👉 Breaking Down Salesforce ($CRM)

Together with Alumni Ventures

Hi everyone,

The question underneath every software stock right now is who owns the surface. For twenty-five years Salesforce had one answer, which is that enterprise work happens inside a Salesforce browser tab. That answer built the tallest building in San Francisco.

As a reminder, this specific deep-dive is written by the GRIT team and is not the exclusive work of Head Analyst Austin Hankwitz.

Software spent the first half of 2026 being priced for obsolescence. The argument was clean enough to fit on a slide. Agents do the work, seats evaporate, and every company billing per user per month becomes a melting ice cube. Salesforce was the obvious poster child, because Salesforce sells more seats than almost anybody.

Then last week the company reported a quarter that looked nothing like a melting ice cube. Bookings accelerated, attrition hit a record low, and the stock had its second-biggest day in company history. Benioff stood on a stage next to Dario Amodei and told the market the death notice had been filed early.

My view is that both stories are partly right, and the market has now overcorrected in the second direction about as fast as it overcorrected in the first. So the job of this issue is to separate what actually improved from what only looked like it improved, because in this particular quarter those are two very different lists.

• Why Now 👉 The Stock Just Had Its Second-Biggest Day Ever

• The Overview 👉 The Company That Invented SaaS, Defending SaaS

• How Do They Win 👉 Twenty Years Of Data Nobody Wants To Migrate

• Business Units 👉 Two Segments Growing At Very Different Speeds

• How Do They Make Money 👉 Seats, Credits, And A Meter They Invented

• By The Numbers 👉 A Record Quarter With A Visible Asterisk

• Bonus Deep Dive 👉 Claudeforce And The Circle Salesforce Just Drew

• Risks 👉 What Would Break This

• Wrapping Up 👉 My Read On What Just Happened

Co-Invest With Firms Like a16z, Bessemer & Y Combinator in Vetted Startups

Rate of Return and Alumni Ventures are giving readers early access to high-potential startup opportunities across today's most active sectors — co-investing alongside firms like Andreessen Horowitz (a16z), Bessemer, and Y Combinator.

You get:

• Curated deal flow of high-potential startups

• Co-investment alongside name-brand venture firms

• No cost to view deals

• No obligation to invest

Why Now 👉 The Stock Just Had Its Second-Biggest Day Ever

On August 26 Salesforce reported fiscal Q2 2027 and announced an expanded partnership with Anthropic called Claudeforce. The next day the stock rose almost 23%. It closed the week at $256, up from about $205 before the print. That was the second-largest single-day move in company history and roughly $38B of market value created in a session.

Two things landed at once. Adjusted EPS came in at $5.90 against a consensus of $3.27, an 80% beat. And Marc Benioff stood next to Dario Amodei and told the market the SaaSpocalypse was over.

Here is the part most of the coverage buried. Of the $2.63 Salesforce beat by, $2.53 came from gains on its strategic investment portfolio. Salesforce holds a stake in Anthropic now carried at roughly $5.1B. Anthropic raised in May at a reported $965B valuation. Salesforce marked the position up and ran the gain straight through both GAAP and non-GAAP earnings.

Strip it out and adjusted EPS was about $3.37 against a $3.27 estimate. A 3% beat. The 80% headline was an accounting event.

The quarter underneath it was still good. cRPO grew 14% in constant currency, CFO Robin Washington said net new annual order value was the strongest in four years, and attrition sat near a record low. Those are real. But the market just repriced a $200B company on a print where the loudest number tells you the least about the business, and that is worth pulling apart.

The Overview 👉 The Company That Invented SaaS, Defending SaaS

Salesforce is the company that made software a subscription. Benioff founded it in 1999 on the argument that enterprise software should not ship on discs. The "No Software" logo was the entire pitch.

Twenty-seven years later it is a roughly $210B market cap doing about $46B a year, with roughly 820M shares outstanding and about 83,000 employees. It sells CRM to most of the Fortune 500, and it is a Dow component, which is unusual for a pure SaaS business.

FY26 revenue was $41.5B, up about 10%. FY27 guidance is now $46.1B to $46.4B. That is the profile of a mature large cap growing in the low double digits.

The problem is that the market spent the first half of 2026 deciding low double digits was the optimistic case. Between January and February the software sector shed something on the order of $2T in market value on the argument that AI agents make seat-based software obsolete. Salesforce fell roughly 28% in that window and kept sliding to a low of $146.32 on June 22. It has nearly doubled off that bottom.

The bear thesis was specific. If agents do the data entry, the pipeline logging, and the case management, you need fewer seats. Fewer seats means less revenue per customer, and Salesforce sells more seats than almost anyone. The stock traded like terminal value was genuinely in question. Claudeforce is Benioff answering that agents still need somewhere to get their data, their permissions, and their audit trail, and that somewhere is Salesforce.

How Do They Win 👉 Twenty Years Of Data Nobody Wants To Migrate

The moat is not the software. It is the accumulated state.

A company running Salesforce for a decade has millions of customer records, hundreds of thousands of process entries, thousands of custom fields, and a permissions model that took years to negotiate internally. Replacing that is not a software migration. It is a reconstruction of how the business defines a customer. That is why attrition sat near a record low this quarter even while the narrative said the category was dying. Enterprises complain about Salesforce constantly and renew anyway.

The second piece is Data 360, the data layer sitting on the roughly $8B Informatica acquisition. In Q2 it ingested 104 trillion records, up 355% year over year, including 82 trillion through Zero Copy. The bet is that the context layer matters more than the model layer. If an agent needs to know which contract governs which account and who is allowed to change it, the model does not have that. The system of record does.

The counter-argument is that none of this stops an agent from reading Salesforce through an API and writing the answer somewhere cheaper. Connectors exist. Protocol standards for tool access exist. Nothing about owning the records forces the workflow to stay inside Salesforce, and the AI-native CRM startups are built on exactly that premise. Salesforce’s response is that reading is the easy half. An agent that can pull a pipeline is a demo. An agent that can update it needs permissions, an audit trail, and an administrator who can revoke it at two in the morning. That is the part enterprise buyers have actually been waiting on, and it is the part that is hard to rebuild from scratch.

Here is my read. Salesforce is converting a defensive asset into an offensive one. Owning the data and the governance is an excellent position if agents become the primary interface. It is a poor position if agents can reach the same data through cheaper connectors. The entire thesis turns on whether governed action is a durable product or a feature that gets commoditized inside eighteen months.

Subscribe to GRIT Premium to read the rest.