What Federal IT Leaders Are Choosing Amid New AI Capabilities

Ryan Haylock is Chief Information Officer at Evolver, a federal technology and cybersecurity firm headquartered in Reston, Virginia. He has led IT and cybersecurity organizations across commercial, private-equity, higher education, manufacturing, and elder-care sectors with a focus on operationalizing cybersecurity efficiently, smart process automation, and building highly performing teams. 

Federal IT looks different in 2026 than it did a year ago. As workforce numbers compress and procurement consolidates, new AI capabilities are landing on federal platforms. This shift is happening whether the supporting operating models are ready or not. The question I find sitting underneath most federal IT conversations this year is not which tools to buy or which talent to hire. It is what the operating mix looks like across internal staff, contracted partnerships, and supporting technology, and whether that mix will actually deliver. 

This is at the heart of the Capability Mix question: At what rate are these inputs now moving? 

Federal IT conversations often describe the capability-mix decision as a sequence of separate procurement and staffing choices. A contractor task order on one workstream. A managed service contract on another. A platform upgrade somewhere else. A hire when the budget clears. Each decision gets evaluated on its own terms. 

The federal record over thirty years tells us where that approach ends. The transactions are individually defensible. The aggregate is something nobody designed. 

The framing I have come to find more useful, across HIPAA, SOX, FERPA, and ITAR-adjacent regimes before this seat, is that the capability mix is a portfolio of partnerships. Internal staff with their own leaders. Contracted partnerships with managed service providers, integrators, and capability vendors. Technology partnerships with platforms that augment human capability. The forms are different. The disposition that makes any of them perform is not. 

This is familiar territory for federal contracting. Twenty years of federal acquisition literature has emphasized the buyer-seller relationship as the core unit of federal IT delivery, with explicit attention to trust, expectations, and honoring commitments inside a contract structure that defines the work.  

The literature holds up. Contract administration doctrine has engaged with post-award performance management for decades. What I would add to that long-standing conversation is the operating-disposition lens: the year-over-year work of making a partnership perform, inside whatever contract structure shapes it, is what determines whether the capability-mix decision delivers on paper or in practice. 

The contract sets the conditions. The disposition operates within them. 

The Government Accountability Office (GAO) reported in June 2026 that the civilian workforce across the 22 CFO Act agencies declined by approximately 256,000 employees, more than 11 percent, from December 2024 to January 2026 (GAO-26-108583). Eighteen of the 22 agencies posted declines above 10 percent. Some agencies declined by more than 30 percent in a single year. 

Federal IT spending has not contracted alongside that workforce. GAO reported in July 2025 that the federal government spends more than $100 billion annually on IT and cyber-related investments, and that agencies typically spend about 80 percent of that on operations and maintenance of existing systems (GAO-25-107795). For fiscal year 2025 specifically, that came to roughly $83 billion of $105 billion. 

The EY 2026 Government and Public Sector Federal Trends Report surveyed 131 federal agency leaders and named the workforce skills gap as the single most commonly cited barrier to modernization, at 44 percent of respondents. Slow procurement and cybersecurity threats followed at 32 percent each. 

The mission has not contracted either. CMMC Phase 2 begins this November. Federal post-quantum cryptography migration runs on a NIST-published timeline through 2035. Procurement consolidation under the OneGov strategy is reshaping the vehicles through which federal IT capability is acquired. Compliance obligations are growing. Internal capacity to deliver them has shrunk. 

Federal CIOs are absorbing the gap. The choice they are making, often without naming it as a single choice, is what their operating capability mix will look like across internal staff, contracted partnerships, and supporting technology. 

AI augmentation in federal IT operations is no longer a future question. The GSA and ServiceNow announced a OneGov agreement in September 2025 that made AI-augmented IT service management broadly available to federal agencies at substantial discount through September 2028. Federal agency leaders themselves confirm the direction: the EY 2026 report found 92 percent view AI as a critical tool for improving efficiency, and 86 percent report barriers to scaling it agency-wide. Only 38 percent said they have a comprehensive, unified AI governance strategy in place. 

Read those numbers together and the picture is clear. Federal IT leaders see AI as necessary, but most are not yet operationally ready to use it well. 

AI delivers real value in federal IT operations when the operating model around it is built to absorb it. Routine triage, summarization, pattern recognition across incidents and requests — these are exactly the kinds of workloads that scale well with AI and free analyst time for the judgment-heavy work that only people can do. That is augmentation, not replacement. Done well, it realistically lets  federal workers spend their time on the work that matters. But without an operating model behind it, AI produces an expensive layer that adds complexity rather than absorbing it. We’ve seen this happen with every wave of federal IT tooling purchased ahead of establishing the discipline needed to run it effectively. 

Federal AI carries real constraints: Data sovereignty; Governance; Model auditability; Integration with existing identity and access frameworks. These are not arguments against federal AI. They are the conditions under which federal AI works. Federal IT leaders who are getting value from AI in service operations today treated those constraints as design inputs from the start, paired the AI with the operating discipline to govern it, and made the technology part of a broader partnership portfolio rather than a freestanding purchase. 

The capability-mix lens applies to AI the same way it applies to everything else. Technology earns its place when the operating model around it is real. 

Four anchors for any federal CIO or CISO making capability-mix decisions in 2026. These are not Evolver-specific. They are conclusions any federal IT operator arrives at after watching enough capability mixes succeed and fail. 

The capability-mix question is not new. It has been a central question of federal IT delivery for a generation. What is new is a workforce that compressed in a single year, procurement consolidation reshaping the acquisition landscape, and AI capability widening the options available faster than most operating models are absorbing it. 

The federal CIOs who will deliver in this environment are the ones who treat the capability mix as a deliberate portfolio decision instead of the accumulated residue of separate transactions. AI augmentation is part of that mix now. It will deliver where the operating model is built for it, and it will disappoint where it is not. 

The contract sets the conditions, the disposition operates within them, and the work compounds across years. 

So does the absence of it. 

Sources