Invest1 distinct publisher3 min readPublished
Agentforce and Data 360 annual recurring revenue is real and it recurs. Almost everything else in the quarter landed within a rounding error of consensus, or came from paper marks.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
Run the ARR line backwards and the base is small. Growth of more than 210 percent puts last year's Agentforce and Data 360 figure at no more than roughly $1.26 billion, which means about $2.6 billion was added in twelve months [3][14]. Now the top line: $11.35 billion at 11 percent growth implies a year-ago quarter near $10.23 billion, an increase of about $1.12 billion, or roughly $4.5 billion annualised [1][15]. The AI and data run-rate therefore grew by an amount equal to close to three-fifths of the company's annualised revenue increase [16]. ARR is a run-rate snapshot and revenue is recognised over time, so the two never reconcile cleanly, but that gap is too wide to be timing alone. A material part of the $3.9 billion is money Salesforce was already collecting under a different product name.
The rest of the print is a study in narrow margins. Revenue beat by $20 million on $11.33 billion of consensus [1]. Current remaining performance obligations came in at $33.5 billion against $33.4 billion expected, a beat of about 0.3 percent [10][21]. The third-quarter guide midpoint of $11.46 billion sits 0.4 percent above the FactSet number [12][19], and the raised full-year midpoint of $46.25 billion is 0.3 percent above the $46.112 billion analysts carried [13][20]. The second-half reacceleration case rests largely on CFO and COO Robin Washington's statement that net new annual order value posted its strongest growth in four years, a claim the company did not attach a figure to [11].
The earnings line is where the marketing and the accounting part company. Adjusted EPS of $5.90 against a $3.27 consensus reads as an enormous beat, but $2.53 of it was a paper gain on strategic investments, which is about 96 percent of the surprise [5][6][17]. The operating number, $3.37, cleared consensus by ten cents [22]. The report attributes the gain to the strategic portfolio generally rather than to any single holding [5]; that portfolio includes Anthropic, Databricks and Stripe [7], and Salesforce announced a partnership with Anthropic the same day that puts Claude inside its platform [8].
What survives all of that is the scale of the ARR itself. Just under $3.9 billion is about 8.4 percent of the midpoint of the raised full-year revenue guide [18], which is past the point where a number can be dismissed as pilot spend or a Benioff talking point, even though he used it as one, telling investors ARR is "about to cross $4 billion" [2]. The market paid for it: an 11 percent move in the extended session, on top of a 33 percent recovery from the late-July low that still leaves the stock down 23 percent for the year [4][9]. Agentic AI has cleared the recurring-revenue test at one large vendor. It has not yet cleared the harder one, which is showing up in the consolidated growth rate.
Ranked by verification strength, evidence, and original report placement.
Annual recurring revenue from Salesforce's Agentforce and Data 360 products grew over 210% relative to a year earlier, reaching just shy of $3.9 billion.
Salesforce and Anthropic announced a strategic partnership called "Claudeforce" on Wednesday, integrating Claude directly into Salesforce's platform.
Salesforce CFO and COO Robin Washington said net new annual order value reached its strongest growth in four years, keeping the company on track for second-half organic revenue reacceleration.
Salesforce posted $11.35 billion of revenue for the second quarter of fiscal 2027, growing 11% year over year and surpassing Wall Street estimates of $11.33 billion.
CEO Marc Benioff said in a press release: "AI is delivering value across every layer of our platform. We're seeing incredible demand for our AI and data products, with ARR about to cross $4 billion."
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Company-reported figures, one publisher, no independent verification
All quantitative claims are precise and internally consistent, and the source discloses the composition of the EPS beat rather than hiding it, which raises quality. But the cluster has a single publisher relaying company-supplied numbers and management characterisations; the pivotal AI/data ARR figure is a non-GAAP, self-defined metric with no Agentforce/Data 360 split, no incrementality reconciliation and no third-party audit or corroborating outlet.
Large disclosed recurring revenue, thin usage detail
Adoption is evidenced in monetary terms rather than in usage terms: nearly $3.9bn of recurring AI and data revenue growing over 210%, plus $33.5bn of 12-month contracted backlog and a claim of the strongest net new annual order value growth in four years. Against that, there are no customer, seat, agent-run or consumption figures, the ARR bundles data products with agents, and the Claudeforce integration is an announcement with no disclosed rollout or usage. Real money is recurring, but the depth of agent usage inside it is unmeasured.
Headline AI narrative outruns the underlying quarter
The framing is 'AI momentum fuels revenue growth' and 'value across every layer of our platform', while the company grew 11%, guidance was raised by roughly 0.3-0.4% above consensus, cRPO beat by about 0.3%, and roughly 96% of the headline EPS beat came from a paper mark on strategic investments including an Anthropic stake announced as a partner the same day. The $3.9bn ARR line is genuinely large and fast-growing, which keeps the gap moderate rather than extreme, but it is about 8.4% of full-year guidance and is not broken out between agents and data products.
Self-reported quarter with disclosed investor-partner overlap
Every material figure originates with the company on the day it wants an AI narrative believed, delivered via a promotional CEO quote and a metric of its own construction. Salesforce also holds an investment in Anthropic, booked a large paper gain on strategic investments that supplied nearly all of the EPS beat, and announced the Claudeforce partnership with Anthropic the same day - an overlap of investor, partner and reporter roles the source states but does not interrogate. The publisher itself carries no disclosed stake and does flag the paper gain, which tempers the score slightly.
Facts solid, interpretation single-sourced
Confidence in the reported numbers is high because they are precise, mutually consistent and drawn from a company's own quarterly disclosure relayed by an established financial newsroom, and the derived ratios use only those figures. Confidence in the wider read-through is limited: one publisher, one reporting day, no independent verification of the ARR construction, no product-level breakout, and no disclosure of Agentforce usage or Claudeforce deployment scope.
invest
Salesforce's 14% cRPO is the best agent-demand read yet. Organic growth is 6.5%2 distinct publishers
invest
Salesforce bought $27.1B of its own stock in a quarter it grew 13%2 distinct publishers
invest
Claudeforce: Salesforce hands Anthropic the seller's screen and keeps the write path2 distinct publishers
invest
Salesforce's double digits, minus Informatica: agentic AI is real and still 2% of revenue1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 26, 2026