 Hi everybody, welcome back to our wrap-up day three here, HPE Discover 2023. This is theCUBE's awesome coverage, three days wall-to-wall coverage. We're extracting the signal from the noise. Dave Vellante here with Rob Streche. Lisa Martin was here earlier in the week, as was John Furrier, who's now in, I said DC before. He's in New York City, almost did it again. Do it covering Mongo, DB, check out all the coverage at thecube.net and siliconangle.com, but we're going to break it down right now. HPE's annual customer event discover here in Vegas, 10,000 people. Really, it was a message of unification, particularly around GreenLake, and what's powering GreenLake is that Aruba Central IP, we just talked to Bill Matrim about that, and about some of the innovations that are coming out of Aruba. It's the crown jewel right now of HPE's business, despite the awesome server business that it has, it's a really strong compute business. But the big news here was LLMs as a service, basically HPE using its supercomputer Mojo as acquisitions of Cray and SGI to really try to differentiate and move up the stack a little bit with a cloud-based offering, but strong message of edge-centric cloud-enabled, data-driven, something that Antonio Neary, CEO has been on for quite some time, and a story of differentiation, Rob, and big bets that HPE is making, and it's starting to come together, isn't it? Yeah, no, I think it is, and I think what's really good is that they're starting to also simplify their message and starting to consolidate that message within GreenLake, because even some of the discussions we had, it was GreenLake for this, GreenLake for that, and now they're going and simplifying that GreenLake message down to things like GreenLake for LLMs as a service, and bringing real service as a service there, and I think that's really key. Yeah, things that are consumable as opposed to customizable, right? I mean, it could be anything you wanted. You could do a PowerPoint and say, well, we can do that, we can do that, we have services organizations, but that's not what the cloud operating model is. The cloud operating model is I got a console, and I'm going to pay by the drink if I want to. Exactly, and I think that's been the key is that you can get that actual application, you're not just getting a blueprint to the application, you're getting on that application, you're being able to use an LLM on a supercomputer infrastructure, not just having, oh, here, you can buy this, all these parts and pieces, and we give you the instruction manual. It's moved on from there, which is great. So, Pebble's unpacked this LLMs as a service. What exactly, in your view, is it, it's not infrastructure as a service, right? It's up the stack for specific use cases and workloads that are very HPC intensive. So they're starting from a position of strength and trying to be unique and differentiable, but exactly what was your takeaway in terms of what it is? I'd call it a PAS, a platform as a service. I think that they're very strong in the knitting of infrastructure, the software they got from Cray and from the acquisitions of SGI, and I think that they're bringing that know-how of building all of these massive supercomputers that they've been building for governments and corporations for years, bringing that to bear on this LLM space, which has, I think, been really key. Yeah, Dr. Goh was sort of explaining that if you think about the typical cloud model, it's a single virtual instance that's shared amongst a lot of different workloads. If you think about the supercomputer model, it's a workload with a lot of different GPUs that have to be kind of programmed to take advantage of that parallelism. And to your point, Cray and HPE, they have developed software over the years that actually runs end to end, so you don't have to restart the jobs and consume a lot of money, energy, and time. And so that's something that HPE is putting forth as an advantage relative to the way the cloud works. Now, we're going to talk tomorrow on our breaking analysis with Andy Turai as to, well, what can the cloud guys do to respond? How will they respond? Is this really a unique advantage for HPE? And what's the preview? What do you think? Yeah, I think it's at least a advantage in the near term because I think the checkpointing and the actual software that they have in there for managing these workloads at scale is definitely differentiated. If you think about it, it's come out of years of experiencing the grids, the large grids that were all the rage pre-cloud and pre-open stack. And I think that will give them an advantage, definitely in the near term. I think how sustainable that is also comes down to the infrastructure that they build out, the Cray infrastructure. As everybody tried to hit us over the head and I felt like at the end of the day, I almost wanted to become a plumber with the water cooling systems that are in the Cray. I think that you really need to look at this and say, okay, that could be differentiated, how long it may take somebody to get there. Yeah, and sustainability was a huge theme at this show. I mean, I've said it before, every large company has a sustainability story, but what I learned this week and Antonio doubled down on it, John Fry, is HPE sustainability is, I'll call it, science certified. I'm kind of making that term up, but there's an external third party, independent scientific community that actually certifies that what they're doing is not just buying carbon credits, not paying their way and elbowing their way into sustainability, but working toward a true scientific-based net-zero posture. Yeah, and I think they could even take a stronger approach and if you look at this greenwashing as it would be of a lot of other companies that they are putting out, hey, here's your carbon footprint tool and it'll tell you that, but again, HPE and over multiple different discussions talked about supply chain, that you can't go to many vendors who can talk to, here's where I partnered with this processor company and they helped enable me to understand how my supply chain, so you can get to scope one, scope two and scope three, those types of requirements that are science-based, math-based stuff, not, you know, I pay for a carbon credit because I took a flight here kind of thing. Right, now the other notable moment was on the keynotes where Antonio was on the stage with Matt Wood, right from Amazon and it was a tale of two stories. You had Matt Wood saying, well, in front of an audience of 10,000 HPE customers said, well, we still believe in the fullness of time that all workloads are going to go into the cloud, but in the meantime, you know, we'll partner up with HPE, which I thought was stunning and of course Antonio smiled at him and said, yeah, it's a hybrid world and it's going to basically stay that way, I'm paraphrasing, and so I had, I think an interesting tweet on that that got some action, but I think actually Antonio's going to be on the right side of history of that one. I mean, unless AWS redefines the definition of cloud as to include things like GreenLake and Apex and other as-a-service sort of approaches that have the cloud operating model, everything's not going to go into the cloud and then the, you know, one of your thoughts in this, the Equinix announcements, we saw that from Dell, you know, in May, we saw it here with HPE basically pre-populating GreenLake inside of Equinix. You know, that's a form of cloud and it's right next to a Amazon data center and so you've got your latency down and you don't have to build your own data center and your own infrastructure, so that's like the best of both worlds. Yeah, I mean, if I can steal an Equinix saying that they have, they're cloud adjacent. That's how they like to say it. And I think having private cloud enterprise and the private cloud business edition, which is PCBE, which PCBs are bad, so we've got to get away from that, but you know, other than that, I think it's good in the fact that it is where the customers want to be and I think it makes a lot of sense. I also think it builds on the fact that they had this relationship with digital reality already and this helps give them a much larger addressable market in that and I think it just makes total sense. Yeah, so EKS anywhere now on GreenLake and it's kind of funny because in 2000, I think it was 18 when Amazon announced Outpost. We're like, okay, folks, here it is. And that was like a wake-up call to the HPEs of the world to actually get their act together and build it as a service model and it took them a while. My opinion that should have been done, they should have started this years before, but fine, it was a wake-up call. But now what you see, Rob, I'd love to get your take on this, is this is what Outposts should have been. Yeah, right. I don't think it's an outright admission that Outposts haven't done what they wanted to do. I think Outposts are great for little local zones and things like that where you're using the hardcore AWS services but when you're looking at EKS anywhere and Kubernetes in particular, that doesn't just live on an Outpost and I think that when they start to look at EKS anywhere, you start to look at the data behind it, it's nowhere and I think that this is when they're trying to go out and compete against their partners like we had Red Hat on and others that are out there building inside of AWS. I think it's a very shrewd move by AWS to bring EKS anywhere to GreenLake. It makes total sense. The other thing, no, actually let me add to that. So the other thing is you didn't hear anything in the keynotes, at least not much than I heard about Microsoft and Azure. And here's the funny thing, Rob, as you know as well as I do. Azure Stack for years was this sort of cobbled together. It wasn't that great. It was a classic Microsoft, yeah, it kind of works but it really didn't. But Azure Arc is kick-ass. I mean it really is a strong hybrid story and I know for a fact that HPE is doing a lot of business in that hybrid area with Microsoft. We didn't really hear much now but on the show, Flo, you heard a lot about it. So they're actually doing business in the field. So that was kind of an interesting little. And I think what's really interesting about that is you didn't hear about AKS, which is the Azure Kubernetes service, their open-source version of it, where that's actually one of the largest growing products at Microsoft. I would expect that maybe we're sitting here next year talking about AKS for GreenLake products. Absolutely. The other thing is the ebb and flow of the tech industry. You saw Ragu up there on stage with Antonio and they kind of tongue-in-cheek, but maybe not. Hey, it's been a while. Because you had Dell-owned VMware, so if I'm HPE, I'm going, well, okay, how hard am I going to lean into a VMware partnership when they're owned by Dell? They spin them out and now Broadcom's buying VMware. It looks like that has a chance of going through, a good chance of going through. HPEs, which is a huge partner of VMware, they move a ton of VMware licenses, it's like, hey, let's get back into it. We got VMware, we have Red Hat, we got SUSE, is it another partner that we heard from? Not a ton of open source on stage, but you and I I think brought open source to the conversation. Yeah, I mean, open source is one, but at the same time, you still have these lots of VMs out there. It's like everybody saying that this is going to go away overnight and everything's going to be cloud native is just not true. And I think that, having that relationship, being part of PCBE, what I believe is the one it's part of, so when you start to look at the different private cloud offerings that HPE is putting out there and how they're going to market with VMware as part of that, along with Red Hat and SUSE and others, I think it makes total sense that they get closer together again at this point. Yeah, and a couple of other sort of highlights. I mentioned Dr. Go and Antonio Neary was clearly a highlight of the week, always great on theCUBE. We also had Alpa, I think I'm saying that right, German startup company. We had their CEO and founder on who basically is an AI expert and has developed and he's worked on large language models for years and years and years. They've developed an explainable, trustable, transparent LLM that HPE is utilizing for specific use cases like legal as an example. So where you want to build your own LLM, you want to train your own models and have your own data, not have it leak into the public domain and that's something that was kind of interesting because they're working with the startup to do that. Yeah, and I was pretty good interviewer. I thought it was very interesting that they were doing that. Yeah, and then, of course, you're also seeing a lot of M&A out of these guys. I mentioned there was four this year, OpsRAMP, which was a little unclear what they're doing there, but I think Clarity will come in the next several months. Pachyderm, which is AI tools, Access Security, we were just talking to Phil Matrim about that a moment ago, which is an Israeli-based company, Athanet, which we talked about at Mobile World Congress. That's when that deal went down. Determined AI was done maybe 2022, 2021, but again, another AI tuck-in software company that HPE has acquired, and then, of course, there's Esmeral, which is kind of that software division. Of course, HPE software is, I guess, Jason Newton was saying, is more than just Esmeral. Yeah, they're putting in there the cloud-operating model tools into that, as well as part of Esmeral and being part of it. So I think that, again, you get a lot of different things that are coming together, and the fact that I was here for the three years when it broke apart and software went away, and I think now that it's coming back, but in a very, I would say, a very determined and a very specific way that they're bringing back software, that it's really not trying to do everything and be generic software, but how do they support that operating model, that cloud-operating model? Yeah, it's interesting. They, I think, sort of basically cut out the entire software business pretty much and said, oh, here you go. It included the data protector software and included the security business. The open-stack stuff. The open-stack stuff. Basically everything, and a lot of that was sort of legacy, which should probably work pretty well for MicroFocus because they can treat it like a PE firm, but then that gave HPE the opportunity to start from scratch, essentially, and build up a much more focused and modern software portfolio. We'll see where that goes. I think they still get some work to do there and to make it a really noticeable business. Yeah, and I think they have a great advantage in the fact that you hit on the platform being the Aruba Central and building on top of that and I think having that as their north star and aiming for that makes a lot of sense. Rob, thanks for sitting in today. For Lisa, it's fantastic to be working with you. You've got to bring the knowledge of HPE, you worked at AWS, you worked at Zerto, you're seeing that. That was the other interesting thing that we didn't really talk about. You had, we had Veeam on and then you got the Zerto integration and so that's always interesting. Veeam's killing it with AR. They're still the largest part of it. Yeah, that's not going to change. Love Veeam and so. All right, we're going to wrap here. Okay, so go to thecube.net for all the on-demand videos. SiliconANGLE.com has all the news and analysis. As I say, John Furrier this week at MongoDB in New York City. Next week, we'll be back in Vegas for Snowflake Summit. I'll be here with Lisa Martin and George Gilbert. As well, John Furrier is going to be in San Francisco at the Databricks, Databricks AI Summit. Rob, you're going to be there. John will be there. George is actually going to fly out to that one as well. So we're going to have data week and that's going to be amazing. So check all that out next week at thecube.net. Thanks for watching everybody. This is the day three wrap HPE Discover 2023. Thanks for watching The Cube, the leader in enterprise tech coverage. We'll see you next time.