Salesforce Rises 23% on Strong Q2 Earnings and AI Growth

Article by: Tide Research

After hours on August 26, Salesforce delivered earnings that left short sellers speechless: quarterly revenue of $11.35 billion, up 11% year-over-year; adjusted earnings per share of $5.90, 80% higher than Wall Street’s expectation of $3.27; and full-year revenue guidance raised to $46.1 billion to $46.4 billion. The stock surged 13% in after-hours trading, and by the next day’s close, the cumulative gain reached 23%, adding over $40 billion in market value overnight.

On the same day, another SaaS giant, Intuit, reported earnings that also beat expectations, yet its stock plunged 12%. Its user growth guidance for the next fiscal year for TurboTax was only 2% to 3%, prompting JPMorgan to slash its price target from $605 to $331. Adobe and ServiceNow each declined by approximately 3%.

AI is forcing the SaaS industry to undergo a brutal species divergence.

In February 2026, the SaaS industry experienced a sell-off comparable to a mass extinction.

Within a week, the software sector lost over $1 trillion in market value. SAP plunged 16%, ServiceNow dropped 11%, and Salesforce, Adobe, and Datadog were all hit hard. The S&P 500 Software & Services Index fell to a nine-month low.

A brutally simple logic has struck at the heart of every SaaS company: if AI agents can do human work, businesses won’t need as many software seats. Workday announced an 8.5% workforce reduction, citing AI-driven efficiency gains; MongoDB’s stock price halved because AI can now directly help developers generate lightweight databases. The market’s consensus is that the per-seat SaaS business model is being fundamentally undermined by AI.

Salesforce was the hardest-hit stock amid this panic. From the beginning of the year to the end of June, CRM's share price fell a cumulative 43%, closing lower for 13 consecutive trading days—a record—and becoming the worst-performing component of the Dow Jones Index for the year. Bernstein downgraded its rating, Barron’s withdrew its recommendation, and the short interest ratio rose as high as 6.4% of the float.

But this earnings report on August 26 slapped a sharp, clear slap across the face of all short-selling narratives.

The key metric driving a 180-degree shift in market sentiment is Agentforce. This AI agent platform, which was mocked by institutional investors a year ago as a "PPT product," has now surpassed $1.5 billion in annualized recurring revenue this quarter, a 240% year-over-year increase. Combined with Data 360 and Informatica Cloud, Salesforce’s entire AI and data business is on the verge of crossing the $4 billion ARR threshold.

More critical data is hidden in usage.

In Q2, customers consumed 3.2 billion Agentic Work Units through Agentforce and Slack, a 97% increase quarter-over-quarter. The number of Slackbot users surged more than 150% quarter-over-quarter. Orders for Agentforce Advanced and App editions doubled quarter-over-quarter. These figures clearly indicate that Agentforce is no longer a demo tool that customers try and then set aside—it is becoming a true productivity tool embedded in enterprise workflows.

The signals at the contract level are equally strong.

The current remaining performance obligations (CRPO) reached $33.5 billion, representing a 14% year-over-year growth at constant exchange rates; CFO Robin Washington called this the strongest quarter for net new annual order value in four years.

The announcement of the Anthropic partnership on the same day elevated this story to a new level. Salesforce and Anthropic jointly launched "Claudeforce," directly integrating Salesforce’s data, workflows, and business logic with Claude via the MCP protocol, with 37 pre-built sales skills launching in the first phase.

Benioff and Dario Amodei appeared together on stage; Patrick Stokes, President of Sales and Marketing at Salesforce, made a statement worth reflecting on: "When people no longer use Salesforce through traditional interfaces but through agent-based interfaces, Salesforce’s value will significantly increase."

A UI company publicly acknowledged, "We no longer need UI," sending a strategic signal that Salesforce no longer positions itself as "the software you use," but rather "the database that AI calls upon."

Returning to Intuit, with similarly exceeded earnings and the same declaration of embracing AI, why did the market assign a completely opposite valuation?

The answer lies in a harsh arithmetic problem. Wall Street institutions calculate that the cost of processing a tax filing token using AI is approximately $0.12, while TurboTax charges users an average fee of $162. This 1,350-fold price difference represents the structural risk facing Intuit.

The value of TurboTax lies in "translating complex tax rules into an interface guide that ordinary people can understand," which is precisely what large language models excel at. In June this year, Goldman Sachs downgraded Intuit's rating to sell, slashing its target price from $519 to $276.

Salesforce faces a completely different situation. Its CRM system holds decades-worth of accumulated customer relationship data, sales pipelines, business rules, and approval workflows—assets that AI cannot replicate. The more powerful AI agents become, the more they rely on accessing these datasets to execute real business actions. By shifting from “selling software licenses” to “charging based on AI consumption,” Salesforce is effectively moving from selling water to charging for the water source.

This is the species divergence occurring in the SaaS industry in 2026:

One category is the "data fortress." Salesforce, ServiceNow, and SAP are embedded in the core IT, financial, HR, and customer management processes of enterprises, making replacement extremely costly.

ServiceNow's AI product annual contract value has surpassed $1 billion, and SAP's cloud backlog has grown 26% at constant exchange rates. These companies won't be replaced by AI—they'll instead charge tolls to AI.

Another category is the "function factory." Their value proposition is "helping you complete specific tasks with a better interface," while AI can now perform these tasks directly. MongoDB saw a 42% decline due to the rise of AI-assisted programming, and Atlassian’s stock plummeted nearly 40% after experiencing its first-ever drop in enterprise seat numbers.

Adobe represents an interesting intermediate case. Its primary threat comes from direct competition with AI-native creative tools, rather than seat compression. However, the ARR contribution from Adobe Firefly and AI Acrobat Assistant has tripled year-over-year, indicating that it is moving toward the “data fortress” side.

For investors following this round of SaaS pricing restructuring, the four targets form a clear observation matrix:

CRM (Salesforce): The core value of this earnings report lies not in the quarterly figures, but in its validation that the business model of "enterprise data moat + AI consumption-based monetization" is viable. The next key milestone will be the renewal rate data for Agentforce and enterprise adoption of Claudeforce following its public beta in September.

NOW (ServiceNow): Annual contract value for AI products has surpassed $1 billion; the deep integration of IT workflows has created a strong data barrier. The key question is whether platform expansion can extend from IT to HR and customer service domains.

WDAY (Workday): The HR software category is facing the most direct pressure on seat compression. If AI can handle a large volume of low-judgment HR administrative tasks, Workday’s revenue per employee could be structurally compressed. A revaluation would require a compelling path toward a consumption-based transition.

ADBE (Adobe): Competition in AI-native creative tools remains the biggest variable. Firefly’s data is worth continuous monitoring, but Adobe’s moat lies in its integration into enterprise-level creative workflows, not just in image generation capabilities.

In the AI era of 2026, the key variable determining a software company’s fate has shifted from the quality of features to the depth of data.