Image: AndersonPiza/Envato
HPE’s private cloud refresh brings Morpheus, Zerto, and Alletra MP into a unified AI-ready stack, while Juniper’s role remains unclear.
Hewlett Packard Enterprise’s latest private cloud and storage solutions align its portfolio around a unified operating model for hybrid cloud and AI. The announcement is well-timed, as most customers are rethinking where their workloads run and how they’ll operationalize AI at scale.
From my conversations with customers, I’ve found they’ve struggled to cobble together infrastructure from multiple vendors. HPE’s solution is designed for turnkey, simplified deployment.
HPE is positioning this as the fourth generation of its private cloud, consolidating a confusing set of SKUs, including Private Cloud Business Edition, Private Cloud Enterprise, SimpliVity, and others, into a single HPE Private Cloud offer with multiple form factors (PC 1000 hyperconverged, PC 3000 disaggregated, and PC 7000 fully managed “as a service”).
The primary benefit is a single control plane, powered by HPE Morpheus, for VMs, Kubernetes, and AI workloads, whether they’re running in the data center, a colo, or at the edge. As mentioned above, customers have struggled to stitch together separate stacks for virtualization, containers, and AI. They want one opinionated platform that still preserves the choice of hypervisor and deployment model.
This matters because the economics and architecture of virtualization have shifted dramatically over the past 18 months.
The biggest factor has been Broadcom jacking up VMware hypervisor licensing costs, and customers are also repatriating workloads from the public cloud and need to quickly stand up AI. This conflicts with brittle, VM-centric architectures that were never built for distributed, GPU-heavy workloads.
HPE is trying to tap into that frustration by promising a lower-friction, multi-hypervisor private cloud that feels like the cloud but runs where the data is.
Although HPE did not name Broadcom’s VMware during the press briefing, the intent is clear.
This private cloud stack is designed as an exit ramp for customers reeling from VMware’s pricing and licensing changes. By placing HPE Morpheus at the center as a multi-hypervisor control plane, HPE enables enterprises to introduce their own hypervisor and other options alongside VMware, then shift workloads over time rather than through a risky big-bang migration.
Zerto 10.9 adds a mobility and protection layer that continuously replicates workloads and orchestrates failover between VMware and HPE platforms, enabling customers to prove resilience before cutting over. Combined with PC 3000 disaggregated systems for core data centers, PC 1000/SimpliVity at the edge, and a VMware-like self-service and automation experience, HPE offers a way to modernize for AI and hybrid cloud while steadily shrinking, rather than abruptly ripping out, the VMware footprint.
At the core of the news is HPE’s insistence on “one operating model” and a “single pane of glass” for all infrastructure actors: developers consuming VMs and containers, platform teams setting policy and security, and ops teams managing hardware lifecycle.
Key elements:
For customers grappling with “AI sprawl,” including islands of GPU nodes, shadow projects across business units, and fragmented toolchains, this unified model solves many issues. HPE’s positioning is that customers can’t standardize on a single AI framework or public cloud, but they can standardize on a single operational fabric that spans them.
This creates significant choice at the infrastructure and cloud layers.
Another pillar is cyber and operational resilience, enabled by HPE Zerto. HPE is reframing resilience as a board-level capability, not just an IT checkbox, and is baking it into the private cloud story rather than treating it as an add-on.
New Zerto capabilities include:
Combined with integrated backup, ransomware protection, and StoreOnce-based recovery across the private cloud and the Alletra MP portfolio, the message is that resilience is not something customers bolt on later; it is built into the platform and exposed through the same control plane.
Given that HPE has now closed its $14 billion acquisition of Juniper Networks and doubled the size of its networking business, the relative absence of Juniper and Mist in this announcement was surprising.
When asked directly why networking, and particularly the Juniper assets, were missing from the private cloud architectural slide, HPE’s answer was that it will “be shipping opinionated networking” for the private cloud interconnect, using HPE networking to hide complexity from customers while allowing them to plug into heterogeneous aggregation networks (Cisco, Arista, and others).
That raises several open questions:
Given HPE’s own description of the Juniper deal as creating a “full, modern networking stack” that is “purpose-built with AI and for AI,” the lack of a clear, explicit networking layer in what is otherwise a very complete private cloud and data platform is an obvious gap — and one customers will be asking about.
From a customer perspective, private cloud, storage, and AI infrastructure are only as good as the network paths connecting GPU clusters to data stores and edge sites. Latency, microbursts, and congestion don’t care how elegant your control plane is.
HPE’s answer, that it will provide a safe, managed interconnect for the private cloud while accommodating whatever aggregation network the customer already has, is pragmatic, but it undersells the potential of an integrated, AI-driven network stack spanning from top-of-rack to the WAN and campus.
For this portfolio to be truly differentiated, HPE will need to:
Until that happens, networking remains a promise, not yet fully realized, within HPE’s “one operating model” vision.
The expanded hybrid cloud stack is certainly the right direction, but several open questions remain. I already addressed networking, so I won’t rehash it. Below are other things customers should consider.
HPE emphasizes the choice of hypervisor and support for its own HPE VM inside Morpheus. Customers should ask for specifics on supported combinations, migration tooling, and licensing impacts as they unwind VMware dependencies.
The Data Fabric and AI-driven governance capabilities are promising, but how deeply will they integrate with third-party catalogs, privacy tooling, and LLM governance frameworks?
Can customers declaratively express where sensitive data may or may not be used for training versus inference, and have the platform enforce that end-to-end?
Read more: What enterprises need to know about sovereign AI.
HPE asserts that hybrid is “more hybrid than ever” and that AI is pulling data back on-prem or to the edge, but hard metrics comparing the TCO of this private cloud stack with public cloud AI services will matter. Customers should ask for documented case studies in which HPE’s disaggregated private cloud, plus Alletra MP, has delivered measurable savings and performance improvements for AI workloads.
Zerto’s AI assistant and Data Fabric’s conversational controls are early examples of agentic AI in infrastructure operations; the natural next step is agents capable of safely executing changes. What guardrails, audit mechanisms, and rollback capabilities will HPE provide as it moves from insights and recommendations to autonomous remediation across compute, storage, and eventually networking?
For HPE, this launch is an important milestone: it unifies Morpheus, Alletra MP, Data Fabric, and Zerto into a more coherent story about building a modern, AI-ready private cloud that lives where the data is and can withstand real-world failure modes.
The next test – and the opportunity to fully capitalize on the Juniper acquisition — will be to make networking as visible and integrated into that story as compute, storage, and data protection already are.
Also read: Google Cloud’s agentic AI roadmap shows how enterprise platforms are being rebuilt around data, security, governance, and AI workloads.
Zeus Kerravala is an eWEEK regular contributor and the founder and principal analyst with ZK Research. He spent 10 years at Yankee Group and prior to that held a number of corporate IT positions. Kerravala is considered one of the top 10 IT analysts in the world by Apollo Research, which evaluated 3,960 technology analysts and their individual press coverage metrics.
