Why federal agencies can’t build a strong AI strategy on outdated endpoints

Federal leaders have spent the last two years building AI strategies around models, platforms, data and governance. But a critical factor is often overlooked: The devices employees use every day.

New Forrester Consulting research commissioned by Dell Technologies and Microsoft Copilot+ PC shows a growing gap between federal AI ambitions and endpoint readiness. The study surveyed 324 IT decision-makers across federal agencies, systems integrators and government contractors responsible for endpoint procurement. Nearly 60% said AI PCs will be important to their organizations’ success over the next two years, yet only about one-quarter said their workforce is fully prepared to adopt AI-powered tools.

For agencies focused on mission outcomes, compliance and public service, closing that gap should be a priority.

The readiness gap is real

Outdated endpoints create more than performance problems. They can slow adoption of modern software, increase IT support burdens, weaken security posture and make it harder to provide consistent user experiences across the workforce.

The consequences are particularly serious in federal environments, where sensitive data, mission continuity and requirements such as FISMA and Zero Trust shape technology decisions. In the Forrester study, 56% of respondents said their organizations would struggle to meet mission requirements without endpoint modernization. Nearly half also said they needed a large-scale device refresh but were concerned that it would significantly disrupt operations. Procurement complexity, approval cycles, legacy application dependencies and competing priorities all contribute to the challenge.

Waiting, however, does not preserve the status quo. It increases the likelihood that agencies will deploy new AI-enabled software on hardware that was never designed to support it.

Endpoint strategy is now AI strategy

Modern AI PCs include dedicated neural processing units (NPUs) designed to handle AI workloads efficiently on the device. This capability is becoming increasingly important as software vendors move AI features closer to the endpoint.

For agencies evaluating the Microsoft Copilot app, hardware capability is directly connected to the user’s experience. Copilot-capable devices must meet defined hardware standards, including sufficient NPU performance, to support the full range of intended features. Devices that do not meet those thresholds may deliver limited or inconsistent functionality. Agencies should also account for Microsoft Copilot+ vendor funding requirements when evaluating eligible configurations and refresh plans.

On-device processing can be especially relevant when data sovereignty, connectivity limitations, classification requirements or restricted operating environments limit what can move to the cloud. Nearly three-fourths of Forrester respondents said local or hybrid AI approaches are essential to supporting security needs. This does not mean local AI replaces cloud AI. It means agencies can align processing architecture to the sensitivity of the data, the risk of the workflow and the needs of the mission.

Start with mission outcomes

Federal organizations do not need to modernize every endpoint overnight. A deliberate, use-case-led approach can help agencies capture value while managing operational risk.

First, identify workflows where AI can make a measurable difference. Forrester respondents pointed to workflow automation, visualization and assistance with writing, editing and summarizing documents as leading near-term use cases. These practical applications can deliver value while agencies strengthen governance and build workforce confidence.

Next, segment users by role, workload and risk profile. Not every employee needs the same device configuration or level of local processing. Evaluating application requirements, data sensitivity, security controls and online or offline operating conditions can help direct investment where it will have the greatest impact.

Finally, use controlled pilots before scaling. Targeted deployments can reveal compatibility requirements, support needs, adoption barriers and security considerations. Modernization should also include training, responsible-use policies and governance for sensitive information.

The endpoint is no longer a background IT concern or a routine refresh exercise. It is part of whether agencies can securely adopt and scale the next generation of AI-enabled software. Agencies that align endpoint modernization with mission priorities, workforce readiness and security requirements will be better positioned to turn AI ambition into measurable federal outcomes.

Read the full report on Dell.com