Salesforce reports its second fiscal quarter, the three months ended 31 July 2026, after the US close on Wednesday 26 August 2026, with the call at 2:00 p.m. Pacific. The company gave that date in a notice on 5 August. It is worth stating plainly because it is a week earlier than last year — the equivalent quarter was released on 3 September 2025 — and a preview built on the wrong week is worth nothing.
Consensus sits at $3.28 a share on about $11.33 billion, which is +12.7% and +10.7% against the $2.91 and $10,236 million Salesforce reported for the same quarter a year ago. Those are press-reported figures from third-party estimate feeds, not a series this site stores. The company itself guided $11.27–11.35 billion of revenue and $3.25–3.27 of non-GAAP earnings per share. Consensus is one cent above the top of the guide. There is no disagreement here to resolve.
The interesting number is what sits underneath the twelve percent. Take the midpoint of Salesforce's own earnings guide and multiply it by any plausible share count, and the profit behind it is flat to down.
| Diluted shares | Implied non-GAAP net income | vs. $2,795M a year ago |
|---|---|---|
| 871M — the April quarter's actual | $2,839M | +1.6% |
| 845M | $2,755M | −1.4% |
| 823M | $2,683M | −4.0% |
| 819M — shares outstanding on 21 May | $2,670M | −4.5% |
Those are our figures, not Salesforce's; the company guides earnings per share and does not publish a net income guide. The share count is the only assumption in them, and it is bounded on both sides by disclosure. It cannot be much above 871 million, because that is what the April quarter averaged while the buyback was still landing mid-quarter. It is unlikely to be far below 819 million, which is what the cover of the last quarterly filing counted on 21 May. Everywhere in that range, the guided twelve percent of earnings growth is a smaller denominator rather than a larger numerator.
The points
- Salesforce repurchased $27.1 billion of its own stock in the three months ended 30 April 2026 — the company's own words in its May release, and the 10-Q for that quarter puts total cash paid at $27.2 billion. This is money spent in one quarter, not an authorisation, and not a cumulative figure stretching back years.
- $25 billion of it came from newly issued debt. In March 2026 Salesforce sold $25.0 billion of senior notes maturing between 2028 and 2066 and, in the 10-Q's words, "used the net proceeds from the March 2026 Notes to fund an accelerated share repurchase program."
- The board authorised $50.0 billion in February 2026, replacing the previous remaining authorisation. $22.9 billion of it was left at 30 April. So there are three different numbers in circulation — $50B authorised, $27.1B executed in a quarter, $22.9B remaining — and only the middle one describes what actually happened.
- The net share count fell 14.3% in a year, from 956 million on 21 May 2025 to 819 million on 21 May 2026, both counted on the cover of the quarterly filing. That is net of everything Salesforce issued in stock compensation, which matters: it has $9.03 billion of unrecognised stock-compensation expense still to run.
- Salesforce cut its own full-year free-cash-flow growth guidance from approximately 9–10% to approximately 4–5%, and said in the same line that the change was "to reflect the impact of the $25 billion debt issuance for the ASR."
- Interest expense went from $68 million to $317 million year over year in a quarter that only carried the new notes for part of its length, and the company expects the increase to persist through the year.
- There is no disclosed Agentforce revenue line, in the 10-Q or anywhere else. There is $1.2 billion of Agentforce annual recurring revenue, disclosed in the May release — 2.6% of the year Salesforce has guided to.
- Slightly more than four points of this quarter's guided 10–11% growth is the Informatica acquisition. That is not our estimate; Salesforce puts it in the guidance table.
What the $27.1 billion actually is
The figure has been circulating as "$27 billion in buybacks" as though it were a pledge. It is better than that, and more specific. The mechanics, from the filing:
In February 2026 the board authorised $50.0 billion, inclusive of what follows and replacing the prior remaining authorisation. In March, Salesforce issued $25.0 billion of senior notes with maturities out to 2066, netting $24.8 billion. It then entered accelerated share repurchase agreements with several banks for an aggregate $25.0 billion, paid the whole $25.0 billion up front, and took immediate delivery of approximately 103 million shares at an average price of $198.34 — about 80 percent of the total shares expected under the agreements. Alongside that it bought 11 million shares in the open market at an average $192.00, for $2,145 million. Final settlement of the accelerated programme is expected in the third quarter of this fiscal year, which is the one after Wednesday's.
So the $27.1 billion is one quarter's executed spend, on shares delivered, at prices the company has published. It is roughly 16.0% of Salesforce's market capitalisation — $169.1 billion on a price of $205.43 at the 20 August close and the 823 million diluted shares the April quarter carried, which is our arithmetic and moves with the price every day. It is 1.85 years of trailing free cash flow and 63% of a year's revenue.
None of which it was paid for with. Free cash flow in that quarter was $6.56 billion. The buyback was four times it, and the gap came from the bond market.
What the buyback cost, in the one number this site is built on
Here is the part nobody has put together, and it does not need an estimate because Salesforce did the arithmetic itself and published both halves three months apart.
On 25 February 2026, initiating fiscal 2027 guidance, Salesforce guided full-year operating cash flow growth of approximately 9% to 10%. On 27 May 2026, it updated that to approximately 4% to 5% for both operating and free cash flow, and attributed the change to the debt issuance that funded the buyback.
Run it against the base. Fiscal 2026 closed with $14,402 million of free cash flow on $41,525 million of revenue — a 34.7% margin. Revenue guidance for this year did not move: $45.8–46.2 billion in February, $45.9–46.2 billion in May, both around 11% growth. So:
| Free cash flow guide | Implied FY2027 FCF | Implied FCF margin | Rule of 40 |
|---|---|---|---|
| February: +9% to 10% | $15.77B | 34.3% | 45.3 |
| May: +4% to 5% | $15.05B | 32.7% | 43.7 |
About $720 million of free cash flow, 1.6 points of free-cash-flow margin, and therefore 1.6 points of Rule of 40 — because on this site's definition the cash margin is one of the two terms. The buyback did not sit beside the operating story. It came out of it, on the company's own numbers, in the direction the score reads.
That figure is worth holding next to the trailing one. Salesforce currently scores 45.2 on our trailing measure: $42.83 billion of revenue over the four quarters through April, up 11.0%, on $14.66 billion of free cash flow, a 34.2% margin. The company's own guidance for the year now sits below that, and the difference is interest.
The arithmetic above is ours. What is not ours is the direction: Salesforce told everyone in May that the cash line was going to grow half as fast because of the bonds it sold to buy its stock.
Agentforce: the label is on 65% of the revenue, the revenue is not disclosed
This is the load-bearing question for anyone previewing this quarter, and it has a clean answer.
Effective with its first quarter of fiscal 2027, Salesforce changed how it disaggregates revenue. Subscription and support is now reported in two buckets: Agentforce Apps, and Data 360, Headless Platform, and Other. Agentforce Apps was $6,910 million in the April quarter, 65% of subscription revenue. It grew 8.9%.
Read what is inside it, from the filing: "Agentforce Sales, Agentforce Service, Agentforce Marketing, Agentforce Commerce, Agentforce Apps Flex Credits and Slack." That is the Sales Cloud, the Service Cloud, Marketing, Commerce and Slack — the core of Salesforce, renamed. The other bucket, at $3,683 million, grew 24.8%, and $444 million of Salesforce's revenue that quarter was Informatica, which sits in it. Strip the acquisition out of the consolidated line and total revenue grew 8.7% rather than the reported 13.3%.
So there is now a category called Agentforce that contains two-thirds of the company's subscription revenue and grows at roughly the rate the company has always grown. And there is no line anywhere that says what Agentforce sold. The 10-Q does not use the phrase "annual recurring revenue" once.
What Salesforce does disclose, in the May release: $1.2 billion of Agentforce ARR, up 205%; nearly $3.4 billion of combined Agentforce and Data 360 ARR, up over 200%, of which $1.1 billion is Informatica Cloud; Agentforce One Edition and Agentforce for Apps bookings up nearly 60%; 3.8 billion "Agentic Work Units" delivered; 28.6 trillion tokens processed.
Annual recurring revenue is not revenue. It is an annualised snapshot of active contracts, on a definition the company sets, and it is not audited into the income statement. Put it against the year anyway, because that is the only denominator available: $1.2 billion is 2.6% of the $46.05 billion Salesforce has guided to for this year. The combined AI and data figure is 7.4%, and a third of that is an acquisition.
How much revenue did Agentforce earn in the July quarter? Nobody outside the company knows, and we are not going to invent a number to fill the gap. The disclosure is an ARR figure and a bookings growth rate; there is no revenue line, no consumption revenue split, and no way to derive one from what is published. The strongest thing that can be said is the shape: a product cycle the company names 29 times in a quarterly filing, embeds in the label on 65% of its revenue, and quantifies only as a snapshot metric equal to about one fortieth of the year.
That is not an accusation. Consumption products are genuinely hard to disclose early, and Salesforce's own risk factors say so — that the "markets and monetization strategies" for Agentforce and Data 360 "remain relatively new and uncertain." It is simply the boundary of what is checkable, and a preview that pretends otherwise is fiction.
What Wednesday does to our model
We published a forward model for Salesforce on 21 August. Its base case is $287.15 against $205.43 at the 20 August close, with a bear of $164.70 and a bull of $382.98.
The model does not have an Agentforce line, deliberately, and the reason is the section above: the $1.2 billion is ARR, and that ARR is already earning inside the subscription revenue the model projects. A separate line would count the same dollars twice. Agentforce appears where it honestly can — in the growth rate of the contracted book.
Which is where the reconciliation gets interesting, because the model's revenue and the company's guide agree almost exactly, and their forward books do not.
The model's driver is Salesforce's own forward measure: current remaining performance obligation, the contracted revenue due in the next twelve months, times the rate it converts to revenue. That conversion has held between 31% and 33% for two years; the model uses 31.5% and grows the book 2.5% a quarter.
- On revenue, that produces $11.39 billion for the quarter reporting Wednesday, against a guide of $11.27–11.35 billion. The model is $44 million above the top of the guide, 0.7% above its midpoint. For a projection calibrated on reported quarters rather than on the guide, that is agreement.
- On the contracted book, it does not agree. The model's 2.5% quarterly step implies cRPO of about $34.4 billion, which would be +17.1% on the $29.4 billion Salesforce reported at 31 July 2025. The company guides approximately 14%, which is about $33.5 billion. The model's forward book is running 2.8% hot.
So: does Wednesday change the model? On the revenue line, no — and the size of the no is the point. One quarter landing inside a guide the model already reproduces to within a percent tells a twenty-quarter projection nothing. What would change it is the cRPO print, because that is the model's only driver and it compounds; a 2.8% gap held for twenty quarters is a different company at the end of it.
The buyback moves the model in one narrow way, and it is small. The model holds the share count at 823 million, the diluted figure from the April quarter. About 26 million shares are still to come — 103 million delivered is roughly 80% of the total expected, so the programme has about a fifth left. Settling it takes the count to roughly 797 million and, holding the equity value constant, the base case to about $296, a 3.2% move. Against a base case sitting 40% above the current price, retiring the last of the buyback is a rounding difference. The contracted book is worth twenty times more to the valuation than the share count, and it is the one Wednesday actually prints.
The gap this quarter's guidance carries at the bottom
One measurement note, because the consensus card at the top of this page is on the non-GAAP line and it has to be.
Salesforce guided this quarter to $3.25–3.27 of non-GAAP diluted EPS and $1.74–1.76 of GAAP diluted EPS. That is a $1.51 wedge — 46% of the non-GAAP figure. A year ago the same wedge was $0.95, or 33%. For the full year the company guides a GAAP operating margin of 20.6% against a non-GAAP margin of 34.3%: 13.7 points.
What is in it is disclosed and unglamorous: amortisation of purchased intangibles, stock-based compensation and restructuring. Stock compensation was $870 million in the April quarter alone, with $9.03 billion still unrecognised. The wedge is widening mostly because Salesforce keeps buying companies and keeps paying in stock — and the second of those is directly connected to why the buyback has to keep running to hold the share count down.
Read Wednesday's headline number against $3.25–3.27, not against $1.74–1.76, and never subtract one from the other.
Grading what we said three weeks ago
On 3 August we wrote that Salesforce's revenue growth had bottomed at 7.6% and re-accelerated to 13.3%, the fastest in thirteen quarters, and read that as evidence that downstream enterprise AI demand was arriving. The reported figure was right. This preview adds the part that piece did not carry: Salesforce disclosed that $444 million of that quarter's revenue was Informatica, which takes the organic rate to 8.7% — and the company's own guidance for this quarter says slightly more than four points of the 10–11% is the same acquisition. Salesforce's phrase for what it expects next is "organic revenue re-acceleration in the second half of FY27," which is a statement that it has not happened yet.
The earlier piece's other finding stands and is worth repeating for Wednesday specifically: the quarterly Rule of 40 score is the wrong instrument here. The equivalent quarter a year ago scored 15.7, because Salesforce's cash flow concentrates in the fiscal first quarter — a pattern the company confirms in the filing, which says its "first quarter is typically our largest collections and operating cash flow quarter" and that "our second or third quarter has historically been our smallest." Expect the quarterly score to fall by something like forty points on Wednesday. It will mean nothing.
What to watch
- Current remaining performance obligation against approximately 14%, or about $33.5 billion. This is the single most valuable number in the release and it is not the revenue line. It is the only forward volume Salesforce publishes, it is what our model runs on, and our model is currently 2.8% above the company's own guide for it.
- Non-GAAP net income, not non-GAAP EPS. Divide it by the year-ago $2,795 million. If it grows less than the earnings-per-share line — and on the guide it grows roughly zero — the difference is the buyback, and it will not repeat at this size once the programme settles.
- Whether full-year free-cash-flow growth stays at approximately 4–5%. February said 9–10%; May said 4–5% and blamed the bonds. A third number on Wednesday would be the clearest read available on what the $25 billion actually costs.
- Any Agentforce figure denominated in dollars of revenue rather than ARR, bookings, work units or tokens. There has never been one. If one appears, it is the most important disclosure of the quarter regardless of its size.
- The organic growth rate, excluding Informatica. Salesforce publishes the acquisition's contribution in dollars each quarter. Subtract it. The company has promised organic re-acceleration in the second half of this year, and this is the last quarter before the half it promised it in.
Salesforce reports the quarter ended 31 July 2026 after the US close on Wednesday 26 August, with the call at 2:00 p.m. Pacific; the date is from the company's own notice of 5 August. Consensus of $3.28 and about $11.33 billion is press-reported from third-party estimate feeds and is not a series this site stores or verifies. All guidance quoted — for the quarter, for the year, and the February guidance it replaced — is Salesforce's own, from its releases of 27 May 2026 and 25 February 2026. The buyback mechanics, the March 2026 notes, the repurchase and authorisation balances, the Agentforce Apps and Data 360 revenue split, the Informatica contributions, interest expense, stock compensation, the contracted book and the cash-flow seasonality are disclosed in the 10-Q and release for the quarter ended 30 April 2026; prior-year comparatives come from the release of 3 September 2025, and the 956 million and 819 million share counts are the cover pages of the respective quarterly filings. The Agentforce figures Salesforce publishes — $1.2 billion of ARR, and the bookings and usage metrics — are annual recurring revenue and operating measures, not revenue. No Agentforce revenue figure is disclosed in any filing, and none is estimated here. Ours rather than the company's: the implied-net-income table, the implied cash-flow margins and Rule of 40 scores, the $720 million and 1.6-point differences, the 14.3% net share reduction, the ex-Informatica growth rates, and every share-of-market-cap and share-of-revenue ratio. The fair values and the conversion and growth assumptions behind them are outputs of our Salesforce model of 21 August 2026, not company forecasts. The price of $205.43 is the close of 20 August 2026, and the market capitalisation built on it moves daily.
