Salesforce's September Reckoning: Can the AI Rally Survive Contact With Reality?

Salesforce shares surge on AI agent growth, but overbought signals and one-time gains raise questions ahead of Dreamforce.

The gap between where Salesforce shares trade and where the company's actual operations stand has rarely been wider — and the next few weeks will determine which side closes that distance. The stock sits roughly 4 percent below its 52-week high of EUR 229.90, having climbed 71 percent off its yearly low, yet the company's year-to-date performance remains marginally negative at minus 2.5 percent. That contradiction captures the peculiar position investors now occupy: the market has repriced Salesforce for an AI-agent future that is still being built in real time.

The immediate catalyst for the recent surge is well documented. Second-quarter results for fiscal year 2027, released on August 27, sent shares up roughly 22.6 percent in a single session. Revenue advanced 11 percent to $11.35 billion, while reported earnings per share of $5.90 blew past the consensus estimate of $3.27. But that headline EPS figure carries a significant asterisk: roughly $2.6 billion of it stems from a one-time paper gain on the company's stake in Anthropic. Morgan Stanley's analysts, led by Adam Wood, were quick to flag the distinction, lifting their price target to $235 while holding an Equal-weight rating. Their argument: the more meaningful metric is the 14 percent growth in short-term backlog (cRPO), which points to genuine demand recovery rather than accounting mechanics.

The operational story beneath the earnings noise is where the bulls find their footing. Agentforce, the company's AI-agent platform, reached an annualized recurring revenue of $1.5 billion — a 240 percent jump year over year. Data 360 posted ARR of $2.4 billion. During the quarter, 3.2 billion agentic tasks were processed, up 97 percent from the prior quarter. Perhaps most tellingly, half of new contract value now comes from usage-based Flex Credits rather than traditional subscription seats — a structural shift suggesting Salesforce is selling actual AI-agent consumption, not just software licenses.

The analyst community has responded with a flurry of target revisions that reads like a referendum on the pivot. TD Cowen raised its target to $280 on Thursday, maintaining a Buy rating. Goldman Sachs and JPMorgan followed with targets of $271 and $265 respectively, with JPMorgan explicitly citing the "dissipation of disruption narratives" around AI and SaaS business models. Wolfe Research went further, upgrading the stock to Strong Buy on Saturday, pointing to the successful pivot toward agentic AI and a broader re-rating of the software sector. Mizuho moved its target from $250 to $265, while Citigroup edged up from $204 to $233 with a neutral stance. Truist Financial reaffirmed its Buy recommendation and lifted its target from $280 to $300. The consensus now stands at $261.15, skewed toward positive ratings.

Yet the technical picture gives even sympathetic observers pause. The relative strength index sits at 80.5, firmly in overbought territory. The stock trades roughly 40 percent above its 50-day moving average and 29 percent above the 200-day line, with a 30-day gain of approximately 34 percent. The annualized 30-day volatility of 78 percent adds another layer of fragility. These are the kind of stretched readings that historically resolve through consolidation or a sharp pullback — even when the underlying narrative remains intact.

Management is doing its part to buttress the operational case. A $25 billion share buyback program remains active, supported by free cash flow of $1.1 billion in the quarter, up 81 percent year over year. The newly unveiled Claudeforce — integrating Anthropic's Claude into Salesforce and Slack with 37 pre-built sales functions — extends the company's reach beyond its own AI infrastructure.

The proximate test arrives quickly. From September 15 to 17, Salesforce hosts its Dreamforce conference in San Francisco under the banner "Agentic Enterprise." There, the company must demonstrate that the announcement wave around Claudeforce and the planned acquisitions of Contentful and Fin translates into a tangible product portfolio — or risk the market ratcheting expectations higher without operational support.

What makes this moment genuinely uncertain is the tension between two competing readings. One sees a structural re-rating of a company successfully transitioning from legacy SaaS licensing to autonomous AI agents, backed by real usage metrics and a functioning monetization model. The other sees an overbought reaction to a skillfully communicated narrative, inflated by a one-time accounting gain and vulnerable to the inevitable gap between product announcements and enterprise adoption timelines.

The stock's distance from its 52-week low — 71 percent — reflects how much optimism has already been priced in. Whether that optimism is justified will become clearer in September, when promises are scheduled to become products. For investors holding the stock, the coming weeks represent a test of conviction rather than an invitation to chase momentum.