With regulators and boards paying nearer consideration to the place delicate information sits, Fred Lherault, Area CTO EMEA/Rising Markets at Pure Storage, outlines why hybrid methods and selective cloud repatriation are prone to speed up as AI scales.
After two years of accelerated AI experimentation, rising expectations, and speedy vendor growth, I consider 2026 will mark an vital inflection level for organisations constructing fashionable information infrastructure. Many enterprises at the moment are shifting previous the preliminary hype cycle and specializing in what’s required to operationalise AI reliably and at scale.
That shift is already seen throughout clients evaluating how AI will combine into manufacturing workflows. If we extrapolate from these traits, a number of themes are prone to affect how organisations design their information pipelines, storage architectures, and cloud methods within the yr forward. The next displays my perspective on how these dynamics might unfold.
From hype to manufacturing: information readiness and inference change into the precedence
Whereas some organisations are nonetheless convincing themselves how important AI is, most at the moment are reasonable about what they do, and, crucially, don’t deploy. The change in focus from coaching to inference implies that, with no strong inference platform, and the flexibility to get information prepared for AI pipelines, organisations are set to fail.
As AI inference workloads change into a part of the manufacturing workflow, organisations should guarantee their infrastructure helps not simply quick entry, but in addition excessive availability, safety, and non-disruptive operations. Not doing this will probably be expensive, each from a outcomes perspective, and an operational one.
Nonetheless, most organisations are nonetheless battling the info readiness problem. Getting information AI-ready requires going by way of many phases, comparable to information ingestion, curation, transformation, vectorisation, indexing, and serving. Every of those phases can usually take days or perhaps weeks, and delay the purpose when the AI undertaking’s outcomes could be evaluated by the enterprise.
Organisations who care about utilizing AI with their very own information will deal with streamlining and automating the entire information pipeline for AI – not only for sooner preliminary outcomes analysis, but in addition for steady ingestion of newly created information, and iteration.
This stays some of the vital boundaries to AI adoption. Enterprise information is usually dispersed throughout legacy programs, cloud environments, and archives, which makes it troublesome to entry and put together on the velocity AI workflows require. In 2026, we will anticipate this problem to change into extra pronounced as organisations look to extract worth from all of their information, no matter location. Handbook preparation is not going to scale to satisfy these necessities. Automated pipelines, richer metadata, and built-in information platforms will change into important foundations for organisations aiming to make use of AI with steady, repeatable outcomes.
AI and information sovereignty will reshape cloud technique, and speed up selective repatriation
The twin problems with AI and information sovereignty are driving issues about the place information is saved, and the way organisations can preserve belief, and assure entry within the occasion of any points. As a way to extract worth from AI, it’s essential for organisations to know the place their most vital information is, and that it’s prepared to be used.
Issues about information sovereignty are additionally driving extra organisations to rethink their cloud technique. Rising geopolitical tensions and regulatory strain will form nations’ information centre methods in 2026 in response. Governments, specifically, need to minimise the danger that entry to information may very well be used as a menace or negotiating tactic. Organisations must be equally cautious, and put together themselves.
We’re already seeing early indicators of this shift. Boards and regulators are paying nearer consideration to the place delicate and strategically vital information resides, pushed, partially, by evolving regulatory frameworks comparable to GDPR, DORA, and steerage rising from the EU AI Act. This scrutiny is prompting many organisations to reassess cloud methods that after prioritised price or comfort over sovereignty and resilience.
In consequence, hybrid fashions are prone to increase, with extra AI-critical datasets and workloads positioned nearer to the place they are often ruled, audited, and managed. This isn’t a retreat from the cloud, however a extra deliberate, workload-specific leveraging of it.
KubeVirt will scale into mainstream manufacturing
The latest adjustments to VMware licensing that adopted Broadcom’s acquisition have kickstarted a dialog round different approaches to virtualised workloads. KubeVirt, which permits administration of digital machines by way of Kubernetes, offers one such different—a platform that encompasses each virtualisation and containerisation wants—and I anticipate it would take off in 2026.
The KubeVirt providing has matured to the purpose the place it’s appropriate for enterprise wants. For a lot of, shifting to a different virtualisation supplier is a large upheaval, and, whereas it might ultimately lower your expenses, it all the time comes with a set of limitations and constraints, particularly relating to all the pieces that surrounds the virtualisation platform (information safety, safety, networking, and so forth).
KubeVirt permits organisations to leverage the rising Kubernetes ecosystem to extra shortly realise the worth in a platform which offers the capabilities to handle, orchestrate, and monitor not simply VMs, but in addition containers, no matter how the proportion of these evolves over time.
KubeVirt’s momentum displays a broader shift in how organisations need to function their infrastructure. As containerisation turns into customary and AI workloads scale, many groups are searching for a unified operational mannequin that reduces complexity, and avoids long-term platform lock-in. Consolidating digital machines and containers underneath a single management aircraft aligns with this course.
If adoption will increase as predicted, storage and information providers will evolve in parallel, with better demand for persistent, low-latency, Kubernetes-native storage that may help mixed-workload environments.
2026 will probably be about self-discipline, not disruption
If the previous two years have been outlined by speedy disruption, pushed largely by AI, 2026 is prone to be a yr the place organisations prioritise the operational basis required for long-term success. Enterprises will:
- Transfer from AI experimentation to constant, production-grade inference fashions
- Modernise information pipelines to help steady information readiness
- Reassess cloud methods with a sharper deal with sovereignty, governance, and resilience
- Consider VMware alternate options, comparable to KubeVirt, which help a unified strategy to digital machines and containers
The organisations in a position to take these shifts of their stride will probably be greatest positioned for achievement in 2026.
This text is a part of our DCR Predicts 2026 collection. The collection will formally finish on Monday, February 2 with a particular bonus prediction.
