Flexential’s annual report finds power shortages, network limitations, and distributed AI architectures are reshaping how organizations deploy AI
DENVER — Reliable grid power availability now plays a determining role in where 89% of organizations deploy their AI workloads, and 55% consider electricity pricing to be the most important consideration when choosing a region, according to findings from Flexential, a leading provider of secure and flexible data center solutions.
Those findings, captured in Flexential’s “2026 State of AI Infrastructure Report: AI-Ready Isn’t Infrastructure-Ready,” offer a comprehensive picture of how businesses are approaching AI deployment. Energy access has become the governing constraint as power availability, networking capacity, trade policy, and compute access converge into a connected set of limitations that most organizations can’t resolve independently.
The report also found that network performance issues affected 96% of respondents over the past 12 months, underscoring how constraints between facilities, instead of within them, have become a growing source of friction for AI deployments.
Even so, the anxiety that accompanied early AI adoption has largely subsided. The share of IT leaders who feel overwhelmed by AI implementation dropped from 29% last year to 11%, while those reporting nervousness fell from 13% to 3% and uncertainty from 10% to 2%. Although they’re contending with real limitations, they’re clear-eyed about the problems rather than rattled by them. In fact, 82% say they are excited by their organization’s AI initiatives — the highest level in the report’s history — and 73% describe themselves as inspired, up from roughly half in previous years.
Now in its third year, the report draws on a comprehensive survey of more than 350 IT decision-makers at the director level or above at companies with more than $100 million in annual revenue, including respondents from companies exceeding $2 billion. All had knowledge of their organization’s AI implementation and related infrastructure buildouts.
Their concerns over power extend well beyond site selection. Nearly three-quarters (72%) of those surveyed express moderate or extreme concern about electricity price volatility affecting their AI operating costs, and 42% identify power availability as the sustainability-related constraint posing the greatest long-term risk to AI infrastructure growth. When asked where the energy for AI will come from, 67% believe renewables will supply most of it within five years, with 42% expecting solar and wind to be paired with battery storage. Only 14% point to natural gas as the dominant future source, despite its growing popularity for on-site generation.
“Power determines what’s possible with AI right now. If it isn’t there to support compute, networking, and deployment timelines, nothing else matters,” said Ryan Mallory, CEO of Flexential. “Organizations that have the budget and talent are showing up with ambitious AI roadmaps and finding out that they can’t get megawatts in the market they need. Where power is available, and who can deliver it, has become as important as any technology decision these companies are making.”
The survey also found that:
- When asked to identify the greatest barrier to expanding AI initiatives, 40% of respondents pointed to IT infrastructure and just 1% cited budget. Amid persistentconversations about AI spending and whether the investment is worth it, these organizations have found that money isn’t the problem.
- U.S. tariffs and import restrictions have already reshaped procurement strategies, with 54% ofthose surveyedincreasing their reliance on domestic suppliers, even as most view it as a short-term adjustment rather than a long-term solution. At the same time, 40% are delaying or scaling back purchases, and 40% are signing longer contracts to manage exposure.
- Excessive latency, whichwas cited by just 32% of those surveyed in 2024, now impacts 71% as data transfer struggles to keep pace with increasingly distributed architectures.
- AI compute and AI data are migrating in opposite directions. GPU deployment in the public cloud rose from 30% in 2024 to 54% today,while thenumber of businesses housing AI data in the public cloud fell from 59% to 26%. That widening separation helps explain the latency and performance issues respondents are reporting.
- The share of organizations expecting measurable financial benefits within a year fell from 51% to 36%, while those already seeing returnsessentiallyheld flat at 20%. Cost reduction and operational efficiency (55%) replaced revenue growth (42%) as the primary measure of AI success, reflecting an environment where rising costs are reshaping how companies define value from their AI investments.
Together, these findings describe an environment where the barriers to AI growth have less to do with the technology and more to do with the infrastructure, supply chains, and policies surrounding it.
“Organizations used to be able to plan around power, networking, compute, and trade policy independently. Those days are over,” said Tom Bailey, Vice President of Energy at Flexential. “Today, a power shortage affects where you can deploy and your networking options, and when you layer on tariffs and procurement timelines, and you have to manage it all simultaneously, you’re in a considerably different environment than you were a year ago.”
That environment is changing how organizations evaluate infrastructure partners and what they expect from them.
“Customers are making multi-year infrastructure commitments, and they want to know that the capacity they need will be there when they need it,” said Patrick Doherty, Chief Revenue Officer at Flexential. “They’re willing to be flexible on where and when they deploy, and they’re increasingly focused on whether a provider can solve multiple infrastructure challenges at once.”
Flexential’s full report, including additional findings on networking, compute, tariffs, investments, revenue, and sustainability, can be accessed by clicking here. For more information on Flexential’s suite of high-density infrastructure capabilities and customized IT solutions, visit flexential.com.
About Flexential
Flexential empowers the IT journey of the most complex businesses by offering customizable hybrid IT solutions designed for today’s demanding high-density computing requirements. With colocation, cloud, interconnection, data protection, and professional services, the FlexAnywhere® Platform anchors our services in 40 data centers across 18 highly connected markets on a scalable 100Gbps+ private network backbone. Flexential solutions are strategically engineered to meet the most stringent challenges in security, compliance, and resiliency. Experience the power of IT flexibility and how we enable digital transformation at www.flexential.com.
