Scaling AI Beyond Pilots: HPE and Vultr on Networking, Partnerships, and the Path to Production

Scaling AI and cloud infrastructure is no longer a question of whether enterprises can access compute. It's a question of how fast they can get it without absorbing operational complexity that cancels out the speed advantage. That's the exact tension the Hewlett Packard Enterprise and Vultr partnership is built to address.

At HPE Discover 2026 in Las Vegas, Patrick Moorhead and Daniel Newman sat down with Phil Mottram, EVP and Chief Sales Officer at HPE, and Nathan Goulding, SVP of Engineering at Vultr, to discuss how the HPE–Vultr partnership is powering the next wave of enterprise AI deployment.

Goulding explains why Vultr—the world’s largest independently held cloud computing company, with more than two decades of operating history—is often miscategorized as a “neocloud” and how partnering with HPE on Helios, scale-up networking, and liquid-cooled Tomahawk 6 in the rack gives Vultr the foundation to bring next-generation AI infrastructure to customers ranging from foundational model labs to financial services to “vibe coders.” He also unpacks why agentic AI, which he describes as “a CPU application with a bad GPU habit,” is driving demand inside the four walls of corporate IT faster than supply can keep up.

Mottram, drawing on his networking background, breaks down why the network has become the underlying fabric for AI performance—covering capacity, throughput, latency, and security—and why the timing of the Juniper acquisition was perfect for HPE’s networking portfolio. He also explains how self-driving network technology from Juniper Mist and Aruba Central can already resolve roughly 90% of customer network issues without human intervention, with 100% within a couple of years. The conversation closes with both leaders laying out what enterprises will prioritize next: speed, security, data sovereignty outside the US, and predictable, controllable cost—especially as token economics start to feel uncomfortably familiar to anyone who lived through early cloud sticker shock.

Key Takeaways:

🔹 Enterprise AI is moving from pilots to production. Mottram notes that most customers are now running real production use cases, with CEOs themselves acting as the new chief AI officers and applying pressure to move quickly.

🔹 Vultr is not a typical “neocloud.” Goulding emphasizes that Vultr is the world’s largest independently held cloud company with more than two decades of operating history, which is why foundational model companies, financial services firms, and enterprises rely on it for compliance, security, and scale.

🔹 The HPE–Vultr partnership extends well beyond this week’s announcements. Vultr has been running Juniper on its edge network for over a decade, and the companies just announced a joint NVIDIA NVL72 RackScale deployment through HPE.

🔹 Networking is now a determining factor in AI performance. Mottram explains that capacity, throughput, latency, and security in the network fabric directly shape whether AI workloads—training or inference—actually perform at scale.

🔹 Self-driving networks are already resolving most issues without humans. Juniper Mist and Aruba Central can resolve roughly 90% of network issues with no human intervention today, with Mottram projecting 100% within a couple of years.

🔹 Agentic AI is reshaping enterprise demand. Goulding describes agentic workloads as “a CPU application with a bad GPU habit,” driving an enormous demand surge across foundational models, financial services, and consumer-grade vibe coding.

🔹 The next wave of AI customers will be won on four priorities. Mottram and Goulding align on speed, security, data sovereignty (especially outside the US), and predictable, controllable cost—particularly as token economics begin to mirror the early cloud cost-management challenge.

Watch the full video at sixfivemedia.com, and be sure to subscribe to our YouTube channel so you never miss an episode.

Disclaimer: Six Five Media is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.

‍

PHIL MOTTRAM:
AI is important from a networking perspective to build solutions, but also AI can really help as a platform technology. I mean, you're now at a point where the network platforms can kind of resolve like 90% of the issues that customers see with the network with no need for any human intervention. And within a couple of years, we'll be at 100%.

PATRICK MOORHEAD: 

Six Five On The Road here in Las Vegas, HPE Discover 2026. It's been a great show so far. We're talking about full stack.

DANIEL NEWMAN: 

In reality, it's just been a great couple of days here. Keynotes from Antonio Neri, Rami Rahim. But, you know, the energy here, I think I've heard that this is the largest HPE Discover yet. A big celebration tonight for those that celebrate. And of course, in the background, we've gotten to watch the World Cup. You know, that's every time I walk through the casino, I see the games on. It's a little distracting for me as a soccer guy, but that's okay. We have the AI World Cup going on right now. And you know, the folks that we have here today are playing in the game right as we speak. That's right. And Daniel… Good analogy.

PATRICK MOORHEAD: 

It was a good analogy. We've always talked about on the show how it takes, everybody can't do it themselves, it takes partnerships, and we've seen that more than ever with agentic AI, and this is the focus of this conversation. Phil, welcome back to The Six Five, and Nathan, welcome to The Six Five.

NATHAN GOULDING: 

Thank you.

PATRICK MOORHEAD: 

Yeah, let's dive in here.

DANIEL NEWMAN: 

Vulture as a, you know, quickly going cloud AI provider, you've got a lot of great announcements, deep partnership here. We'll talk about that a little bit, but you're also here, you know, as a partner of HPE and they've had a lot of announcements over the last couple of days. Just curious from your perspective at Vulture, kind of like what stood out so far from what HPE has announced here at the event?

NATHAN GOULDING: 

Yeah, absolutely. Well, first of all, thanks for having me. I'm really excited to be here. You know, Vulture is the world's largest independently held cloud computing company. And many people don't know that we've actually been operating as a company for over two decades. And so we often get lumped into the neocloud category. And obviously, AI is a huge demand driver for us. But we've been doing this for a long time. And we have a key to our success is relying on partners like HPE. And so some of the things that we've seen here, I mean, Antonio mentioned it in his keynote, but specifically some of the scale up networking that we're seeing inside of the Helios rack is really exciting. Being able to bring a fully liquid cooled Tomahawk six into the rack with Helios is extremely exciting. That's one of the things that we're excited about as we bring Helios to market. Being able to partner with HPE is, Our success really is reliant on the partners that we choose. And so HPE is probably one of the only companies in the world that has the longest and richest history with deploying some of the world's largest and fastest supercomputers on planet Earth. And so that gives us confidence as we bring this technology to our customers, being able to partner with folks like HPE.

PATRICK MOORHEAD: 

That's great. So, Phil, two hours ago, I was doing a presentation to a board about where are we on the AI chart, right? Are we truly stuck in pilots to production? Are we scaling? And a lot of the announcements, a lot of the themes here are about getting customers farther along that curve. What do these innovations ultimately mean, though, for customers trying to scale AI and ultimately make money off of it? it?

PHIL MOTTRAM: 

Yeah, I think in terms of where we are on the journey, by the way, I think it's a lot more than pilots now. I think it is largely moving to production cases with lots of customers. Albeit, I think, lots of pressure from the board. I mean, we were talking to some customers yesterday that said, you know, for the most part, the CEOs become the new kind of chief AI officer. So there's a lot of pressure in organizations to do things and do things quickly. But as I say, when they're trying to do things, speed's important, being really careful about where your data's going to be and making sure it's not compromised in any way, and also being able to control the costs, right? So the cost of tokens could get out of control for some customers. That's what we're hearing is important to customers. And you know, our solutions and working with partners like Vulture is where we're trying to help the market move along the AI journey. Yeah, it makes sense.

NATHAN GOULDING: 

Yeah, it's a really interesting point though. I feel like, I would say maybe even just a year or two ago, the mandate from the board of the CEO was, hey, how can we figure out, there's this big AI wave, how are we as an organization supposed to be adopting it? And the first iteration of that was maybe an LLM that was the chatbot that was maybe hooked up to your public documentation, but really was unenabled and not that useful. And now with the massive boom in agentic AI being incorporated inside the corporate network that actually has access to all of the applications and the MCP servers, is actually able to take actions and perform tasks. That's just driving enormous demand within the enterprise. And so I don't, you know, maybe a year or two ago, we were on the tip of the spear in terms of the enterprise adoption, but we're certainly seeing an enormous amount of growth within more traditional enterprise as agentic AI is coming into the, you know, kind of the four walls of corporate IT.

DANIEL NEWMAN: 

It's really interesting you said that in just such a short period of time, you know, we've kind of had all these kind of this emotional loop. Is AI going to replace everything? And, you know, is it even or is it just all slop and garbage? I mean, I think, like I said, there was this inflection maybe about six, seven months ago around Opus 4.6 that suddenly the utility really started to shine. And then it was like, oh, this is so good. It's going to wipe all the knowledge work out. And then we actually see is that in a lot of companies, the ones that are applying it really well, there's this growth of the numerator, right? We're always talking about the denominator, like, oh, well, if all things stay the same and we can bring costs down, but what if we can drive massive amounts of new productivity? And I think we're starting to see what's possible from some of these companies. It's been really interesting. And by the way, you know, it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's a it's What is driving your growth right now? And talk a little bit, you know, Phil's here. We got to do this for Phil.

DANIEL NEWMAN: 

Talk a little bit about how the HPE partnership is really driving it. I'm just taking my job over. I'm meant to be the sales guy. Oh, no, no. Do the selling for Phil.

NATHAN GOULDING: 

No, I mean, truly, we could not. We've been deploying, obviously Juniper now, part of the HPE family, but we've been deploying Juniper for over a decade. Juniper has been running our edge network at Vulture again for over a decade. And so our partnership with HPE dates back a long time ago. And now, obviously, we're considering, we just had an announcement today, partnering with HPE and NVIDIA. to deploy NBL 72 RackScale solutions through HPE. In terms of the demand that we see, I would say it lumps it into a few different categories. I would say there's foundational model companies who are coming to us, looking to us. Again, because we have a rich operating history, we have things like compliance, things like security, which are of utmost importance to foundational model companies, financial services companies, companies where, again, it's not just, how do we incorporate AI into our business? These are companies who, specifically in the case of financial services, the models that they're able to train using AI is a huge driver of their business. And then, downstream from that, they're and try to consume as much as we can deliver as fast as we can deliver it. And so there's an enormous amount of demand for that. And it's not just, I mean, obviously vibe coding and kind of consumer grade AI is obviously also experiencing an enormous amount of growth and placing tools into your hands where you can vibe code a health app. You can vibe code an application that is able to connect to your phone and message. He calls himself a full stack.

PATRICK MOORHEAD: 

Welcome to full stack development. My apps last about one day before they break. 

Security holes, but I'm a developer.

NATHAN GOULDING: 

Yeah, no, exactly. And truly, the ability to be effective with these tools is your ability to articulate the problem to the agent on the other side. And so anybody who is able to articulate that is able to then manifest the solution that they want, which is very exciting for these sorts of consumer grade AI solutions. But it really runs the gamut, and I think that's really the turning point that's seeing a lot of uptake in the enterprise and organizations, which then, again, from our perspective, there's such a, you know, there's, the demand and the supply has not yet equaled out. And so, you know, all of these agentic applications, you know, they're, it's fundamentally, it's a CPU application with a bad GPU habit. And it's just these CPU applications are just consuming more and more GPU resources. And so that's driving a ton of demand for us. And it runs across many different verticals from, again, foundational model companies to financial services to some of the other cutting edge platforms that are out there. And we strive to be the platform for AI platforms. What are the things, if you're deploying something at hundreds of megawatt or gigawatt scale, what can we do as an AI infrastructure provider to provide the best platform and product for them and how they want to consume their AI infrastructure?

PATRICK MOORHEAD: 

Makes sense. So Phil, you must have had a big smile on your face. Antonio kicks it off and basically the headline is networking. And you know a little bit about networking. A lot about networking. In fact, the last time we had you on the show, you were running networking. But I mean, it seems very intuitive to me that any workload, you need to have a properly sized compute, storage, and networking. But it's not obvious to everybody. talk about why you have to have that for performance, efficiency, and ultimately for outcomes.

PHIL MOTTRAM: 

Yeah, exactly. Right. Well, I think there's quite a few. One that you didn't mention there is security. I think that's really becoming a big issue now for organizations, particularly in an AI context. Where are their security flaws and weaknesses. But yeah, I think a network is the underlying fabric that's going to link all of this together. And then you need something that's secure. But also you need something with the right amount of capacity throughput and also latency, because latency is really important for some of these applications. So Yeah, I mean, I think Antonio's timing with the Juniper acquisition was perfect around, you know, kind of enhancing the HPE networking portfolio. And so we're super excited about that. And then obviously the other aspect with AI and networking is how we're using AI in the network platforms to deliver a better experience to customers. I'm sure you heard the self-driving networks story, but I mean, What Juniper did with Mist, what Aruba did with Aruba Central is years ahead of what the competition has been doing. You're now at a point where the network platforms can resolve 90% of the issues that a customer may see with the network with no need for any human intervention. And within a couple of years, we'll be at 100%. So, you know, AI is important from a networking perspective to build solutions, but also AI can really help as a platform technology to deliver a better solution for customers.

PATRICK MOORHEAD: 

Yeah, I know there was some debate out there on networking is the lead, but it's like when I look at, OK, I can have this great rack with a bunch of great compute in it. I could do scale up, scale out, scale across. It is the outlet to make non-efficient workloads on a GPU or an XPU. come together for a combined solution that's required. It's almost like an outlet piece, and in a perfect world, hey, it's all scale up. And it's like, well, okay, we're going to scale out, and if we can't scale out and achieve, we're going to scale across. And anyways, networking is just front and center, and markets have responded, and I think enterprises realize that they need to upgrade. And I'm glad, though, you hit on that portion, whether it's Mist or Aruba Central in automating all of those things, because I remember in cloud, it took me a day to spin up a CPU. It took me two days to spin up storage, and then like 10 weeks to spin up networking, right? But I know we're beyond that, but that's an important point.

NATHAN GOULDING: 

I think it's very important to have, you know, we obviously are in the business of making hardware move at software speed, and I think a huge part of that is the network, especially with these massive AI clusters. And there are different technologies, you know, have taken a different approach. Like, you look at some of the custom silicon vendors, you know, their strategy is build a bigger wafer and, you know, see how large, like, get more models onto a single wafer. And then you look at other strategies of, okay, well, we do have to intercompete these together. How can we get the networks that are connecting these together to be transferring data faster and faster? And it's really, it's not a mention of what's more important, it's where's your bottleneck. And so if you're doing training, it's getting the data from persistent storage onto the GPU over the network. If it's inferencing, it's getting, how can I optimize my KD cache for better performance over the network? So the network is truly super critical and the thing that really makes it all go. And without that, you have something that isn't going to be effective.

DANIEL NEWMAN: 

Well, you sort of started answering my last question, but look, we know demand is in the enterprise is going to continue to explode. So maybe a quick from each of you to wrap this up today, but like what are, start with you, Nathan, what are you prioritizing over the next few years to attract that next generation AI and cloud customer?

NATHAN GOULDING: 

It's a great question. I think that as the adoption curve has gone from tip of the spear to more mainstream, the things that are going to be important are security, data governance, sovereignty. These are things that are incredibly important to the enterprise, especially as they consider bringing these AI tools inside of their corporate network. Partnering with a company like Vulture, the things that we're focused on are being able to deliver those tools and capabilities to our customers so that they can adopt AI in a safe and compliant and controlled manner. That's going to be probably the single biggest thing that's going to drive the next wave. And what can we do as an infrastructure provider of AI to enable organizations to do that?

PHIL MOTTRAM: 

Yeah, I think roughly the same, but speed. Companies are under a lot of pressure to deliver and deliver quickly, so that's definitely one. Security is a big thing now, and I think it's only going to get bigger. Data sovereignty, particularly outside of the US. If you go and speak to other global customers in other markets, it's a huge thing. And then last but not least is cost. And it's having a predictability around the cost. So it's, you know, cost in overall terms, but also being able to be predictive and contain your costs. So that's where I think it's at for the future.

PATRICK MOORHEAD: 

Having flashbacks of cloud.

DANIEL NEWMAN: 

Yeah, exactly right. Exactly right. Token maxing to token optimizing. And it's like, two months. Yeah. Wow.

PATRICK MOORHEAD: 

I can swipe my credit card and you know, I get $50,000 in tokens.

PHIL MOTTRAM: 

You're exactly right. It is. It's literally in the last two months, isn't it? I mean, four months ago, I don't think people were talking about this. Now they are. It shows how quickly this thing's moving, doesn't it?

DANIEL NEWMAN: 

Nathan and Phil, I want to thank you both so much for joining us here on The Six Five. I look forward to having another conversation soon, tracking the progress that Vulture makes. And of course, we'll be watching the numbers very closely to make sure you're watching. Everyone's watching the numbers. Everybody's watching what Phil's doing. Exactly. All right. And, you know, thanks, everybody, for joining us. We'll be right back.

‍