on the cheap, because they would actually rather put their money into the people than into the
machines. And when you say
awareness, sorry, on the awareness that even if it's three generations out, it's still functional, and they're prepared to have the potential outage or downtime, so that they're buying the spare parts at the same time. Yeah.
Basically inventory, yeah. Well, the other part there is, you know, they see what their Azure bill would be. Maybe because they can, it's very unpredictable. So the other aspect of this is, yeah, I'm trading off certain aspects, and yes, I'm going to have potentially more downtime. Hopefully it's scheduled downtime, but it's downtime, right? I can afford that. And the other thing that I can I can bank on, is a lot more predictability, yes, and that is, I know what it's going to cost me pretty much every month for the next foreseeable period. That for a lot of small and medium enterprise companies, I can see as being a, you know, a real benefit. And, you know, basically making the, you know, the health span of the CFO a little bit better, because they don't have they're not sitting there, you know, being jerked about, you know, every month waiting for the bill to show up.
Well, I think that there's a bit of a pendulum swing happening, and maybe this is a harbinger of more green data centers overall, or or this notion of UPS upscaled data center, which is, I don't mind trading before it was CapEx for OpEx. Now it's the other way around, because the cost of money once is so high. Yeah, is is going to be cheaper than the cash flow situations of having to pay for OPEX over and over and over again, where you have no control, where it fluctuates so dramatically from month to month, etc, etc. So I think that we're going to see some of this, and we're. Search for the next year. More More so than
great. You know, it's worth trying to draw somebody's attention to that, you know, just to kind of look at the numbers. Where were you getting your numbers? Joanne, where were you
saying? Oh, I did a bunch of, I did a bunch of research. And I kind of, I mean, that's it goes anywhere. But depending on which research house you're using, it goes between like 3.6 on the K gar to 5.7 you know, like I went across the board, there is some big yeah, there's a big swing, because it depends on their definition of the hardware, like a server minus the boards, or everybody has a different way of doing that. And I'm not talking like Gardner, IDC, whatever. There's arc, there's abi, there's LNS, there's, I mean, you name it. And there's a lot of market swings, and the biggest the outside the US, so the biggest buyer for this stuff, you will be surprised at, not China, Africa.
It's where, India. Oh, okay, that doesn't surprise doesn't surprise me, actually, yeah.
Well, there's a rumor that that Tata is buying a lot of this stuff, packaging their software onto it, whatever they're distributing or creating, and
then selling systems, selling systems, exactly, whole systems. Yeah. All right. Now that's an interesting that's an interesting business model, because what you're doing there. So in selling whole systems like that is, you're also about to give yourself a recurring revenue stream on support, on the you know, total support. Get it all right here, yeah, and most appliances are way, way under utilized from that perspective. So if you're selling it as a bundled system, and you're just, you just hit your You just dropped your cogs on that pretty dramatically, right?
And we're looking at it from, you know, rugged devices for the shop floor, on logic.
That's a, that's a question. I mean, there that was the, one of the questions I was going to ask you, because that's net new, that's not replacement.
No, they would buy back the replacement and replace it with net new. We would buy the stuff from them.
The when you say we utilize stuff is is really as much about the motherboard or the components, it's much more about the housing perspective. So, yeah, they're taking older gear, which is time proven in a sense, then you can, you know, you could conceivably certify that you could lock down the components and certified as as ruggedized. Yeah, would you? Would you actually consider doing that at real AI,
we have been looking at it because we were approached by um on the let's call it a appliance gear company that does custom rugged boxes. They're very large. They're out of the UK, and to us, it makes perfect sense to throw the software on it, or at least the platform software, because it is a ruggedized box. You don't need the the capability of the GPUs, because we can do that on a server, and we have the capability of, you know, pushing messaging very quickly. So there
is shop this is really shop floor, ruggedized environment, yeah, yeah. Okay. It's interesting, because it starts to open up a, you know, kind of a whole new set of both companies, but also markets, business, business models, and it's in response to markets, you know, buying out, buying into, you know, kind of next generation tech, which you know, OT is way overdue for it. So, you know, this, this may end up being the, yeah, kind of the place where all of the the the analog to dark fiber following, you know, web 1.0 is bubble bursting and where that went. We may find that, you know, when this one starts to, starts to greater, this is, this is the kind of place, you know, it may be, ot where, where a lot of this ends up.
Well, that
would make, that would make sense.
I'm glad you agree it. But OT is notorious for very antiquated systems. I mean, I've been in factories even recently where win 95 Yeah, yeah, is still prevalent. NT is still prevalent. And you look at that, you go, are you out of your mind? Why wouldn't you spend a few bucks to get a cheap server?
It's, it's not, it's, it ends up not being a few bucks for these folks. That's the problem. It's not as flat. It's not a it's not a one to one swap out. You really do have to and cat. You almost have to encapsulate it. You have to containerize it. You're going to have to manage the the end of life on all of, you know, whatever, quote, whatever, passed as a server. You know, when you put the, you know, the NT box in there, and that's, you know, it's, it's not just the it's not just the hardware, it's not just the capex. And, you know, it's, I don't know, you know, the the question is, how, how desperate are they, in the sense that they are spending, you know, more than 50 or 60% of their IT budget, just keeping these things alive,
well, depending on the situation. So I can give you a more tangible example. There is a company that we know of that is using NT servers to run pieces of equipment, most of them coming from Rockwell and Siemens
the I'm sorry, what's coming from Rockwell
and the gear, the pieces of equipment that they're running, and they have weird connections, because they never anticipated this. It happens to be a large bakery chain, commercial bakery, and so their ovens are about 50 years old, so they jury rigged ports on the back of the oven through NT servers as the next layer, and that is connected to a piece of equipment that happens to come from Rockwell that is about 10 generations behind the times. And it's literally Rs 232 and RJ 45
say, big, big fat cereal. Yeah, cereal plugs that look, you know, like that. Okay, right?
And the reason, and their reason for doing it then, was because the ovens were so expensive to replace, couldn't afford to replace the oven, so they had to jury rig a capability. Now it's so bad that their connectivity to those ovens where they've tried to add sensors to them, even for something like heat, you know, to measure the fluctuations in heat between the bottom layer, the middle layer and the top layer in a big commercial oven or the stack is so bad that they're like, how do we fix this without buying a new oven? And so because of some companies that we're already working with and what they're doing in this notion of refurb and upscaling, we started looking to see how big is this market, and where are people actually spending their money. And the other reason is literally, not only because of old equipment and jury rigged stuff, but the other reason for companies wanting to do this is because the power consumption on some of the older boxes is so high that they're looking for something more energy efficient as well. And statistically, almost every manufacturer that is replacing equipment the sustainability aspect and power consumption is high, like number two on their list of what they don't what they're looking for, not the capability of the equipment. It's the power consumption. Because they're running 724, 364,
right? But that's not going to lend them towards older gear.
Well, no, no. Like, if it's so antiquated,
if they can, if they can get a replacement, if they can get a replacement that comes close, whoa. So it's, you know, instead of being six generations out, there are only two or three, right? Right, okay, yeah, that that's inconsistent with what I'm seeing, especially with the coming VMware replacement wave refresh. Yeah, people are accelerating their refresh cycles.
But this is a, you know, there's gonna keep there's a good hook for you in here. Okay? And the hook is, if you're if you don't, if you have to do the refresh, start looking at the stock and what's making up your stock at the same time
as as the person refreshing away, or the person buying the refurb gear,
which the person buying, the person who could consider buying the refurb gear? Ah, okay. And it's a great mid market play. Not that you necessarily want to go down that road, but if you do, that's one way to do it, because I think you're going to find people are going to become even more cost conscious in 2026 than they are now.
And if you start throwing some Terra for other issues on top of it, it makes a ton of sense to me. It's always made sense to me. It's it's a question of finding the decision makers and the people who are sort of, sort of in the middle of that. The thing, the thing that I'm finding, we talked about this yesterday, frustrating from the the open shift migration piece, but I put some of these back in the in the in the agenda list. So is right? The the way Red Hat's pitching OpenShift virtualization is not cheaper. So Right? There is a cheaper path that they've created. They just aren't selling it very hard. So we'll we're waiting for market feedback to enable that, and we're about to start a pretty big push explore. I literally the meeting before this, we were just talking about, you know, our push on trying to see why people aren't up taking here. There's actually a, you know, I think this is my play to your question about refurbs and is I what we're wondering, is, if we've been underplaying the consulting angle on bare metal expertise, and that, you know, there's, you know, a lot of these companies who want to do this. They don't, they don't really have the expertise. Or they, they have some expertise, but they, they need, they need some additional support on it based and so, yeah, we really have not tried to sell any or build any expertise as the bare metal experts. But the more the market walks down this path with cloud, the more we're recognizing that that we have unique skill sets. So, you know, they, they, you know, kind of forward deployed, kind of field engineering, right is, you know, call them what you want, but that that's a model that I think is going to that's a that's a business model that I think is going to become more and more prevalent. And your point about, you know, between tariffs, between energy costs, between the new, you know, the new, you know, buying net buying new equipment, as opposed to certified refurbed for what they need to do. It starts to make all of that look a lot more reasonable. Your your story about Tata, possibly, you know, buying up, buying up equipment, packaging, the whole, you know, you know, kind of the whole system, it's got one logo on it. And, by the way, our field engineers do all, all the prep, do the install, and then Ben guys are going to show up at some point, you know, just to, you know, keep you up and running, right?
You know it, like I said, the pendulum is swinging back the other way. And if what I also find very interesting, and you may find it as well. I know you guys have heard me say this before. I follow J McBain. He's channels and partnership partners and all of that on LinkedIn. I've known him for like, 20 odd years. He's also from my side of the border, and he's what I would call he is the authority on channel management and partners and whatever, whatever. And what I find very interesting, Rob, to the point that you made that you. Partner with other companies is there is a decline now in what partners are able to do. There's no more. Oh, let's get everybody on the same page, and everybody together will all go into the opportunity together. And blah, blah, blah, no, because if you saw his post from two days ago, and I suggest you follow him, because he's really quite astute about these things. What's, what's J McBain here? I'll put it in the chat.
Oh, cool. I'd like to see it. Yeah, me too.
The company is, it was called canalsis, and now it's been virtual on media, on media.
Channel insights, I see it.
Yeah, yeah.
Very interesting. Follow him or or channel insights.
Follow Him, because you'll get both, okay, but he really starts to paint the picture of not the demise, but the definite decline of the way partner channels and all the peace parts of the partner channels and all of that nonsense have started to kind of backfire in people's faces. He's not against the channel or channel management, or any of the piece parts thereof. What he was talking about yesterday was the decline in revenue that people are seeing by going at it with too many partners together. So to the idea of what Rich was just saying about Tata, this is also why I making my projection that you're going to have amongst the top 15, maybe not the top 20, global as size, buying this hardware, packaging and bundling it and shipping it out.
Ooh, s eyes. The idea of, well, the idea of s i is doing it when they've never been, they've never even been close to that business model before. I just, you know, whoa. I mean, I can see some, some of the bigger guys. I mean, a Tata, which, you know, yeah, there's nothing they don't do. So, you know, it's, it's kind of like I can buy into that interesting
top Gemini, to a certain extent, Accenture.
They're, they either have cloud or they need, everybody needs a more consistent, uh, user target experience. Is the challenge. There's just no other way to right? And if all, if the only thing to this, the point of this conversation. If the only thing you're talking is cloud as the your target for digital transformation, given what we've just described, you're you're missing out on some significant portion of the market. Now, is it a portion of the market you want, I don't know yet, it also leads, it also leads to the idea of perhaps a different, a different group of vendors, of you know, that start to look more like, less like size and more, like, you know, kind of a whole solution sellers,
yeah, and it is a gated model, but it's being refreshed. Because if, if you look at, excuse me, it's not the right forum. But if you look at the way, look at the way, Microsoft just realigned all of its channel partners, the minimum bar to be the long tail of the channel, they just got rid of about 15,000 partners.
Oh Lord, and by the way, aw enter the AWS
as well. Yeah,
AWS has been doing that kind of pretty interesting speed for a couple of years now. Yeah, so it's literally impossible to bootstrap a business, a channel business for these clouds anymore. Yeah, everybody who says we're going to get channels and they're going to sell for us. It's like, it's an incredibly I've good luck. Yeah, I've given up on that business altogether.
It's especially, I mean, I heard through the grapevine, shall we say, about a 25 million. Loss in AWS with one of its software partners that it never came to fruition. There was an another one about six months ago that was 250 million. What? I'm sorry, reliance,
where they What do you mean? What do you mean as a loss with this money?
Yeah, AWS put up the cash for the software, you know, for one of its partners to go open a new market in the channels, and it was an epic fail, and 300 people got laid off as a result. The models are broken. That's the problem they had. You know, especially when you look at the as size, like whether it's an AWS or an Accenture or whatever, they're little fiefdoms, right? Partners create these little fiefdoms. They argue over who's going to get what piece of what pie, and it becomes an absolute nightmare. If you're an us, or you're a smaller firm, and you're trying to get into one of those because they want you to have not only X amount of revenue already booked in ARR, but you have to have the sweet and the pot stuff at the same time, and the cost of managing those channels with the relationship managers and the sales force and the whatever, whatever, it's insane. So the one of the stats that Jay put out was 56% of the companies that are partnered with the top 20 as size never saw a diamond revenue in 2024
problem, yeah, that that's, I'm not surprised, but it's hard to hear that.
Wow, one of the things that's amazing to me about having this group together and letting the conversations flow, not trying to stick to an agenda necessarily, but saying, Wow, that's really interesting. Let's explore like we just did. It's really fascinating. It takes us to some interesting places, huge markets, as Joanne was talking about, but also profound implications for how the forces in the industry are being reshaped. We each have such unique perspectives, and I really love bringing together. And if you're listening to me at this point, then you must enjoy it too. Love to hear from you. If this is working for you, if this is the type of content that you're interested in what appeals to you. As always, you can get in touch with me, just email me. Rob@rackn.com love to hear what you think about the podcast, and as always, invite you to come in and join the conversation. Thanks. Thank you for listening to the cloud 2030 podcast. It is sponsored by RackN, where we are really working to build a community of people who are using and thinking about infrastructure differently, because that's what RackN does. We write software that helps put operators back in control of distributed infrastructure, really thinking about how things should be run and building software that makes that possible. If this is interesting to you, please try out the software. We would love to get your opinion and hear how you think this could transform infrastructure more broadly, or just keep enjoying the podcast and coming to the discussions and laying out your thoughts and how you see the future unfolding. It's all part of building a better infrastructure, operations community. Thank you.
