U.S. Federal AI Adoption: Momentum, Gaps, and the Road Ahead
Key Findings from the 2026 Nutanix Enterprise Cloud Index — U.S. Federal Edition
By Sherry Walshak, Director – Public Sector Solutions Marketing, Nutanix
Federal agencies face one of the most demanding technology balancing acts in government today: maintaining the legacy systems that underpin critical services— from Social Security payments to air traffic control—while simultaneously adopting cloud-native architectures, containers, and AI at mission speed. The challenge isn’t choosing between the two. It’s carrying both at once, with finite budgets, lean teams, strict compliance requirements, and little room for error.
Pressure is coming from every direction. Agency leadership is setting AI adoption targets. OMB guidance is pushing toward cloud-smart and zero-trust architectures. Program offices are standing up AI-enabled capabilities faster than enterprise IT can govern them. And all of this is happening against a backdrop of urgent infrastructure decisions—including a forced migration away from VMware to new platforms by October 2027—that are compressing timelines even further.
To understand where federal IT leaders actually stand, Nutanix commissioned Wakefield Research to survey 100 U.S. federal IT and engineering executives in late 2025. The findings, published in the 2026 Nutanix Enterprise Cloud Index: U.S. Federal Government Report, offer a clear ground-level view of how agencies are approaching containerization, hybrid infrastructure, and AI workload deployment—and where the gaps remain.
Infrastructure Isn’t Ready—and AI Is the Stress Test
Hardware procurement used to follow a predictable cycle. Agencies standardized on a processor architecture, picked a memory configuration, bought servers from a preferred vendor, and ran them for a defined lifecycle. AI has broken that pattern entirely.
New GPUs, accelerators, and memory technologies are arriving faster than most organizations can evaluate them. The ECI data reflects this urgency: 74% of U.S. federal IT leaders say their on-premises infrastructure is not fully ready to support AI workloads.
The hardware challenge is compounded by the pace of the AI software ecosystem. With new large language models released constantly, the models agencies deploy today may not be the ones they rely on six months from now. Frameworks update, dependencies shift, GPU requirements change—and many of those models run in containers, even when the surrounding environment is still VM-based. This creates a specific operational question federal IT leaders are wrestling with:
How do you absorb continuous hardware and software change without re-architecting your environment every time the industry moves?
The answer lies in platform choice. Agencies that select a foundation capable of supporting both virtualized and containerized workloads—with unified management and built-in governance—can absorb change at the infrastructure layer rather than being forced to respond to it with each new deployment.
Containers Are Becoming the Common Language of Federal Modernization
Across civilian, defense, and intelligence environments, containerization has emerged as the most practical path between where agencies are and where they need to be— not because it is new, but because AI has compressed the timeline.
The ECI numbers tell the story: 85% of federal IT leaders say AI is a significant driver of container adoption. 82% are already building new applications in containers today, and 88% expect container use to grow over the next three years. The logic is straightforward: containers allow agencies to bundle application code and dependencies in portable, secure environments, enabling consistency across deployments and the agility to iterate quickly—without rebuilding the underlying environment each time.
The practical reality for most federal agencies, however, is a mixed estate. 72% of federal organizations are currently running AI-enabled applications on a mix of virtual machines and containers. VMs aren’t going away; they run the established workloads agencies depend on every day. The infrastructure must support both— without separate teams, separate tools, or separate management planes.
The federal deployment context adds further complexity. Unlike commercial enterprises, agencies must routinely operate at the edge—forward-deployed environments, tactical networks, disconnected enclaves—where managed cloud services are not an option. 62% of federal organizations expect their containerized applications to be running on-premises or in private cloud infrastructure within three years, the same share doing so today. Containerized applications that run consistently on-premises, in a cleared cloud, and at the edge are the only architecture that works across the full operational spectrum.
Looking ahead, 48% of federal IT leaders expect their organizations to be running more than five AI-enabled applications within three years, including 26% who anticipate more than ten. The agencies building toward that future need to be laying the infrastructure foundation now.
The Governance Gap: When AI Adoption Moves Faster Than Policy
One of the most striking findings in the Enterprise Cloud Index report is the prevalence of Shadow AI—unsanctioned AI tools that employees adopt independently, outside of IT oversight or organizational policy. 96% of federal IT leaders report encountering AI applications implemented by employees in non-IT functions.
This isn’t a compliance failure—it’s a signal. Employees are turning to AI tools because they’re accessible and genuinely improve their work. The goal isn’t to discourage that momentum; it’s to channel it through governed, sanctioned pathways. That requires a platform capable of enforcing policy at the infrastructure layer, not just through policy documents.
Organizational silos make this harder. 84% of respondents say silos between business units and IT make it difficult to execute technology initiatives effectively. When teams aren’t aligned, enterprise AI adoption fragments—and risk compounds. A unified platform gives teams a sanctioned, governed way to innovate without working around IT.
Agentic AI: The Next Capability Federal Agencies Are Already Sizing Up
The AI conversation in federal circles has moved well beyond chatbots and search augmentation. 67% of federal IT leaders expect to deploy agentic AI or autonomous agents within the next three years—the most anticipated new capability in the survey, ahead of generative AI (61%) and conversational AI (56%).
Agentic systems—which can execute multi-step tasks, learn from context, and act on behalf of an organization—align naturally with the scale and complexity of federal operations: benefits adjudication workflows that cross multiple agencies, intelligence analysis pipelines synthesizing high volumes of disparate inputs, and logistics systems that anticipate demand rather than simply responding to it.
63% of federal IT leaders expect AI agents to transform their business processes and operations, and 55% see them as a primary driver for developing next-generation mission capabilities—not just improving efficiency, but fundamentally changing how federal work gets done.
Agentic AI requires more than compute. It requires data environments that are governed, auditable, and tightly controlled—and platforms that can enforce policy at the infrastructure layer. Because agentic workloads will run alongside existing virtualized applications, not replace them, the platform must manage the full workload spectrum consistently. Agencies that invest in that foundation now will be positioned to deploy agentic capabilities as the technology matures.
Data Sovereignty: From Compliance Requirement to Strategic Foundation
Federal agencies have always prioritized data sovereignty. For U.S. federal agencies specifically, it means operational control: where data lives, how it is protected, who can access it, and what happens when the network is contested or disconnected.
The Enterprise Cloud Index data reflects how deeply this is felt: 90% of federal IT leaders say data sovereignty is a high priority or must-have factor in infrastructure decisions—higher than any other segment in the broader global survey. 60% say they need to run infrastructure within a single country, driven by security requirements, data protection regulations, and classified operating environments.
What is changing is the urgency. As AI systems become more deeply embedded in mission operations, the data those systems depend on becomes correspondingly more sensitive. An AI system is only as sound as the data governance beneath it. Sovereignty can no longer be an afterthought,it must be the foundation on which AI infrastructure is built.
Distributed sovereign cloud architectures—where data governance, key management, identity control, and network policy are enforced at the infrastructure layer and behave consistently across on-premises, cloud, and edge environments—are emerging as the practical path forward. Critically, those architectures must accommodate the full application estate: not just containerized AI workloads, but the virtualized applications agencies depend on every day.
The Hidden Cost: Time Lost to Maintenance
Federal agencies spend approximately 80% of their IT budgets on operations and maintenance of existing systems—patching, upgrading, and troubleshooting legacy infrastructure—leaving only 20% for modernization and innovation, according to the U.S. Government Accountability Office.
AI raises the cost of that lost time. Every hour spent on maintenance is an hour not spent on:
- Training or updating AI models
- Deploying new mission-critical services
- Integrating the latest hardware and LLMs
- Strengthening data governance
- Improving constituent outcomes
Platforms with automation for lifecycle management can reclaim that time. Organizations that shift resources from upkeep to innovation will move faster and deliver better outcomes for the constituents, warfighters, and mission partners they serve.
Where This Leaves Federal IT Leaders
The Enterprise Cloud Index data points to a consistent theme across every finding: the agencies best positioned to deploy AI—at scale, securely, and in compliance—are those that have already made deliberate choices about their infrastructure foundation. Not point solutions assembled around individual workloads, but unified platforms capable of supporting virtual machines and containers, on-premises and edge environments, today’s established workloads and tomorrow’s agentic AI.
Three imperatives stand out from the research:
- Bridge the hybrid gap now. AI applications are increasingly built cloud-natively, but most agencies operate across hybrid infrastructures. Closing that gap is a prerequisite for delivering AI at scale.
- Treat infrastructure as a governance layer, not just a compute layer. AI workloads demand governed, auditable data environments. Modern software delivery models—supporting multi-tenant environments and cross-agency requirements—are no longer optional; they are foundational.
- Proactively rethink infrastructure strategy. Agencies that balance performance, security, and compliance in their infrastructure decisions today will be better positioned to deploy AI responsibly,and at mission scale,tomorrow.
AI is moving too quickly for static infrastructure, siloed teams, or architectures that weren’t designed for change. The window to build the right foundation is now.
Background and Research
The Nutanix Survey was conducted by Wakefield Research among 1,600 IT and engineering executives (minimum seniority: manager) at organizations with 500 or more employees across 14 markets, with an oversample of 100 U.S. Federal workers, between November 13–23, 2025, using an email invitation and online survey.
For the global sample, results vary by ±2.38 percentage points at the 95% confidence level; for the U.S. sample, ±4.9 percentage points.
References
- U.S. GAO. Artificial Intelligence: Generative AI Use and Management at Federal Agencies. GAO-25-107653. Published July 29, 2025. https://www.gao.gov/products/gao-25-107653
- 2026 Nutanix Enterprise Cloud Index: U.S. Federal Government Report. Survey of 100 U.S. Federal cloud, IT, and engineering executives conducted by Wakefield Research for Nutanix.
- U.S. GAO. Information Technology: Agencies Need to Plan for Modernizing Critical Decades-Old Legacy Systems. July 2025. https://www.gao.gov/products/gao-25-107795
© 2026 Nutanix, Inc. All rights reserved. Nutanix, the Nutanix logo, and all Nutanix product and service names mentioned herein are registered trademarks or trademarks of Nutanix, Inc. in the United States and other countries. All other brand names are for identification purposes only.
