AI data strategy for enterprise value: PwC

Dan Priest

Chief AI Officer, PwC US

Email

Rima Safari

Principal and Data Engineering and Analytics Practice Team Leader, PwC US

Email

Rajeev Krishnan

Managing Director and Offering Leader, PwC US

Email

Key takeaways:

You don’t need to make your entire data environment “AI ready” before creating value
Leading organizations are targeting high-stakes business decisions first, using focused data unification and AI-enabled workflows to drive faster outcomes.
AI and data strategies create more value together than separately
Organizations can combine quick-win AI deployments with governed, reusable data products that strengthen enterprise capabilities over time.
A shared execution layer can accelerate scale while reducing complexity
Reusable AI agents, workflows, connectors, and governance controls can help organizations modernize data, improve consistency, and compound value across future AI initiatives.

If AI is on your agenda, then data is likely top of mind: How to give AI, quickly and cost effectively, the right data to deliver your business priorities. But most companies today are falling into one of two traps. Some aim to modernize data first, then add AI. But that can keep you waiting months or even years, and it typically produces data sets that may be modern, but don’t match business needs. Others try to run AI fast, then patch data. That may produce isolated small wins. But they almost never compound.

There’s a better way: Build a data-for-AI and AI-for-data “flywheel” that avoids both traps as you advance on two fronts at once. Starting with a few high-stakes areas, you give each AI use case the precise data it needs to deliver business value. At the same time, in the background, AI continuously transforms data for enterprise enablement too. With reusable assets, each new deployment builds on what came before. Delivery keeps accelerating. Your data environment becomes self-improving. And you can spot new ways to rationalize systems and cut technical debt.

The result can be quick wins that deliver big value and a sustainable foundation for high-impact AI strategy. According to PwC’s 2026 AI Performance Survey, the top AI performers (20% of the survey sample) capture nearly three quarters (74%) of AI-driven value. What sets winners apart is often data. If you’re a CEO, CIO, or another senior leader, here’s what you need to know.

Fast results and data products at scale: How to invest in data for AI

Two investment logics dominate AI, but there’s no need to choose. You can bet on both. For high-value use cases, you quickly unify the data that AI needs. At the same time, you gradually build a platform of scalable, governed data products. To drive lower costs and faster outcomes, both approaches share an activation layer. Over time, the paths converge. You get past vendor dependency and own the key parts.

It starts with a deliberate strategy for investing in two approaches, side by side.

Unification: Enable high-stakes, cross-domain decisions

For cross-domain decisions with an enterprise impact, such as how to grow revenue across business lines, address risk across functions, or manage supply chains, financial closes, or post-merger integrations, AI can change the game.

If you select the right decisions, you can identify the precise data needed and unify it across domains through a shared ontology (a common definition of business concepts and their relationships) and semantic layer that applies these definitions consistently across systems. You can then have more up-to-date insights, reliable forecasts, and realistic scenario modeling for complex decisions. AI can automate (under human oversight) simpler ones.

Benefits can be vast. At one global manufacturer we assisted with AI decision intelligence for supply chain management, first year savings from reduced on-time-in-full (“OTIF”) fees have been roughly US$700 million. These and other benefits can keep growing and scaling. But it’s important to make the right decisions:

The stakes are high. Your first unification bets may be high cost and high touch. For attractive ROI, start with complex, cross-domain decisions where faster, better choices could really move the needle for revenue, costs, risk, or strategy.
You can scope precisely. If you have to do too much data work up front, you may get bogged down. Focus on decisions where you can precisely select and quickly transform the specific data sets needed to drive business value.
You can rinse and repeat. If you can reuse the patterns for one decision for other, similar cross-domain priorities, you can create a reusable capability: Each new decision intelligence can come faster and be cheaper than the last.

Platform: Enable enterprise-wide reuse and control

How to get scalable, dependable, “industrial-grade” AI? It starts with governed, reusable data products that have clear ownership, lineage, security, and standards. But creating these products can be a lot of work: reconciling data models and master data, and reskilling people.

But modern tools, powered by agents, can help. You can move domain by domain, creating value at each step. You can usually use your current hyperscaler platform. And the more you do, the better the economics can get, as more teams and agents consume the same data products.

At one global technology company, we helped stand up reusable data ingestion pipelines, curated data layers, reporting foundations, and a common data model with built-in governance. In eight months, the company cut its legacy table footprint 73% and rolled out over 500 gold-layer reporting tables. With reusable tools and patterns, this firm’s data modernization is still accelerating. But with the “platform bet” too, it's wise to start where certain conditions are met:

The same data serves many teams. If you start where the path to scale is both clear and clearly valuable, your platform can create more value more quickly.
Trust, compliance, and consistency are vital. If mediocre data is good enough for a set of use cases, leave those for now, Instead, start where robust and clear governance, lineage, and ownership can make new value possible.
The economics work. In some domains, the cost / benefit ratio can improve especially quickly as more teams (and AI agents) use the same data products.

Activation layer: Power both investment paths on your existing systems

A single, reusable activation layer can power both your unification and platform bets, improving consistency and lowering the cost of scale. It contains AI agents, modular skills, connectors, and human-in-the-loop gates, and it offers major benefits:

Connect new and old systems. Thanks to agents, the activation layer can connect and integrate data sets and workflows across new systems and the ones you already have, even if they weren’t originally designed to work together.
Centralize and customize governance. As part of AI-ready architecture, the activation layer holds your chosen governance and controls, so you can grow stakeholder trust.
Orchestrate and create. If your activation layer includes a suitable platform and tools, you can orchestrate AI across your enterprise, design new workflows, and create new agents for data modernization as needed.

Since this layer can work on existing systems, its assets are reusable, and its agents can help with data quality, lineage, migration, and workflow execution, ROI can come quickly. At one national retailer we assisted, agents now unify fragmented signals from store operations, conduct root-cause analysis, and recommend actions for district and store leaders. The company estimates over $20 million in annual value, with indirect benefits (such as time savings for senior personnel) adding up too.

Turn AI and data into ever-growing value: 4 steps to take today

When you combine a dual investment model with a shared activation layer, benefits compound. At one global healthcare company we assisted, the unification bet connected previously silo-ed trial and patient data to drive better care decisions and research. With the activation layer’s help, the platform bet then scaled that same data into governed assets that over 50 domains are now using. Care teams can now get critical analytics in half the time, enabling better patient outcomes. The company is expecting new value of over $50 million annually.

4 steps can get you started on this path of compounding AI and data value:

Set your ambitions and align them with your present state. Gather stakeholders to align on goals for AI. Assess these goals in light of where your data, systems, and governance stand today.
Start with high-value outcomes and key decisions. Identify a few business outcomes where better, AI-enhanced decisions could be gamechangers. Break down each outcome into their key decisions and supporting data domains.
Route each decision to the right bet. Using the criteria above, such as value and reusability, assign each decision to the unification bet, the platform bet, or both.
Act fast and in parallel. With the help of agentic activation, you can quickly (often in 60-90 days, depending on data readiness and decision scope) launch at least one unification workflow and one platform domain.

Together, these four steps can start the AI data flywheel spinning: As unification delivers early wins and momentum, your platform strengthens your data foundation. Governed data products spread. Costs drop. Decision speed and quality rise. New AI-enhanced decisions and data, with your chosen governance built in, come faster and faster. Soon, you won’t just have a data platform. You can have a decision factory in which you own the control plane and value keeps compounding. Your strategic unit can then shift from functions to decisions which cross functions.

This AI data flywheel is not only possible, it’s the necessary foundation for bi-modal transformation and an agentic enterprise, and you can start building it today.