By Jim Takatsuka | October 2026
—
If you’re a CIO moving with intent on the cloud, the situation often looks like this: most of the infrastructure still runs on-premises, a couple of SaaS applications are in use without much strategy behind them, and there’s a mandate to move faster and show enterprise value soon. You also know the bill is coming, and it will be appreciable.
You’re not asking whether to move, or whether to phase it. You know both. The question worth sitting with is what should lead. And for a move you want to accelerate, I’d argue the answer is data, for reasons that aren’t obvious at first.
Three Tracks Moving at Different Speeds
It helps to stop thinking of this as one program and start thinking of it as three tracks that move at different speeds and pay off on different timelines.
The first is data: getting it reachable by cloud services and getting governance and lineage right. This isn’t just moving bytes, it’s making data trustworthy and accessible enough that other things can be built on it. It’s the track everything else compounds on.
The second is infrastructure: migrating and modernizing what you actually run. This is necessary, but it’s the slowest to show business value, and it’s the place where “we moved the data center” gets mistaken for progress the business can feel.
The third is SaaS, the buy-versus-build decisions like CRM. These are the fastest visible wins, and also silos-in-waiting if adopted without a strategy.
All three happen. The question is what leads and what follows.
Why Data Leads
Data is the only one of the three that compounds. A CRM solves one problem and then plateaus, that value is real but it’s bounded. A reachable, governed data foundation makes every later capability cheaper, analytics, AI, and the SaaS applications themselves. Work you do here keeps paying off across everything that comes after.
It’s also what unlocks AI, and AI is where the enterprise-value story lives right now. Moving your data to where cloud services can reach it opens the capability set your board is genuinely excited about. That’s a bigger, more durable story than replacing a tool, and it’s the one most likely to justify the spend you know is coming.
There’s a sequencing reason too. SaaS built on ungoverned data inherits the mess. Stand up a CRM before the data foundation exists and it becomes one more silo, its own copy of customer data, its own access model, its own gap in lineage. Do the data work first and every application after it plugs into a governed foundation instead of spawning another island.
And data has the longest lead time of the three. SaaS can be stood up quickly whenever you choose. The foundation can’t be rushed the same way, so starting it late is what hurts most. It’s the track that has to start now precisely because it’s the one that can’t be compressed later.
The Catch: Data-First Done Wrong Is Invisible
Here’s the honest failure mode. Months of pipeline and governance work with no user-facing outcome is its own trap. You trade one problem for another, the business sees nothing for a long stretch and patience runs out right when you need it. Data-first only works if it’s anchored to something people can see.
The Fix: Scope Data to a Use Case
Don’t build the foundation speculatively. Pick the first high-value decision or capability that matters, often something AI-enabled, and build the slice of data foundation that use case needs. The same effort delivers a visible win and a compounding foundation at once.
Choosing that first use case is most of the work. Look for high business value, data you can realistically reach in months rather than years, and an owner who actually wants the outcome. Get those three and you have a project that proves the approach and leaves real foundation behind.
On SaaS: Adopt With a Strategy, Not by Accident
Since the starting point here is SaaS without a strategy, this deserves real attention.
Decide buy-versus-build deliberately. Buy for undifferentiated capability, CRM, ITSM, HR, where a vendor does it better than you ever will. Build only where you have genuine edge worth owning.
Get identity and access right from the first application, before sprawl sets the pattern. One identity model across your SaaS estate, not one per vendor, is far easier to establish early than to retrofit.
Treat every SaaS application as a data source that feeds the governed foundation, with integration and data ownership settled up front. The alternative is a standalone island whose data you promise to “deal with later,” and later rarely comes.
Go in clear-eyed on cost, too. SaaS shifts spend from capital to recurring subscription, which changes the shape of the budget your board is approving. That’s not a reason to avoid it, just a thing to name before the renewals arrive.
Managing the Cost Narrative
The spend is real and appreciable, and you know it, so the move isn’t to soften it but to pair it with visible outcomes on a timeline the board can watch. The use-case-scoped data work gives you an early win that buys patience for the slower infrastructure spend behind it.
Sequence the money so each phase has a demonstrable outcome attached, rather than a long quiet stretch followed by a promise. And be transparent that foundation investment is what makes later spend smaller, the early cost is buying leverage, not just capability. That framing tends to land better with a board than either extreme, boiling the ocean on one side, death by endless pilots on the other.
A Place to Start
Map your initiatives on two axes: time-to-visible-value and cost-of-inaction. Start the data foundation now regardless, but scope it to the one use case that scores high on both. Let SaaS fill the fast-win slots as it feeds the foundation, and let infrastructure migration follow where real cost or risk drives it, not the urge to finish the move for its own sake.
Getting your data reachable and governed isn’t only AI preparation. It’s the move that should lead an accelerating cloud strategy. Sequence it so the foundation and the first visible win come from the same effort, and the wins arrive before the doubts do.
If You Want to Go Deeper
A few things worth reading if this resonates:
- A Leader’s Guide to Cloud Transformation (AWS Executive Insights). Enterprise strategists on navigating a cloud move, with an emphasis on trusting your people and adapting culture, not just the tech.
- Cloud Migration Strategy: A Modern Approach for Enterprise Teams (AWS Executive Insights). A conversation on migration lessons: start promptly, keep it simple, begin with low-risk workloads, and treat speed as a way to reduce risk.
- Debunking Seven Common Myths About Cloud (McKinsey). A useful reframe, starting with the myth that cloud value is mostly about IT cost reduction rather than business capability.
- Five Learnings From CTOs and Tech Leaders on Their Cloud Strategies (McKinsey). Peer perspective on the business-case, architecture, and operating-model questions a cloud strategy has to answer.
—
Jim Takatsuka works with Hawaii’s enterprise organizations on AI and cloud strategy. Reach him at jimtak@amazon.com.
