Part of The State of Data and AI series—featuring insights from the Global Data Insights Survey 2026 (GDIS)

Artificial intelligence (AI) has entered its economics era. The shift from strategy to execution is well underway, and the payoff is real. But not every organization is positioned to capture it. The infrastructure beneath AI has become the gatekeeper between promise and payoff, and the organizations that recognize this are the ones pulling ahead.

IDC forecasts that AI will generate $22.5 trillion in cumulative global economic impact by 2031. The enterprises seeing the clearest gains today share one defining trait. They’ve moved from buying access to AI to owning it, controlling their models, data, and infrastructure on dedicated, private AI environments built for scale. The next phase of AI isn’t about access. It’s about control.

Digital Realty’s 2026 Global Data Insights Survey (GDIS) tracks exactly where the gap sits. American enterprises are out front. A full 41% have already achieved measurable business outcomes from AI, including revenue growth and cost reduction. That compares with 36% in Asia-Pacific (APAC) and 28% in Europe, the Middle East, and Africa (EMEA).

Look at near-term returns, and the picture sharpens further. Only 3% of enterprise leaders report seeing no measurable return on investment from their AI initiatives, and 63% expect to see ROI within the next six months to two years. IDC estimates the total economic impact of AI in the Americas will reach $14.1 trillion by 2031. That's more than three times the projected impact of either EMEA or APAC individually.

The question worth asking isn’t just why. It’s what leaders everywhere can do about it.

Advantage the Americas: What’s driving the ROI gap

The Americas advantage didn’t arrive by accident. It doesn’t come down to a single policy, program, or market condition. The GDIS points to a broader pattern that is harder to shortcut.

Enterprises in the region are simply more likely to have access to the infrastructure foundation that converts AI pilots into sustained, repeatable business outcomes. That foundation has three pillars, and together, they define what private AI at scale requires:

The ability to place data and workloads where they can perform securely.
The capacity to scale high-density AI environments.
The interconnection needed to move data across clouds, regions, and partners without introducing complexity at every step.

Get all three right and AI stops being a project. It starts to become a competitive capability.

Scale and infrastructure depth give organizations more ways to win

The Americas operates from a large, mature digital infrastructure ecosystem built up over decades. AI workloads demand far more than raw compute. They require coordinated access to capacity, connectivity, power planning, cooling, security, and operational expertise working in concert.

It’s worth being direct about one important nuance here. Power availability, permitting, and land use are real constraints across the Americas. Major hubs like Northern Virginia are already facing government and utility pressures that shape deployment timelines. Tax incentives and economic development programs influence where investment flows. They don’t, by themselves, solve power procurement or grid capacity challenges.

The real story is more useful. The Americas offer significant digital infrastructure depth alongside significant demand. That gives enterprises more viable deployment options: established high-density campuses, distributed regional architecture, and emerging AI-ready metros. Constraints don’t disappear, but there are more options for designing intelligently around them.

Power has become a board-level question

Energy availability used to be an operational concern. AI has made it a strategic one.

As organizations move from pilots into scaled inference and production workloads, securing the right power profile becomes critical to cost, performance, and deployment timelines. The organizations planning this early are the ones avoiding bottlenecks that derail AI programs at the worst possible moment.

The strategic question has shifted. It’s no longer simply where compute can be hosted. It’s where data, power, cooling, connectivity, and operational resilience come together. The goal is sustained AI growth well into the future, not just quarter-to-quarter. For enterprises in markets where power is constrained, the priority is partnering with providers that can plan across multiple metros and avoid single points of failure.

Our Americas platform offers dense interconnection hubs, emerging AI-ready markets, and established campuses built for varied workload profiles at scale. The advantage isn’t immunity from constraints. It’s the range and expertise to work around them at pace.

The infrastructure layer that makes AI work

A lot of AI conversation focuses on models, applications, and AI fabric. Those layers matter. They sit on top of a physical and digital foundation that either supports them or doesn’t. Without that foundation, AI programs stall, not because the model is wrong, but because the digital infrastructure can’t carry the load.

For AI investments to produce repeatable outcomes, enterprises need three capabilities working together. This is what our global data center platform, PlatformDIGITAL®, was built for:

High-density capacity to support AI workloads with the performance, cooling, and resilience they require. As GPU-intensive inference loads grow, the data center environment must be purpose-built to handle the demands of production AI. High-Density Colocation, up to 150 kW per cabinet, from a single rack to full data halls, is the entry point for enterprise AI at scale. It’s no longer a niche requirement.
Private interconnection to make data and workloads usable across clouds, regions, partners, and providers. Open, carrier-neutral connectivity allows AI to operate on real enterprise data rather than siloed samples. It also lets organizations reach multiple cloud providers and data sources without rebuilding complexity at every boundary.
Geographic distribution to keep data close to users, applications, and regulatory requirements. Distribution isn’t just a sovereignty play. It’s a performance and latency strategy as AI inference moves closer to the business processes that depend on it.

Organizations winning today built them in from the start, using platforms like ServiceFabric®. ServiceFabric brings together cloud providers, networks, and digital infrastructure into a single programmable platform built for this kind of scale. Its dense, on-demand connectivity lets enterprises reach multiple clouds, networks, and data sources without the cost and complexity of provisioning dedicated circuits at every boundary. Because that interconnection is available across a globally distributed footprint of metros and campuses, organizations can place AI workloads and data exactly where performance, latency, and sovereignty demand, then shift them as those requirements evolve.

Geopolitics is actively reshaping where AI value gets created

Geopolitical dynamics aren’t a background variable in the AI ROI story, rather an active force shaping infrastructure decisions in every region, and they’re accelerating.

Data sovereignty is the clearest example. The 2026 Global Data Insights Survey shows that 86% of enterprise leaders worldwide are either exploring or actively implementing sovereign AI options, and over half are already hosting their AI models in private clouds. For the Americas, that cuts both ways. It reinforces the region’s advantage in local data control, while also raising the stakes for maintaining distributed infrastructure across a large and diverse set of markets.

Advanced AI chip controls add another layer of complexity. Infrastructure decisions are increasingly being made based on where capacity, hardware, data, and compliance requirements can align. Organizations with platform access and digital infrastructure relationships needed to adapt their sourcing strategy are far better positioned to absorb these disruptions.

The rise of distributed AI makes all this more urgent. As inference moves into regional deployments closer to the business processes that use the outputs, the need for geographic distribution and open interconnection intensifies. Unlocking the full value of distributed data requires infrastructure that can follow the data wherever business and compliance logic demand, without sacrificing performance or security.

This is one of the dynamics shaping regional AI adoption today. The Americas benefit from infrastructure depth that can support distributed AI architectures right now, while other regions continue to build toward that same opportunity.

The pattern of AI adoption is also following a familiar trajectory. Much like cloud computing before it, AI is gaining traction first in technology-intensive sectors such as financial services, healthcare and pharmaceuticals, and media and entertainment. While private AI adoption is often driven by the need for greater control over data and the rising costs associated with AI workloads, enterprises are increasingly finding value in hybrid architectures that balance private environments with access to external data, models, and innovation ecosystems. As organizations continue to develop the talent and expertise needed to execute AI strategies at scale, these hybrid approaches will help unlock new use cases and accelerate adoption across a broader range of industries.

Digital infrastructure AI-readiness is the deciding factor

The gap between organizations realizing AI returns and those still waiting comes down to one thing: readiness. For organizations still working to turn pilots into sustained, repeatable outcomes, the survey is clear about what’s holding them back: digital infrastructure bottlenecks, data storage challenges, and interconnection gaps. Not model quality. Not data science talent. Not executive commitment. In fact, only 3% of leaders report no measurable ROI at all, and 63% expect returns within the next six months to two years. The gap isn’t ambition. It’s readiness.

This does however present a solvable problem, but only for enterprises willing to be honest about where they stand.

AI ROI isn’t created by the model alone. It is created when data strategy, digital infrastructure strategy, and business strategy are designed together from the start. The organizations winning today have access to a foundation capable of supporting AI at scale and adapting as power, policy, and market dynamics continue to evolve.

For enterprise leaders ready to close the gap, the first step is an honest assessment. Can your digital infrastructure support the density, interconnection, and geographic distribution that production AI demands? Does your platform open access to every cloud and data source that matters, or does it lock you into a single vendor ecosystem? Do you have a partner with the global reach and operational depth to move quickly across every market you serve? See where your organization stands across the AI maturity scale.

We bring together the global data center platform, interconnection ecosystem, and operational expertise to help enterprises move from AI ambition to measurable outcomes. Whether the challenge is power planning, building a distributed architecture, or closing the gap between a promising pilot and production at scale, the foundation is already in place.

Explore the full findings from Digital Realty’s 2026 Global Data Insights Survey and see where the real opportunities to close the gap remain.

About the survey

The Global Data Insights Survey 2026 is Digital Realty's biennial research study, based on insights from 2,131 IT decision-makers across 19 countries and 11 industries. The 2026 report explores how organizations are moving from AI strategy to execution, with a focus on infrastructure, data governance, sovereignty, and measurable ROI. Its findings reveal four key trends shaping the AI economics era and the critical role of intelligent digital infrastructure in enabling AI success.

Read the full report to explore the data, insights, and recommendations driving the next phase of enterprise AI.

Digital Realty Innovation Lab São Paulo Expands AI Infrastructure Testing in Latin America