The U.S. Department of Veterans Affairs awarded Salesforce a $1.6 billion, three-year Agentic Enterprise License Agreement on July 24, 2026. The contract, structured as a one-year deal with two optional renewal years, will deploy Salesforce’s Missionforce platform and Agentforce AI agents across a healthcare system that served more than 17 million veterans through 170 medical centers and over 1,100 outpatient clinics in 2025, according to Salesforce’s announcement.

Nine days before the VA deal was announced, KeyBanc Capital Markets published a CIO survey that found enterprise customers view Agentforce unfavorably. The investment bank’s managing director for software equity research, Jackson Ader, wrote in the report: “Our checks and customer conversations have not been strong, nor has the feedback been on Agentforce.”

The gap between federal enthusiasm and private sector hesitation frames one of the central questions in enterprise AI adoption: who is actually buying AI agents at production scale, and why?

The Contract

The VA’s Agentic Enterprise License Agreement covers three areas of deployment, according to the Salesforce press release.

First, AI-powered 24/7 contact center support. Agentforce Public Sector and Agentforce Health will handle patient triage, intake, and care coordination. The AI agents will surface knowledge during live calls, triage cases automatically, and automate benefits verification. This builds on VA Health Connect, which has already handled more than 40.6 million veteran calls on Salesforce infrastructure.

Second, a connected collaboration layer. Slack is already deployed across more than 150 VA medical and outpatient centers as what Salesforce calls an “agentic operating system.” The VA’s Unified Patient Scheduling initiative will integrate with Slack to coordinate care across more than 40,000 provider services. Salesforce claims the target is reducing average appointment scheduling time from 28 days to minutes.

Third, unified data infrastructure. MuleSoft and Data 360 will integrate VA’s legacy systems into a single data layer. Tableau will provide dashboards for claims processing, employee performance, and daily operational reporting.

Kendall Collins, CEO of Salesforce’s Missionforce and Government Cloud division, framed it as an efficiency play in the press release: “Every minute a VA employee spends navigating disconnected systems is a minute not spent serving a Veteran.”

Federal Momentum

The VA contract does not exist in isolation. Salesforce’s federal footprint has expanded rapidly across multiple branches of the U.S. government in 2026.

On July 8, Salesforce announced that the U.S. Air Force 441st Vehicle Support Chain Operations Squadron is using Missionforce National Security to manage a $13.5 billion fleet of over 84,000 vehicles across 389 locations. That deployment is backed by a $72 million Air Force Enterprise License Agreement.

In January, the U.S. Army awarded Salesforce a $5.6 billion Indefinite Delivery/Indefinite Quantity (IDIQ) contract for Missionforce National Security.

Combined, these three contracts represent a contract ceiling of approximately $8.8 billion across VA healthcare, Army operations, and Air Force logistics. Federal procurement at this scale is unusual for any software category. For AI agents specifically, it has no precedent.

The Skepticism

The scale of federal buying contrasts with what analysts are finding in the private sector.

KeyBanc’s July 15 report, covered by The Register, drew on CIO surveys and partner conversations. The findings were blunt: “Customers’ data is not in order to do meaningful AI work,” and “Agentforce, as a product, just isn’t there.”

The KeyBanc analysts reported that partners are “just now beginning to convert Agentforce proof of concepts into deals in the pipeline.” More CIOs in their survey expect to deprioritize Salesforce within their IT budget than increase spending on it over the next 12 months. The report also cited “aggressive price increases” and noted that the majority of surveyed customers are “not willing to pay for AI capabilities through their CRM provider.”

Salesforce responded to KeyBanc’s findings by calling Agentforce “the fastest-growing product in Salesforce history,” citing customers like Engine, Falabella, and AAA going live “in weeks, not months,” according to The Register.

The numbers support that growth claim. In Q1 FY2027 (ended April 30, 2026), Agentforce ARR surpassed $1 billion for the first time, according to The Silicon Review. Combined AI and data ARR reached $3.4 billion, up more than 200% year-over-year. Two quarters earlier, Agentforce ARR had just crossed $500 million.

The Pricing Question

Beyond product readiness, there is a pricing structure question that will affect every organization considering an Agentforce deployment.

Salesforce introduced the AELA model in October 2025 as a flat, seat-based arrangement that bundles AI agent access at a capped cost. It was designed to make adoption predictable.

In January, Gartner warned Salesforce users that the capped enterprise agreement “may not be available when they come to renew.” The analyst firm’s concern: today’s all-you-can-eat pricing could become tomorrow’s cost shock when contracts expire and usage has grown. Salesforce denied it would move away from capped plans.

For the VA, the contract ceiling of $1.6 billion over three years puts a concrete dollar figure on federal agent deployment. But the one-year base contract with two optional renewal years means the VA could walk away after year one if Agentforce fails to deliver. The total ceiling is theoretical until renewals are exercised.

Why the Federal Government Is Buying First

Federal procurement is notoriously conservative. Agencies require years of security certification (FedRAMP High authorization), HIPAA compliance for healthcare, and audit trail infrastructure that most SaaS vendors cannot provide. The VA’s decision to commit $1.6 billion to AI agents does not reflect recklessness. It reflects the opposite: a calculated bet by an organization that already runs mission-critical operations on Salesforce.

The Veterans Crisis Line was modernized on Salesforce Government Cloud in 2025, according to the VA press release. During go-live, the platform supported responses to more than 10 critical incidents, including two cases where emergency assistance was dispatched to veterans in danger. VA Health Connect has processed 40.6 million calls. SQUARES gives 2,100 authorized users access to veteran eligibility data. These are not pilot programs. They are production systems with years of operational history.

This is the key difference between federal and private enterprise adoption. The VA expanded an existing deployment. Most private enterprises are evaluating Agentforce from scratch, with unstructured data, no FedRAMP-level compliance framework, and no existing Salesforce infrastructure designed for agent workloads. KeyBanc’s finding that “customers’ data is not in order” reflects a starting-line problem that the VA solved years ago.

The Competitive Landscape

The VA contract also signals positioning in a broader federal AI infrastructure fight.

The Register noted that Oracle holds a $9.99 billion VA electronic health record contract signed in 2018, though that rollout was suspended in April 2023 over patient safety concerns and rebooted in February 2025. The same week as the Salesforce VA deal, the Department of Defense signed a $7 billion, ten-year Enterprise Software Agreement with Oracle.

Salesforce and Oracle are now operating in parallel across federal healthcare and defense. Salesforce handles the agent and CRM layer. Oracle handles electronic health records and database infrastructure. Neither vendor has displaced the other; instead, the federal stack is fragmenting into specialized layers, each with its own multi-billion-dollar contract.

Microsoft, OpenAI, and Anthropic are absent from federal agent procurement at this scale. Microsoft’s Perception agentic security platform launched July 27, but targets enterprise cybersecurity, not government healthcare operations. OpenAI and Anthropic have no comparable government-deployed agent infrastructure.

The $8.8 Billion Test

Salesforce now has approximately $8.8 billion in federal contract ceilings for AI agent and platform deployments across three branches of the U.S. government. The VA alone is responsible for 82 million direct-care appointments per year. The Air Force fleet includes 84,000 vehicles. The Army IDIQ spans operational workflows across the Department of War.

If these deployments succeed, they become the reference implementation that private enterprise procurement teams use to justify their own Agentforce purchases. A VA case study with real patient outcomes and scheduling improvements would be more persuasive than any analyst report or product demo.

If they fail, or if Gartner’s pricing warning proves correct and renewal costs spike beyond the original ceiling, the VA deal becomes a cautionary tale about federal AI procurement risk that could chill enterprise adoption for years.

The next 12 months will determine which outcome materializes. The VA’s one-year base contract means the first renewal decision arrives by summer 2027. By then, Salesforce will need to show that 28-day scheduling times have actually decreased, that AI-powered triage is handling real clinical workloads, and that the cost structure holds at scale.

KeyBanc’s CIO survey says the private sector is waiting. The federal government decided not to.