Why HPE Stock Is Surging on the Vultr AI Data Center Win

Hewlett Packard Enterprise is trading around $48.54 after climbing 8.66%, and the move has a specific name attached to it. Vultr, the independent cloud provider, selected HPE alongside NVIDIA to build its next generation of cloud-scale AI data centers, choosing NVIDIA GB300 NVL72 systems delivered by HPE and wired together with NVIDIA Spectrum-X Ethernet networking. That is a design win in the exact place investors have been waiting for one, because it puts HPE inside a neocloud buildout rather than at the edge of an enterprise refresh cycle.

HPE snapshot, July 13, 2026

- Price: roughly $48.54

- 24-hour change: +8.66%

- Catalyst: Vultr selects HPE and NVIDIA for cloud-scale AI data center infrastructure

- Last reported cloud and AI revenue: $7.7 billion, up 22.9%

- Level to watch: the $45 shelf that capped the stock before this leg

The market is treating this as a re-rating rather than a one-day headline pop. Here is what Vultr actually bought, why the win carries more weight than its size suggests, and what has to hold for the move to stick.

What Vultr Picked HPE and NVIDIA to Build

Vultr is not buying servers. It is buying a rack-scale AI factory, and the distinction matters for how HPE books the revenue and how sticky the relationship becomes.

The core of the deployment is the NVIDIA GB300 NVL72, a single liquid-cooled rack that fuses 72 Blackwell Ultra GPUs and 36 Grace CPUs into one coherent compute domain. HPE delivers those racks as part of the NVIDIA AI Computing by HPE portfolio, then connects them with NVIDIA Spectrum-X Ethernet, the fabric purpose-built to keep GPU clusters from choking on their own east-west traffic. Around that hardware sits the part vendors rarely brag about but customers actually pay for, which includes liquid-cooling design, power and thermal engineering, deployment services, and multi-year lifecycle support.

HPE laid out the arrangement in its own press release on the Vultr selection, framing it as infrastructure for enterprise private cloud, model training, and large-scale inference. Vultr's demand profile is the reason it reads as a validation. Vultr sells GPU capacity to companies that need it now, so the operator is committing capital against bookings it can already see.

Liquid cooling is the quiet moat here. Once a GPU rack crosses roughly 130 kilowatts, air cooling stops being an engineering choice and becomes a physical impossibility, and the vendors who can design, install, and service closed-loop liquid systems at scale are a short list. HPE has been building that capability since its supercomputing days, and the Vultr win is where it finally shows up on a commercial cloud order.

Why the Vultr Win Matters More Than Its Headline Size

The dollar value of a single neocloud order is almost beside the point. What changed is the category HPE gets sorted into.

For two years the AI infrastructure trade has been priced as a narrow club. NVIDIA supplies the compute, a handful of networking and memory names ride along, and the legacy enterprise hardware vendors get treated as the losers of the transition rather than the plumbing of it. That framing has been generous to the silicon and stingy to the systems layer. Traders who already follow the compute side through NVIDIA and the NVDA stock story or the memory side through Micron's climb toward the trillion-dollar club have watched those names re-rate hard while HPE traded like a dividend stock with a modest AI kicker.

A cloud operator choosing HPE to build the whole environment breaks that assumption. Vultr could have gone to any of the white-box integrators that dominate hyperscaler procurement, and it did not, because at rack scale the integration risk is the real risk. And when a customer pays for someone else to carry that risk, the margin profile of the order looks nothing like a commodity server sale.

The second-order effect is the pipeline. Neocloud operators talk to each other, benchmark each other, and copy each other's reference architectures. One public deployment of GB300 racks by HPE becomes the sales deck for the next five conversations, which is the same flywheel that turned the networking silicon vendors into AI names. The same dynamic is visible across the sector, from Marvell's AI outlook to the custom-accelerator strategy laid out in the Phemex profile of Broadcom CEO Hock Tan.

The Juniper and GreenLake Backdrop Behind the Move

Today's move is not standing on the Vultr headline alone. It is standing on a quarter that already showed the AI and networking engines working, and the market has spent the weeks since that print slowly deciding it believes the numbers.

HPE's fiscal Q2 2026, the quarter that ended April 30, 2026, is the reference point. The company reported cloud and AI revenue of roughly $7.7 billion, up about 22.9%, while networking revenue reached roughly $2.7 billion, up about 148.2% year over year on the back of the Juniper Networks integration. The full figures sit in HPE's fiscal 2026 second quarter results release and the historical series on its investor relations quarterly results page.

Juniper is the piece most investors underestimated. Buying a networking business at the exact moment AI clusters made the network the bottleneck looked expensive when it was announced, and it looks different now that every GPU deployment needs a fabric decision made at the same time as the compute decision. Spectrum-X handles the intra-cluster traffic, and the campus, data center, and WAN layers around it are where Juniper sells.

GreenLake and HPE Morpheus are the reason this can compound. GreenLake bills infrastructure as consumption rather than a capital purchase, and Morpheus adds the agentic AIOps and private-cloud orchestration layer on top. Hardware revenue is lumpy and gets a hardware multiple. Consumption revenue is recurring and gets a software multiple, and that arbitrage is the bull case that HPE has been arguing for years without much of an audience. It is the same argument that re-rated Oracle's cloud and AI capex story, and the market rewarded that one eventually.

What the Vultr Win Means for HPE Stock From Here

A +8.66% day on a design win tells you positioning was light and expectations were low. That is usually a constructive setup, but it also means the stock is now carrying an expectation it did not carry last week.

Three things decide if this leg holds. The first is follow-on orders, because one neocloud win is a data point and three is a trend, and the market will give HPE roughly one quarter to convert the Vultr reference into named additional deployments. The second is gross margin on AI systems, since the persistent bear case is that AI server revenue arrives at thin margins because NVIDIA captures the value and the integrator captures the logistics. If HPE shows that the services, cooling, and networking attach lifts blended margin, the multiple expands. If AI revenue grows while margin compresses, the stock goes back to being a value name with a story.

The third is the $45 area. That shelf capped HPE before this move, and it is now the line that separates a re-rating from a failed breakout. Reclaimed and defended, it becomes support and the prior highs come into play. Lost on a close, and the Vultr news gets filed as a headline the market already forgot.

One structural note for traders. HPE is not among the tokenized equity pairs available on Phemex, so the cleanest way to express an AI infrastructure view on the platform is through NVDA, currently around $208.49, which is both the tokenized AI proxy and literally the compute inside the Vultr racks. The correlation is not perfect, but when the thesis is that AI data center buildouts keep accelerating, NVDA is the instrument that expresses it directly.

Frequently Asked Questions

Is HPE stock a buy in 2026?

The bull case rests on three things holding together, which are AI order conversion, the Juniper networking contribution, and margin on AI systems. The Vultr win supports the first two and says nothing yet about the third, so the honest position is that HPE has become a credible AI infrastructure name without yet proving it is a high-margin one. Size it as a cyclical infrastructure position rather than a growth position.

What did Vultr actually buy from HPE?

Vultr selected NVIDIA GB300 NVL72 rack-scale systems delivered by HPE, connected with NVIDIA Spectrum-X Ethernet networking, along with the liquid-cooled rack design, deployment services, and lifecycle support that go with them. The deal covers infrastructure for enterprise private cloud, model training, and large-scale inference workloads rather than a simple server order.

Can I trade HPE stock on Phemex?

HPE is not currently one of the tokenized stock pairs on Phemex. Traders who want exposure to the same AI infrastructure theme generally use NVDA, which is available as a tokenized perpetual and sits at the center of the Vultr deployment as the compute supplier.

Why is networking revenue growing 148% at HPE?

The jump comes from the Juniper Networks acquisition being folded into the reported numbers, which adds a full networking portfolio to a quarter that previously counted only HPE's own Aruba business. Underneath the accounting effect, AI clusters genuinely need more networking per unit of compute than traditional workloads, so the organic demand is real even after the acquisition math is stripped out.

The Bottom Line

Hold $45 and convert the Vultr reference into a second named neocloud win, and HPE stops trading as a legacy hardware vendor and starts trading as AI infrastructure, which is a different multiple entirely. Show AI revenue growth with flat or falling gross margin at the next report, and the market reprices it right back to where it sat before $48.54. The $7.7 billion cloud and AI line and the 148.2% networking growth already prove the demand exists, so the open question is no longer if HPE participates in the AI buildout. The question is what it earns for participating, and the next earnings print is where that gets answered.

This article is for informational purposes only and does not constitute financial or investment advice. Cryptocurrency trading involves substantial risk. Always conduct your own research before making trading decisions.