In association with Datapac
In 2026, infrastructure teams are facing a unique convergence of pressures. Senior leadership is asking for credible AI roadmaps and digital acceleration, while finance teams are requiring tighter cost discipline in response to changing infrastructure economics.
At the same time, global demand driven by AI workloads is placing upwards pressure on the cost of core infrastructure components such as RAM and SSDs. In parallel, virtualisation licensing structures have shifted, fundamentally changing baseline renewal costs across many estates.
These challenges cannot be viewed in isolation. They are interconnected and reflect structural shifts in the economics of infrastructure. For most organisations, maintaining the status quo now absorbs more capital while delivering no additional capability.
Right-sizing in a volatile market
In this environment, precision, not simple cost reduction, must guide decision-making. A common issue is historic sizing assumptions carried forward into a different economic climate. Infrastructure deployed during the last refresh cycle was optimised for its time. However, the landscape has changed so fundamentally that the very framework with which decisions are made needs to be updated, new questions asked, and new answers sought.
Workload consolidation, capacity right sizing and architectural density now matter more than ever. Platforms such as HPE ProLiant Gen12 enable higher core density and improved workload efficiency, reducing the physical footprint required to support the same or greater demand. Where licensing exposure is linked to socket count, this can influence long-term costs.
In addition to the physical data centre, hypervisor strategy forms part of this evaluation. Alternatives such as HPE VM Essentials introduce different licensing structures, including per-socket licensing versus the per-core model, which can quickly rebalance cost exposure in high core-count environments. When working with customers, we recommend modelling these options proactively months in advance of the renewal window, as this is always more effective than reacting to renewal shock.
Across the entire infrastructure stack, physical and virtual, the common thread is optimisation. Avoiding overprovisioning is just as important as identifying consolidation opportunity when both licensing and component costs are under upward pressure.
AI readiness: capability and clarity
Alongside these economic pressures, AI introduces another layer of infrastructure complexity. First, there is the infrastructure question. Most estates today were architected before AI became a boardroom discussion and as such were designed for CPU-bound business applications. AI workloads introduce a materially different demand profile, requiring infrastructure built on GPU-first foundations to cope with new demands, including model training, inference pipelines or larger-scale data processing. This means that refresh decisions made today should account for architectural flexibility to meet these new demands, even where immediate deployment is not planned.
Secondly, there is the value definition question around AI, namely “what will it actually do for us?”. Infrastructure capability alone won’t miraculously create business value with AI. Value comes from careful adoption and integration of AI into existing business processes and workflows, and the ‘correct’ answer will vary between organisations, yet most are unsure of what questions need to be asked. Without structured evaluation, AI experimentation can quickly consume capital without producing measurable return.
Through structured infrastructure and AI discovery engagements, Datapac works with organisations to assess architectural readiness, model consolidation and licensing exposure, and define viable AI use cases before additional capital is committed. This produces two valuable outcomes: where opportunity exists, it provides a roadmap to real business integration, and where organisations aren’t quite ready, they find this out as early as possible, preventing misallocation of capital.
Get answers to the right questions
Licensing reform, component volatility and AI ambition are converging, and treating any one of them in isolation will limit visibility and options.
If you are responsible for infrastructure strategy, this is the point to assess consolidation potential, licensing exposure and AI readiness together.
Datapac works directly with infrastructure teams to model hypervisor strategy, evaluate modern platforms, and define viable AI pathways before capital is committed.
Engage with us early so you can answer questions with confidence.
info@datapac.com | 01 426 3555 | www.datapac.com/infrastructureassessment/
Gavin Downey is infrastructure lead technical architect at Datapac
Ltd
