Data Gravity and Enterprise AI
AI is creating a new Data Gravity challenge and its a 3 body problem: data increasingly resists centralization, decisions require distributed context, and the cost of moving information keeps rising.
Data Gravity's Three-Body Problem
Why Centralization Is the Wrong Starting Point for AI
There is a long-standing assumption in enterprise architecture that Data Gravity is primarily a storage problem.
Data accumulates. Applications move closer to it. Organizations centralize information into data lakes, warehouses, or other repositories. Eventually, everything important lives in one place. Well, almost everything.
For traditional analytics, centralizing enough of the data often made sense.
But many organizations, particularly in regulated industries, have always lived with a different reality. Sensitive data is often restricted by security policies, data residency requirements, privacy regulations, classification levels, organizational ownership, or operational constraints. Some data was never meant to move in the first place.
AI changes the economics for everyone.
The challenge isn't that organizations have too much data. It's that AI requires access to more of it than ever before, while the forces preventing centralization are becoming stronger.
What used to be a straightforward architectural pattern has become something closer to a Three-Body Problem.
Organizations are now caught between three competing realities:
- The data won't centralize.
- The context won't stay in one place.
- The economics no longer justify moving everything.
Individually, each force can be managed.
Together, they are reshaping how enterprise AI gets built.
Problem #1: Data Resists Centralization
One assumption worth challenging is that enterprise data eventually ends up in one place.
In reality, most organizations are moving in the opposite direction.
Data is spreading across clouds, business units, operational systems, edge environments, partner ecosystems, and jurisdictions. Some of that data is protected by regulation. Some is governed by security policy. Some is controlled by organizations that have no incentive to share it.
This isn't a temporary problem. It's the operating reality of modern enterprises.
Many organizations respond by creating data lakes, warehouses, or clean rooms. These approaches can be useful, but they share the same underlying assumption: that the data can ultimately be centralized.
Increasingly, that's not true. The challenge isn't figuring out how to move more data. It's figuring out how to create intelligence across data that may never live in the same place.
Problem #2: AI Assisted Decisions Need Context
Organizations aren't investing in AI because they want more data. They're investing in AI because they want better outcomes, faster decisions, and improved operational performance.
The challenge is that decisions rarely depend on a single source of information. They depend on context.
- Operational signals.
- Historical information.
- Organizational knowledge.
- External intelligence.
The information needed to support important decisions has always been distributed. That's not new.
Banks, hospitals, governments, and large enterprises have all managed it.
What's changed is the expectation that AI can help make sense of it.
Context rarely lives in a single system. It exists across applications, business units, partners, operational environments, and organizational boundaries.
This is where Data Gravity becomes more than a storage problem.
The challenge isn't simply where data resides. It's whether organizations can assemble enough context to make, validate, and trust AI-assisted decisions.
So the result isn't simply distributed data. It's distributed decision context.
Organizations struggle to understand what information influenced a recommendation, what evidence was considered, and whether the decision can be explained, validated, or improved over time.
The challenge isn't simply accessing more information.It's creating trusted decisions when the context needed to support those decisions will always be distributed.
Problem #3: Time, Power, and Money
Even when organizations can centralize data, a third challenge emerges.
How long will it take?
And how much will it cost?
Most large-scale data centralization efforts aren't projects. They're programs.
The integrations continue.
The governance evolves.
The migration expands.
The architecture grows.
Meanwhile, AI innovation operates on a completely different timeline.
Models improve monthly.
Competitive advantages emerge quarterly.
Business leaders need outcomes today.
At the same time, moving data isn't free.
Every movement of data consumes resources. Data moves through memory, across networks, between systems, and increasingly between distributed environments. Those movements require infrastructure, power, and ongoing operational investment.
What organizations often underestimate is that the cost of centralization isn't just measured in storage.
- It's measured in time-to-value.
- It's measured in energy consumption.
- It's measured in the opportunity cost of waiting.
The assumption that data can always be moved to where intelligence lives is becoming increasingly expensive to maintain.
Data Gravity Has Changed
The traditional answer to Data Gravity was simple:
Centralize the data.
AI changes that equation.
Organizations need more data than ever before. Yet that data increasingly resists centralization. And even when centralization is possible, the time, power, and money required to achieve it continue to grow.
That's the Three-Body Problem.
The challenge isn't collecting more information.
The challenge is creating intelligence across information that remains distributed, governed, and owned by different people, systems, and organizations.
For years, the answer to Data Gravity was simple: move the data.
AI forces a different question.
What if the data isn't going to move?
The organizations that answer that question successfully won't just build better AI systems. They'll build AI systems that can operate in the real world.