A Double Upgrade That Admits Defeat: Why Wall Street Got Salesforce and AI Completely Wrong

Posted on by

(SeaPRwire) –   By: Oliver Hawthorne

Here is the uncomfortable part nobody at BofA wants to say out loud. The bank just moved Salesforce from Underperform straight to Buy, skipping a notch, and lifted its price target from $160 to $285. The stock responded instantly, trading near $237 in Monday’s pre-market, up roughly 3.5% from Friday’s close of $229.13. But the real story is not the upgrade. It is the confession buried inside it. Back in May, BofA’s bearish case rested on a fashionable thesis, that AI would eat Salesforce alive. The fear was simple enough. Why pay for seats of enterprise software when an AI agent can do the work? That logic sounded smart at dinner parties and on conference panels. Then the analysts actually did their homework. They called CTOs at major companies and asked the question directly. Would you rip out Salesforce? The answer came back almost unanimous. No. Replacing those platforms is highly unlikely, the CTOs said, because they hold business logic, data pipelines, workflows, and compliance frameworks that are brutally hard to move. This is the industry’s central anxiety exposed. Everyone predicted AI would kill the incumbent SaaS giants. Nobody bothered to ask the people who would actually sign the migration order.

Now look at the numbers behind the reversal, because they matter more than the rating itself. Salesforce’s AI-related annual recurring revenue crossed $1.5 billion in the most recent quarter, driven largely by the Agentforce platform. That is not a press release vanity metric. That is real money customers are committing to AI built inside the system they already own, not AI bought to replace it. The August 26 earnings print told the same story. Earnings per share landed at $5.90 against an estimate of $3.27. Revenue grew 10.8% year over year to $11.35 billion, edging past the $11.33 billion consensus. The company guided fiscal 2027 EPS to a range of $16.67 to $16.71 and paid a $0.44 quarterly dividend on October 8, a 0.8% yield. Wall Street’s posture was already tilted positive before BofA moved. MarketBeat counts three Strong Buys, 30 Buys, 14 Holds, and three Sells, with a consensus target of $275.14. BMO went to $285 in mid-September, Argus to $300 in late August. Even the insider tape is split in an interesting way. Director Craig Conway sold 4,500 shares on September 4 at $260.61, while Director David Blair Kirk bought 4,176 shares on September 18 at $239.33. Institutions hold 80.43% of the float.

So what is the actual commercial loop here, and where does it end? The loop runs like this. Salesforce owns the system of record. The data, the workflows, the compliance scaffolding all live inside it. AI agents only become useful when they can act on that data safely and with context. Which means the agent layer does not replace the system of record, it depends on it. Every CTO BofA interviewed understood this instinctively, because they are the ones who would carry the risk of a failed migration. The market is slowly pricing in the same realization. The stock trades above its 50-day average of $225.66 and far above the 200-day at $192.67, still below the 52-week high of $269.11, with a market cap of $188.03 billion. Wednesday’s September CPI report could shake the rate-sensitive tape short term, but it does not touch the structural argument. My read is blunt. The end-game in enterprise AI is not a graveyard of legacy SaaS vendors. It is a toll booth model, where the incumbents who own the data layer collect rent on every agent that runs through it. Watch the migration pipeline, not the demos. The first Fortune 500 company that actually rips out its CRM for a pure AI stack will be the real signal. Until that happens, the bears are arguing with a customer base that has already voted.