Share with your CMO/CRO
Salesforce is betting its future on Agentforce, an AI platform that automates customer service workflows, but the numbers reveal how lopsided that bet still is. With Agentforce annualizing at roughly $800 million against a total revenue base guided to $46 billion in fiscal 2027, the AI story accounts for about 2% of the business while the other 98% grows at 6 to 7%, below what Wall Street wanted. Shares have fallen more than 25% year-to-date, swept into a broader software sell-off that has taken the iShares Tech-Software ETF down 22% over the same period.
What this means for your business
If your commercial stack runs on Salesforce, you are holding a platform in the middle of a painful identity transition. The company that sold you Sales Cloud and Marketing Cloud as the operating system for revenue is now reorienting around AI agents, and the core applications you actually run on are, by UBS analyst Karl Keirstead’s read, “under pressure.” That pressure is structural: enterprise buyers are redirecting budget toward AI and data initiatives, which means the marketing and commerce modules that anchor many CRM deployments are facing headwinds from their own vendor’s strategic pivot.
The 98-versus-2 problem is not just a Salesforce investor story, it is a renewal calculus for any company mid-contract on legacy Salesforce modules. Agentforce’s $800 million annualized figure sounds significant, but spread across Salesforce’s customer base it represents thin penetration, and the analysts who are bullish, Goldman Sachs at a $281 target and Morgan Stanley at $287, are essentially pricing in a second-half 2027 acceleration that has not shown up in the leading indicators yet. Bernstein, rating the stock underperform at a $194 target, frames the deeper risk as share loss from being the largest incumbent in a market now contested by lighter, AI-native competitors. That framing matters for buyers, not just shareholders.
The honest falsification condition for Salesforce’s AI story is whether domain expertise, meaning the customer data and workflow context baked into years of CRM deployment, actually produces better agent outcomes than a competitor standing up a fresh model on the same data. Goldman’s Gabriela Borges makes that the central question, and it’s the right one. If Agentforce can demonstrably outperform on sales or service outcomes because it sits on top of a decade of customer history, the installed base becomes a moat. If it can’t, that history is just switching-cost inertia, and inertia has a shorter half-life every year. The renewal you’re weighing right now is a bet on which of those is true.
Based on reporting from Salesforce’s underwhelming earnings won’t be enough to stem the AI software slide, originally published 2026-02-26 03:00:00.
