For thirty years the SaaS industry was organized around a single unit of economic value: the seat. You counted users, multiplied by a monthly fee, and the math added up. Salesforce built a $300 billion company on that model. This week, the same company is trying to bury it.

Salesforce's Agentic Enterprise License Agreement (AELA)—rolled out through Dreamforce 2025 and now landing its first marquee deals in 2026—replaces per-seat and consumption-based billing with a flat, unlimited-use fee covering Agentforce, Data 360, MuleSoft, and Slack across two- or three-year terms. The Adecco Group, one of the largest staffing firms in the world, signed a multi-year AELA in March 2026 covering 60+ countries with unlimited global access to Agentforce 360. More are in the pipeline.

President Miguel Milano summarized the thesis in a single quote: "AELA is for customers that have already experimented. They're ready to scale. They want to go all in, so we agree on a flat fee, and then it's a shared risk."

That single sentence reshapes the enterprise AI procurement conversation for every CIO and CFO currently staring at a Q2 budget variance driven by unpredictable token and consumption costs. Here is what is happening, why it matters, and what enterprise buyers should do about it before signing the first multi-year AELA of their own.

The Problem AELA Is Solving

Talk to any CIO who rolled out agentic AI in 2025 and you will hear the same complaint: the bill makes no sense.

Consumption-based AI pricing—tokens, API calls, agent actions, tool invocations—sounds logical in a demo. You pay for what you use. But the moment agents succeed at scale, the cost curve goes non-linear. A single agent that handles 50 customer service tickets per hour triggers hundreds of LLM calls, thousands of tool invocations, and a cascade of downstream queries. Multiply by ten thousand employees and a few dozen workflows, and the monthly bill stops looking like software and starts looking like electricity—except electricity utilities publish tariffs, and AI vendors do not.

Finance teams trying to forecast a consumption-metered AI budget for 2026 face a forecasting problem that does not exist for any other software category. You cannot tell your board what this year's AI spend will be; you can only tell them what it was last quarter.

Per-seat pricing, the incumbent model Salesforce itself pioneered, has the opposite problem. Agents do not sit in seats. A fully automated workflow with no human in the loop technically has zero seats and infinite value. Charging per seat for a capability that displaces seats is an obvious strategic contradiction—one Salesforce's competitors (Microsoft, Google, Workday, SAP) have all been navigating with varying degrees of elegance.

AELA attempts to resolve both problems simultaneously: flat fee, unlimited use, shared risk. The enterprise buyer knows the cost. Salesforce knows the commitment. Both sides bet the business value scales faster than the vendor's cost to serve.

What's Actually in an AELA

The specific pricing is not public—AELA is a negotiated agreement, not a rate card—but the structural components are consistent across the deals Salesforce has disclosed:

• Products covered. Agentforce (Salesforce's agent platform), Data 360/Data Cloud (unified customer data), MuleSoft (integration), and Slack (collaboration + Slack AI). Some deals extend to Agentforce 360, the flagship bundle. Marketing Cloud, Commerce Cloud, and core CRM seats are typically separate.
• Term. Two or three years, with renewal terms that are—by most customer accounts—the most contested part of the negotiation.
• Usage envelope. "Unlimited" in marketing language, but most contracts include fair-use language, baseline usage assumptions, and true-up mechanics if the customer materially exceeds documented usage categories.
• Deployment scope. Typically global, covering all subsidiaries, business units, and geographies that the customer operates in. Adecco's 60+ country footprint is the current reference point.
• Commercial structure. A set annual fee or schedule across the term, usually with an escalator, tied to a defined deployment roadmap.

The unwritten component matters just as much: AELA is acquisition pricing disguised as a license agreement. Lisa Singer, VP and Principal Analyst at Forrester, puts it plainly: the model "reframes AI agents as economic actors that generate business value, not metered features that incur variable cost." Salesforce has publicly acknowledged it is willing to lose money on some AELA deals early in the term. The bet is that once agents are wired into customer workflows at scale, the switching cost is structural, and year-four pricing reflects that reality.

Why CFOs Will Like It (and What to Watch)

For finance teams, AELA solves the hardest problem in 2026 enterprise AI budgeting: predictability.

A flat-fee structure converts AI spending from a variable OpEx line into something closer to a depreciable capital commitment. Forecasts become tractable. Variance analysis becomes meaningful. The CFO can present a board with an AI budget that will not surprise the finance committee in six months.

The deeper shift is philosophical. Under consumption pricing, every CFO question starts with "how much will this cost per use?" Under AELA, the question becomes "what is the ROI on a fixed capital allocation, and how do we maximize utilization of the capacity we already paid for?" That is a capital allocation question, not an operating expense question, and it plays to strengths the finance function already has.

Three things to watch, however.

First, renewal leverage sits with the vendor. UpperEdge's analysis of early AELA deals flags the renewal dynamics as the single biggest negotiation risk. Salesforce has every incentive to let customers deploy deeply during the initial term—unlimited usage encourages it—then reprice based on observed deployment breadth at renewal. Finance teams should negotiate explicit renewal caps, multi-year rate protection, and defined exit ramps before signing the first term.

Second, "unlimited" is rarely truly unlimited. The fine print typically includes fair-use thresholds, workflow assumptions, and excluded categories (often the most consumption-heavy ones). Before signing, finance should insist on a written definition of the usage envelope and explicit treatment of categories like multi-agent orchestration, high-volume tool calls, and cross-cloud federation.

Third, AELA amplifies vendor concentration risk. A flat-fee unlimited license is designed to make Salesforce the default place for any new AI workflow. That is great for standardization and terrible for negotiating leverage three years from now. The CFO should treat AELA the way they would treat any long-term infrastructure commitment: negotiable, but not renewable into infinity.

Why CIOs Need to Care Even More

For technology leaders, AELA is less a pricing story and more a platform strategy story. The structure of the agreement forces architecture decisions that will outlast the contract.

Integration gravity becomes strategy. An AELA covering Agentforce, MuleSoft, and Data 360 creates an enormous economic incentive to build all new agentic workflows inside the Salesforce stack. Why deploy a Microsoft Copilot agent for a sales workflow when the marginal cost of doing it in Agentforce is effectively zero? Why buy a best-of-breed customer data platform when Data 360 is already paid for? The rational buying behavior flips from best-of-breed to best-available-within-the-license, and that flip is exactly what Salesforce is engineering.

Governance and observability become CIO-owned problems, not vendor-owned. Under consumption pricing, the vendor's bill is, at minimum, a usage signal: spikes tell you something is wrong. Under AELA, that signal disappears. An agent stuck in a loop costs Salesforce money, not the customer, so Salesforce's alerting catches it before the customer sees it. Agents running outside sanctioned use cases can proliferate silently because there is no cost trigger. CIOs moving to AELA need to build internal agent observability—usage dashboards, workflow attribution, cost-of-serve allocation across business units—that replaces the pricing signal the flat fee removes.

Multi-vendor agent strategies become harder to justify internally. When a CFO has already paid Salesforce for unlimited agentic capacity, a request to buy an additional agent platform for a specialized workflow gets pressure-tested against a zero marginal cost alternative. CIOs committed to multi-vendor portfolios (Microsoft + Salesforce, or Salesforce + specialty vendors like ServiceNow) need a defensible internal framework for when specialization beats free. Without it, procurement will gravitate to the AELA platform for everything, even for workflows where it is not the best fit.

Standards support matters more, not less. Because AELA deepens vendor commitment, the CIO's only credible exit plan runs through standards. Insist on demonstrable support for the Model Context Protocol (MCP), agent-to-agent communication (A2A), and emerging orchestration standards before signing. The more portable the agents built during the term, the weaker the vendor's renewal leverage.

The Competitive Dominoes

AELA is not happening in isolation. The broader AI industry is groping toward pricing models that survive the transition to agents, and each major vendor is placing a distinct bet.

• Microsoft is consolidating its AI capabilities into Microsoft 365 at an enterprise-wide level, bundling Copilot access inside existing E3/E5 agreements rather than charging separately for agents. The Microsoft bet: lock in the E5 renewal, AI becomes an included feature rather than a priced capability.
• Google is embedding Gemini into Workspace with a similar bundling logic, using its productivity suite as the distribution mechanism for agentic features.
• Anthropic and OpenAI are leaning harder into usage-based and hybrid models, with token-priced APIs for developers and enterprise tiers (ChatGPT Enterprise, Claude for Enterprise) priced per user with capacity limits. Neither has matched Salesforce's unlimited-at-flat-fee commitment.
• ServiceNow, Workday, SAP are watching Salesforce carefully. Each has a dominant platform position in an adjacent category—IT service management, HR, ERP—and each faces the same question: can we move to AELA-style pricing without gutting the per-seat revenue that currently funds the platform?

If Adecco's AELA scales successfully through 2026 and 2027, expect competitive responses by Q3: unlimited-agent tiers, fixed-fee agentic bundles, and outcome-based pricing experiments from the vendors who cannot afford to let Salesforce define the category alone.

What to Do Now

For CFOs, the action is to build an AI pricing decision framework that compares the three models side by side on predictability, ceiling exposure, and renewal risk. Do not sign the first AELA your Salesforce rep offers. Run it against a three-year consumption-based projection at 2x current usage, against a Microsoft-style bundled-in scenario, and against a mixed portfolio. Whichever you choose, make the choice explicit.

For CIOs, the action is to inventory agentic workloads by platform fit, standards dependency, and switching cost. The AELA conversation is a one-time opportunity to either rationalize the stack around Salesforce or draw a clear line defending a multi-vendor strategy. Letting procurement accidentally resolve it is the worst possible outcome.

For CROs, CMOs, and COOs, the action is to treat the agent economics as a competitive question, not a procurement question. Competitors signing AELAs and scaling agents across every workflow at zero marginal cost will reprice customer service, onboarding, pricing, and field sales. Laggards by a single budget cycle will be structurally disadvantaged by the end of 2027.

The seat is not dead yet. But Salesforce just wrote its obituary in a pricing term sheet, and the rest of the industry is reading closely.

Want to calculate your own AI ROI? Try our AI ROI Calculator — takes 60 seconds and shows projected savings, payback period, and 3-year ROI.

Continue Reading

Sources

• AI Agents Become Economic Actors: Salesforce Rewrites the Rules of Pricing — Forrester
• Salesforce's New Agentic Enterprise Licensing Agreement (AELA) — UpperEdge
• Agentic Enterprise License Agreement (AELA): AI Pricing Model Explained — BuildMVPFast
• New AI Pricing Models: Avoiding the 'Too Hot' Zone — FourWeekMBA
• Gartner Questions If Salesforce AI Will Stay All-You-Can-Eat — The Register