Infrastructure that’s built for AI
Why AI is forcing a rethink of virtualization
Companies are seeking more adaptable ways to run applications, govern data, and control costs as AI workloads grow.
AI’s voracious compute appetite is putting a physical and financial strain on the systems that support it. Research forecasts that U.S. data center power demand will more than double in 2027, driven by the buildout of AI infrastructure. Companies ramping up enterprise AI workloads must now contend with the time, cost, performance, and power demands of running them. This is driving a rethink of virtualization, the technology that allows multiple applications and operating environments to use shared computing resources. This industry-wide reassessment is seeing businesses seek more adaptable, open models that offer greater control and reduce vendor lock-in. “Traditional virtualization stacks were designed for a different era,” says Brian Gruttadauria, chief technology officer of hybrid cloud at Hewlett Packard Enterprise (HPE), a leader in essential enterprise technology. “Enterprises are now optimizing for hybrid cloud operations, AI-ready infrastructure, and application portability, creating a broad overhaul focused on flexibility, economics, and modernization.”
The ambition, however, is running well ahead of organizations’ ability to act on it. Recent HPE research shows that more than two-thirds of businesses are planning material changes to their virtualization strategies within the next two years, while just 5% say they are fully prepared to make the transition.
5%
For years, virtualization was an efficient way to run business software on shared computing power. As companies expand their uses of AI, that model is being stretched. Effective AI training and inference require intensive graphics processing, rapid data movement, high-speed networks, low-latency storage, and the ability to draw on additional server capacity as demand rises. If architectures built for a steadier operating environment cannot handle this intensity, storage or networking bottlenecks can leave costly processors idling, while poorly placed data can slow AI jobs and raise running costs.
Those technical pressures are forcing IT leaders to rethink the economics of their current setups. Companies must weigh the expense of maintaining existing arrangements against the investment, planning, and disruption involved in modernizing them. HPE’s survey, however, suggests that licensing fees alone are rarely what prompts that reassessment, with just 4% of respondents identifying cost as the primary driver of change. More often, organizations are looking for greater control over performance, placement, and future technology choices. For many, the challenge is how to secure those advantages without adding financial burden, complexity, or risk—a question that is pushing hybrid cloud higher up the modernization agenda.
The case for change
Despite its merits, hybrid cloud modernization can be a slow, exacting process. Budget constraints are the principal obstacle, cited by 28% of IT leaders, followed by technical difficulty at 24%, migration risk at 21%, and skills gaps at 20%.
A staged approach to avoid disruption
The top barriers slowing virtualization progress:
Skills gap
Migration risk
Technical complexity
Budget constraints
20%
21%
24%
28%
To overcome these barriers, organizations need a staged approach. A move toward hybrid cloud cannot be treated as a one-for-one hypervisor replacement as it requires leaders to decide which applications should stay virtualized, which might move to private cloud or containers, and which require dedicated AI systems. It also calls for teams to standardize operations, improve observability, and build the expertise to manage a more varied environment without adding confusion. Specialized hybrid cloud platforms, including HPE GreenLake, HPE Private Cloud Solutions, and HPE Morpheus Software, can help by giving businesses a common layer for operating virtualized systems across a more complex IT estate. That means modernization can proceed workload by workload, with clearer control over migration, management, and cost, rather than through a disruptive all-at-once rebuild.
As AI takes a larger role in enterprise strategy, infrastructure choices are increasingly tied to broader business goals. Companies planning a reset need to know whether the systems they choose today will support future AI use cases, without locking them into decisions that are hard to reverse.
Securing the competitive advantage
Modernization must therefore create room to maneuver. A more integrated hybrid approach does that by helping organizations adapt as their AI needs evolve, instead of forcing each new demand through the limits of an older operating model. “Business conditions and technology requirements are changing too quickly for static infrastructure strategies,” says Gruttadauria. “Organizations need to be able to adapt continuously rather than redesigning their infrastructures every time business priorities evolve. That flexibility is now the competitive advantage.”
“Organizations need to be able to adapt continuously
—Brian Gruttadauria, Chief Technology Officer of Hybrid Cloud, HPE
BACK TO HOME
“Hybrid cloud gives organizations operational flexibility instead of infrastructure rigidity,” says Gruttadauria. “It allows them to place workloads where they make the most sense, based on performance, compliance, economics, or data gravity.” HPE’s research gives a sense of the capabilities IT leaders consider essential to this approach. Among respondents, 70% rate unified backup and cyber recovery as very important or business critical, while 61% say the same of cross-platform governance, and 55% cite integrated observability and AI operations. Together, those priorities reflect the need for visibility, protection, and oversight wherever critical data and software are located.
Platforms designed to simplify hybrid operations can help provide that consistency, allowing companies to modernize their virtualization strategies while retaining greater choice over where their most performance-intensive uses of AI take place.
Hybrid cloud offers companies a way to manage IT resources across on-premises systems, private and public clouds, edge locations, and dedicated AI infrastructure while retaining greater control over where particular tasks are carried out. For an AI application drawing on sensitive corporate data, that could mean keeping information closely governed. For another, it could involve using public cloud scale or processing data close to its source at the point of use. This ability to choose is key because many businesses now need to run established enterprise software alongside newer AI deployments. Instead of treating private cloud, public cloud, edge, and AI systems as separate domains, technology teams can manage them through a more consistent operating model, placing each application where it best fits the business, technical, and regulatory demands around it.
Hybrid flexibility
Only 5% of enterprises are
fully ready for the great
virtualization reset.
time business priorities evolve. That flexibility is now
rather than redesigning their infrastructures every
the competitive advantage.”
BACK TO HOME
Powering Canada’s sovereign
AI ambitions
Watch now →
Why AI readiness starts with the network
Watch now →
Why AI success starts with infrastructure
Read more →
More insights from HPE
