Agent Revenue Is Not Agent Value: What Agentforce’s ARR Does and Does Not Prove

What the headline says

Salesforce reports that Agentforce, its agent product, reached an annual recurring revenue (ARR) of $1.2 billion in the first quarter of fiscal 2027, up 205% year on year². A quarter later it reported ARR above $1.5 billion, up over 240%, with 7.0 billion agentic work units (AWUs) delivered to date across Agentforce and Slack, 3.2 billion of them in the second quarter¹. Those are builder-reported figures: company-defined, unaudited key performance indicators, published by the party selling the product.

For a portfolio owner deciding how much to commit to agent platforms, the question is what these numbers license. They show that one large vendor is monetising agents quickly. They do not show that its customers are earning a return.

What changed between the two quarters

The two ARR figures are not like for like. The second-quarter release states: "Effective Q2 FY27, Agentforce ARR includes our AI offerings, Slackbot and Headless 360"¹. Futurum Research, an independent analyst, notes the same change and describes it as widening the scope of the metric⁴. Neither source quantifies how much of the move from $1.2 billion to above $1.5 billion comes from the wider definition. No like-for-like series is published.

The AWU count has a different weakness. It is cumulative, it is counted across Agentforce and Slack, and the releases give no definition of the size or value of a unit. A rising count shows more tasks executed. It does not show what any task was worth to the customer.

The scale of the claim

Total revenue growth slowed from 13% to 11% over the same two quarters, and both figures include an acquired contribution from Informatica of $444 million and $456 million respectively¹². Salesforce raised its fiscal 2027 revenue guidance to $46.1–46.4 billion, a range that also includes the pending Contentful and Fin acquisitions¹. Management expects organic revenue reacceleration in the second half; Futurum treats that as an expectation still to be tested, not a result⁴.

Set against that guidance, $1.5 billion of ARR is roughly 3% of the midpoint. This is our own comparison of a run rate with an annual forecast, so it is indicative only. It says that agent revenue is material and fast-growing, and that it is not yet what is driving the company's overall growth.

Futurum also reports that about half of Agentforce bookings come from customers refilling credits after consuming them⁴. That is consistent with consumption pricing: the metric then reflects use that customers have chosen to pay for, which is useful, and also reflects a pricing model that converts activity into revenue whether or not the activity produces value.

What buyers report

McKinsey's State of AI 2026, an independent survey, covers 1,719 respondents in 97 nations, fielded from 4 May to 8 June 2026 and weighted by national GDP³. Its findings sit differently from the vendor figures:

• 40% of organisations with revenue above $1 billion are scaling AI agents in one or more functions, up from 27% a year earlier. Smaller organisations are essentially flat, at 22%.

• 37% of respondents attribute any EBIT impact to AI, essentially unchanged from 2025.

• About 6% qualify as AI high performers, attributing at least 5% of EBIT to AI and describing the impact as significant. That share is also unchanged.

• About one in five respondents say their organisation limits AI use because of operating costs, including tokens. Among high performers, costs constrain software coding agents about three times as often as for others.

Scaling is rising among large firms; reported financial impact is not. The vendor figure and the buyer figure measure different things, in different populations.

What this proves and does not prove

It supports three conclusions. One large platform vendor is selling agents at a rate that is material to its own business and compounding. Large enterprises are moving agents into more functions. Reported enterprise financial impact from AI has not yet moved.

It does not support four others. It does not show that Salesforce's ARR growth is organic, because the definition changed. It does not show that vendor ARR equals customer value, because no customer-side cost, outcome or return per AWU is disclosed. It does not link the vendor's customers to McKinsey's respondents, so the 37% cannot be read as a verdict on Agentforce. And it does not show independent verification: we found no audit or analyst challenge that rebuilds the ARR or AWU figures.

A decision test

Treat vendor agent ARR as evidence about the vendor's monetisation and about market demand, and not as evidence about your return. Before committing portfolio spend, require three things from the seller and from your own teams: the ARR or usage figure restated on a constant definition; a unit cost per outcome in your own workflows, including token and human-review cost; and a measured baseline against which an EBIT or productivity effect can be tested after two quarters. If a proposal cannot supply those, the ARR headline is the only evidence it has, and that evidence does not bear on the decision being made.

1. Salesforce — Q1 FY27 results, SEC 8-K Exhibit 99.1 — 27 May 2026 — https://www.sec.gov/Archives/edgar/data/0001108524/000110852426000125/crm-q1fy27xexhibit991.htm

1. McKinsey & Company — The state of AI in 2026: On the road to ROI — 25 Aug 2026 — https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

1. Futurum Research — Salesforce Q2 FY 2027: Can Agentforce Drive Revenue Reacceleration? — 27 Aug 2026 — https://futurumgroup.com/insights/salesforce-q2-fy-2027-can-agentforce-drive-revenue-reacceleration/