The AI Enterprise License Agreement bundles Agentforce, Einstein, and Data Cloud consumption into one commit. The discount is real. So is the forecast risk. Read the buyer side view before you sign.
The Salesforce AELA, or AI Enterprise License Agreement, folds Agentforce, Einstein, and Data Cloud consumption into a single multi year commit. This guide covers how it is structured, how Salesforce prices it, where the risk sits, and the buyer side moves that hold.
Salesforce introduced enterprise AI agreements as generative features moved from add ons to platform. The AELA is the commercial vehicle for that shift. It packages credits the customer draws down as Agentforce, Einstein, and Data Cloud run.
The pitch is one number, one renewal, one discount. The reality is a forecast bet. The buyer commits to a consumption pool before anyone knows the true run rate.
The AELA is an enterprise commitment that pools AI and data consumption into one drawdown balance. It sits above the per product editions rather than replacing them.
Most AELAs bundle three engines. Agentforce for autonomous agents, Einstein for predictive and generative features, and Data Cloud for the unified data layer that feeds them.
Credits are a shared currency. Each action draws a defined number. An Agentforce conversation, an Einstein generation, and a Data Cloud workload all debit the same pool at different rates.
A standard Salesforce enterprise agreement counts seats. The AELA counts consumption. That single change moves the planning burden from headcount, which buyers can predict, to AI usage, which they cannot.
Pricing has three moving parts. The credit rate, the committed volume, and the annual uplift. The discount applies to the rate. The risk lives in the volume.
AELA commercial levers and where the risk sits
| Lever | Vendor framing | Buyer risk | Move |
|---|---|---|---|
| Committed volume | Bigger commit, deeper discount | Pays for unused credits | Size to provable demand |
| Credit rate | Standard price card | Opaque, rarely benchmarked | Benchmark before signing |
| Annual uplift | Protects the discount | 7 to 12 percent compounding | Cap at CPI or lower |
| Rollover | Use it or lose it | Stranded prepaid spend | Negotiate carryover |
| Renewal | Continuity | Resets toward list | Lock renewal cap now |
Agentforce bills on a per conversation basis. Published Salesforce Agentforce pricing lists a rate per conversation, which means cost tracks customer behavior, not seat count. A support spike in one quarter can burn a year of credits.
Data Cloud meters on data processed and stored. As more sources connect, consumption climbs. The AELA pool that looked generous at signing tightens as the data estate grows.
The risk is concentrated in the gap between the modeled forecast and real usage. Three failure modes recur.
The customer buys a large pool for the discount, then uses a fraction of it. The unused balance expires. This is the most common and most expensive error we see.
The opposite case. A usage spike pushes consumption past the pool, and overage bills at a premium. Without a cap, a single quarter resets the budget.
The first term discount is generous to win the commit. At renewal, the rate drifts back toward list and the uplift compounds. The buyer who did not lock a renewal cap pays for it twice.
The standard account team and reseller pitch is that a larger AELA commit is always cheaper because the discount deepens with volume. We disagree. In roughly seven out of ten AI agreements we benchmarked, the discount never offset the stranded value of credits that expired unused, so the effective rate paid was higher than a smaller commit plus negotiated overage. The buyer side move is to commit to a base sized at provable first year demand, attach a pre agreed expansion rate, and refuse to fund a vendor forecast you cannot yet defend with your own telemetry.
Source: Redress Compliance advisory engagement file, 2024 to 2025.
An AELA is not an AI discount. It is a consumption forecast you are paying for in advance. Size the forecast, not the discount.
Five moves separate a defensible AELA from an expensive one.
Pull your own telemetry on support volume, automation candidates, and data sources. A baseline grounded in your data is the only credible counter to the vendor forecast.
Commit to a conservative base. Negotiate a pre agreed expansion rate for upside. You capture the discount without paying for credits you may never burn.
Push for carryover of unused credits, or a true forward that converts unused balance into the next term. Either one removes the use it or lose it penalty.
The credit rate is rarely public and rarely challenged. Independent benchmarking against comparable deals is the highest value step before signature. Salesforce reports its own AI momentum in its investor materials, which signals how hard it is pushing these commits.
A Salesforce AELA is an AI Enterprise License Agreement that pools Agentforce, Einstein, and Data Cloud consumption into one prepaid commit. It sits above the standard product editions rather than replacing them.
The AELA is priced on a credit rate applied to a committed consumption volume, with an annual uplift. The discount applies to the rate, while the financial risk lives in the committed volume.
Usually not by default. Credits typically reset each year with limited or no rollover, so over commitment leaves stranded prepaid spend. Carryover or a true forward must be negotiated.
Agentforce bills per conversation, so cost tracks customer behavior rather than seat count. A support spike in one quarter can burn a large share of an annual credit pool.
No. A bigger commit deepens the discount but often funds credits that expire unused. In most estates we benchmarked, the stranded value outweighed the discount, raising the effective rate paid.
Lock a renewal price protection and an uplift cap before you sign the first term. Without it, the rate drifts toward list at renewal and the uplift compounds.
Sizing the committed pool to provable first year demand is the most important lever. A base plus expansion option captures the discount without paying for unproven usage.
Yes. The credit rate is rarely public and rarely challenged, which makes independent benchmarking the highest value step before signature. It is the clearest way to test the discount narrative.
Agentforce credit modeling, AELA commit sizing, uplift and renewal caps, and the buyer side moves across the full Salesforce AI estate.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement leaders running the next renewal cycle.
