If you're still in a data centre, AI just made it more urgent. Not less.
A Clarity on Cloud blog by our CEO, Helen Gerling
AI didn't change the case for cloud migration. It just stopped letting you ignore it.
Most weeks, I have a version of the same conversation.
Someone shows me a roadmap. The cloud migration is on it. It has been on it since 2019.
They aren't being negligent. They're being rational. The reasons to delay always felt good: capital projects, change fatigue, "if it isn't broken", a wait-and-see on costs.
Quietly, while we weren't looking, the bill for waiting changed.
The cost of staying still has gone up
It isn't just AI, although AI is the loudest signal.
Copilot, agentic workloads and intelligent process automation all need data that's governed, classified and reachable. Fabric pipelines and on-prem gateways can reach a 2012 file server, and many organisations are doing exactly that.
What they can't reach is the DC bill, the patching cycle, or the permissions mess at the source.
The pipeline is a bridge. The cost is paid twice.
And it isn't only AI.
Cyber insurance premiums are rewarding cloud-native posture and penalising legacy estates. Regulators are tightening their grip on both sides of the Channel: the UK's Cyber Security and Resilience Bill, NIS2 across the EU, and the NCSC's Cyber Assessment Framework all expect ongoing assurance, not annual attestations from a server room.
Microsoft's own roadmap is quietly retiring features that don't make it off-prem. And nobody is queuing up to maintain a Windows Server 2012 farm in 2026, which means the talent gap closes itself by attrition.
Each of these pressures is manageable in isolation. Together, they've turned "we'll get to it" into a slow, compounding tax.
The lift-and-shift trap
When laggards finally do move, the temptation is to lift-and-shift and modernise later.
Later rarely comes.
What you've actually done is buy yourself a more expensive version of the estate you started with. The same data sprawl, the same broken permissions, the same shadow integrations, now on a metered bill.
The migration was the expensive part. The value was always going to come from what came after. If "after" never starts, the migration was a cost, not an investment.
What good looks like in 2026
The shape of a good cloud migration has changed. It isn't about getting the workloads off the floor anymore. It's about landing them somewhere that earns its keep.
That means:
Data first. If data isn't governed, classified, and reachable, nothing built on top of it will be either. Including AI.
PaaS where you can, IaaS where you must. Lift-and-shift is a tactic, not a strategy.
Identity hygiene as a prerequisite. (I've written about why elsewhere.) You don't migrate good identity practice. You have to build it.
Governance baked in, not bolted on. Tagging, policy, cost controls and access reviews from day one, because adding them later costs three times as much.
The leadership challenge
If you're a CIO, CTO, or exec accountable for a cloud strategy that's been "in progress" for a while, three things to stop doing:
Stop measuring migration in VMs moved. Start measuring it in workloads modern tooling can reach.
Stop treating "lift-and-shift first" as a strategy. It's a tactic, and a costly one if it's the whole plan.
Stop assuming the cost of waiting is zero. It isn't, and it's compounding.
Final thought
The migration question used to be "should we?". Then it became "when?".
In 2026, it's "how do we do this without recreating the same problems on a more expensive bill?"
The data centre exit isn't the goal. It's the doorway.
What matters is what you walk into.
Cloud migration is no longer a technology milestone. It’s a business readiness question - for AI, cyber resilience, governance and operational agility.
Shaping Cloud works with organisations to move beyond lift-and-shift thinking, helping leadership teams modernise platforms, improve governance and build cloud foundations that support measurable business outcomes.
Book a call to discuss how your organisation can modernise legacy estates and build a cloud strategy designed for long-term resilience, scalability and AI readiness.