AI Isn't Killing Salesforce—It's Making It Stronger
Salesforce ($CRM): Q3 results and updated valuation
Salesforce is one of many software stocks that have struggled this year. The market has latched onto a story that AI will replace or weaken traditional software vendors. The narrative is that companies will (ambitiously) build their own AI-driven CRMs, ERPs, or other tools, or replace existing software with cheaper AI-first alternatives. And since software stocks often traded at premium valuations due to their attractive fundamentals, they are now facing a correction.
The problem is that this consensus does not match reality. A recent MIT study found that roughly 95% of corporate AI projects fail. Most DIY attempts collapse because of weak integration, poor context, and governance gaps.
LLMs like ChatGPT are excellent for individual productivity, but they fall apart inside complex enterprises where data is fragmented across dozens of systems. The MIT research even shows that external partnerships, including those with Salesforce, have twice the success rate of in-house attempts.
Despite this, the bearish narrative around Salesforce has taken hold. It is built almost entirely on speculation and untested assumptions, and it has weighed on the stock. Slower growth and high prior expectations have not helped. The result is that Salesforce shares have barely moved in five years. They are up only 17% over that period, and before last week’s earnings they were close to flat.
But Salesforce’s Q3 results were quite strong, with a beat and a guidance raise. Agentforce is gaining traction, and while most companies are struggling to extract value from AI, Salesforce is delivering AI solutions at scale.
Here is my analysis of Salesforce’s Q3 earnings and my updated valuation.
The AI DIY Problem
Like I mentioned, most in-house AI projects fail. Companies need external vendors to make AI work.
Success depends on three components:
Context: An AI system is only as strong as its inputs. Enterprises may have plenty of internal data, but it is usually raw, scattered, and inconsistent. Useful AI requires data that is cleaned, structured, permission-aware, and updated in real time. Salesforce’s Data 360 (formerly Data Cloud) turns fragmented data across the enterprise into a unified, usable view. Crucially, Data 360’s advantage is that it can pull in data from external sources without copying it. It connects to legacy systems and data lakes directly, which avoids the heavy DIY work of building ETL and ingestion pipelines for every silo.
Determinism: LLMs and AI-agents can follow rules, but they are still probabilistic, meaning they guess. In an enterprise setting, that is a problem. You cannot let an autonomous agent hallucinate while talking to a customer, checking inventory, or issuing refunds. Making an LLM deterministic requires heavy engineering work that chains the model to hard-coded API calls. You end up building a full software engine to keep the AI on the rails. Salesforce already has two decades of codified workflows. Without those rails, controlling AI inside the enterprise becomes far harder.
Embedded Access: AI is most effective when woven into the workflow, not isolated in a standalone app. Salesforce controls the entire CRM platform and Slack, which removes friction.
Alongside the MIT study, Chief Revenue Officer Miguel Milano provided important context around DIY AI when asked whether Gen-AI could displace SaaS vendors:
“Now the problem is they [enterprises] have been experimenting [with DIY projects]... They’ve gone from experimentation now to frustration... and now they’re all saying, you know what? ‘This is hard. This is much harder than we thought.’... LLMs cannot do this alone. And now to answer your question, the last mile is hard. Companies need the context... deterministic workflows... You don’t want the agents to be essentially executing based on what they found in an LLM. You want the agents to execute in a deterministic way... and they need AI that is embedded where the humans are... and only Salesforce can do that.”
Enterprises are waking up to the same conclusion: they need a partner to push their agentic AI efforts forward. This is why Agentforce is by far Salesforce’s fastest-growing segment. Agentforce ARR has accelerated to 330% YoY growth, fueled by customers processing more than 3.2 trillion tokens through the LLM gateway since launch.
Now it’s simply a matter of waiting until the market fully understands that AI is not disrupting Salesforce.
Q3 Results and Guidance
In Q3, Salesforce’s revenue grew 9% YoY to $10.26 billion, while cRPO reached $29.4 billion and RPO reached $59.5 billion, up 11% and 12% YoY, respectively.
Salesforce also renamed its cloud offerings this quarter, although the underlying products did not change:
Agentforce Sales (Sales Cloud): revenue of $2.3 billion, up 8%
Agentforce Service (Service Cloud): revenue of $2.5 billion, up 8%
Agentforce 360 Platform, Slack & Other (Platform & Other): revenue of $2.2 billion, up 19%
Agentforce Marketing & Commerce (Marketing & Commerce): revenue of $1.4 billion, up 1%
Agentforce Integration & Analytics (Integration & Analytics): revenue of $1.4 billion, up 6%
The 360 Platform, Slack & Other segment continues to accelerate: 14% growth in Q1, 16% in Q2, and 19% in Q3. Slack is an important driver. It is becoming the conversational interface for every app, agent, and workflow. Slack can search across Salesforce data, the internet, and internal files, which makes the CRM far easier to use. Instead of navigating dashboards, employees interact with agents directly in Slack channels.
Agentforce and Data 360 reached $1.4 billion in ARR, up 114% YoY. Agentforce alone is now above $500 million in ARR, up 330% YoY. Salesforce has closed more than 18,500 Agentforce deals since its launch in September 2024. That is extremely fast adoption for such a new product. As a reminder, the agentic AI market is estimated at $155 billion by 2030.
Profitability continues to improve. Operating margin reached 21.2%, up from 20% a year ago, and operating income rose to $2.2 billion from $1.9 billion.
Operating cash flow grew 17% to $2.3 billion. Stock-based compensation remained flat at $819 million, essentially unchanged for four years now.
The company bought back an incredible $3.8 billion of stock in the quarter, compared to just $1.3 billion a year ago. At today’s valuation, that equals about 1.5% of the market cap, and Salesforce likely repurchased shares at lower prices. Management expects share repurchases in the second half of the year to be about 50% higher than in the first half. Since buybacks totaled $4.8 billion in the first half, the second half should come in around $7.2 billion, or about $3.4 billion in Q4.
In total, Salesforce is on track to repurchase about 5% of today’s market cap across the entire year. If that’s not a clear sign management believes share are undervalued, then I don’t know what is.
For Q4, Salesforce guides to $11.13 to $11.23 billion in revenue, implying 11% to 12% growth. That includes roughly three points from the Informatica acquisition. cRPO is expected to grow about 15%, with four points coming from Informatica.
For the full year, revenue guidance has increased to $41.45 to $41.55 billion, or 9% to 10% growth. The operating margin is now expected to be 20.3%, down from the prior 21.2% outlook.
At Dreamforce in October, Salesforce also shared key long-term guidance. The company expects organic double-digit growth to return, driven by accelerating net new annual order value.
Excluding Informatica, Salesforce targets 10%+ annual organic growth from fiscal 2026 to 2030, implying revenue of over $60 billion by 2030. In my previous earnings update, I assumed at least $60 billion by 2030, which would translate into an annual shareholder return of at least 10%.
Now, Salesforce has a reputation for being too optimistic, but with Agentforce growing rapidly and becoming a larger part of the business, I don’t think $60 billion by 2030 is unreasonable.
This raises the question: after climbing around 15% last week, is Salesforce still a buy today?
Should You Buy Salesforce Today?
Looking at just the 5-year timeframe from today to 2030, Salesforce would need to grow its free cash flow by nearly 13% annually to justify today’s price.
This assumes a 5-year forecast period, a 10% discount rate, and a 2% terminal growth rate.
With organic revenue growth at 10%+ and the Informatica acquisition—which is entirely ignored—13% free cash flow growth is reasonable. At this rate, investors would see a 10% annualized return, based on the 10% discount rate.
Using conservative assumptions from my previous earnings update—$60 billion in revenue, a 25% profit margin, and a P/E of 25 in 2030—investors can expect an annualized return of around 10%. While there was slightly more value when the share price was in the $220s, today’s price still represents a more than fair value for a strong business.
As before, I believe any price below $250 makes Salesforce a strong buy.
Thanks for reading.
Lucas
Author & Founder, Summit Stocks
In case you missed it:
Disclaimer: the information provided is for informational purposes only and should not be considered as financial advice. I am not a financial advisor, and nothing on this platform should be construed as personalized financial advice. All investment decisions should be made based on your own research.