Salesforce.com (CRM.US), the U.S. cloud software giant specializing in customer relationship management (CRM) software, has issued a current-quarter performance outlook that slightly missed Wall Street analysts' consensus expectations. This has heightened concerns among investors who, since February, have been worried about the potential for artificial intelligence to disrupt the entire software industry under the pessimistic narrative of "AI Disrupts Everything" sparked by Anthropic. Data from Goldman Sachs' 13F filings show that catalyzed by this "AI Disrupts Everything" theme, hedge funds significantly sold off software stocks while aggressively increasing their exposure to semiconductors in the first quarter. Hedge funds' semiconductor long weight rose to its highest level since Goldman Sachs began tracking the data, while their software weight fell to its lowest since 2019. With the leading CRM software provider Salesforce issuing a tepid outlook, software stocks may face another round of intense selling pressure.
The below-expectations performance outlook and RPO (Remaining Performance Obligation) data from Salesforce have triggered not just concerns about a single software company's results, but collective investor skepticism towards the entire application software sector. The fear is that if AI Agents like Claude Cowork and OpenClaw can directly complete entire traditional software workflows—such as customer service, sales, coding, data analysis, and backend analytics—then the profit growth models, and even the entire business models, of traditional SaaS software vendors could be completely upended. These models rely on per-seat licensing, module-based expansion, and manual workflows. Consequently, the synchronized selling of software stocks and pivot towards semiconductors by global hedge funds and mutual funds since the beginning of the year, as shown by Goldman Sachs' fund flow data, represents the capital market's expression of this disruptive logic.
In an outlook statement released on Wednesday Eastern Time, Salesforce indicated that for its second fiscal quarter ending in July, the midpoint of its total revenue forecast is approximately $11.3 billion (within a specific range of $11.27 billion to $11.35 billion). This compares to Wall Street analysts' average expectation of $11.4 billion—an expectation that had even been revised down since February.
Regarding its first fiscal quarter performance metrics, Salesforce's overall results remained on a growth trajectory. However, its forward-looking indicators and second-quarter guidance were insufficient to alleviate valuation fears stemming from the "AI disrupts software stocks" narrative. For Q1 of fiscal year 2027, the company's total revenue was $11.133 billion, a 13% year-over-year increase, surpassing market expectations of approximately $11.05 billion. Adjusted EPS was $3.88, significantly higher than the market expectation of about $3.13. The company also reported that its Q1 Remaining Performance Obligation (RPO)—a key metric for future sales—was approximately $67.9 billion, an 11% year-over-year increase, compared to the Wall Street analyst average expectation of $68.9 billion.
The below-expectations guidance for the next quarter and the quarterly RPO metric highlight that the company's actively promoted AI agent tool, Agentforce, has failed to gain strong acceptance from major clients compared to the viral spread of global enterprise workflows by popular tools like Claude Cowork and OpenClaw (formerly known as Clawdbot, Moltbot).
From CRM Dominance to AI-Era Stress Test: Software Stock Valuation Logic Reshaped by AI Agents Salesforce.com, the absolute leader in customer management software manufacturing, is currently under significant pressure from emerging AI forces like Anthropic to prove it can thrive in the AI era. The company is promoting Agentforce, its proprietary AI agent tool focused on agentic AI workflows, designed to handle a series of core enterprise operational tasks like customer service without human supervision. However, media reports last week suggested the capabilities of this AI agent product do not always match Salesforce's advertising claims.
Due to the disappointing outlook and RPO figures, the company's stock fell nearly 4% in after-hours trading. It closed Wednesday's regular session at $177.510. The stock has fallen approximately 33% year-to-date, a decline similar to other traditional software giants like ServiceNow Inc. and Adobe Inc., significantly underperforming both the S&P 500 and the Nasdaq 100 indices. Salesforce's over 30% cumulative decline year-to-date reflects the "higher growth proof threshold faced by traditional software/internet operations companies in the AI era."
The pessimistic "AI Disrupts Everything"基调 since February stems primarily from growing market concerns that AI agent workflows, like those popularized by Claude Cowork and OpenClaw, could undermine entire software empires built on SaaS per-seat subscription revenue models. This led to rare selling pressure on software stocks in February, with subsequent sell-offs quickly spreading to cybersecurity, online education, traditional finance, insurance, real estate, transportation systems, and any other industry perceived as reliant on per-seat revenue or labor-intensive business models—sectors the market believes will be thoroughly disrupted by AI.
Salesforce management currently expects Agentforce to contribute $1.2 billion in annual revenue, slightly higher than the $800 million forecast in February. The company stated that overall usage of AI large language models within the Salesforce platform more than doubled compared to the previous quarter. The earnings report also showed that the combined Annual Recurring Revenue (ARR) for Agentforce and Data 360 is close to $3.4 billion, representing over 200% year-over-year growth. Agentforce and Slack collectively delivered 3.8 billion Agentic Work Units, a 111% sequential increase, and processed over 28.6 trillion tokens, a 152% sequential increase. Data 360 ingested 52 trillion records in the single quarter, a 136% year-over-year increase, with Zero Copy ingestion accounting for 35 trillion records, a 277% year-over-year increase.
As AI agents like Claude Cowork and OpenClaw increasingly penetrate enterprise operations, the market is demanding that Salesforce prove more rapidly that Agentforce can become the new entry point for the AI era, rather than being bypassed by AI-native agents. Clearly, Salesforce's AI-related business is a bright spot, but the current performance expectations and the closely watched RPO metric failing to meet market consensus highlight that Salesforce's AI-driven growth trajectory is not yet large enough to reverse the pessimistic "AI Disrupts Everything" narrative.
Agentforce is not an ordinary chatbot; it is Salesforce's product aimed at upgrading CRM, Data 360, Slack, and enterprise workflows into an "AI Agent execution layer." Its goal is to have AI agents handle tasks like customer service, sales follow-up, data queries, and ticket routing on behalf of humans. Therefore, Agentforce is essentially Salesforce's AI agent platform/tool for enterprise agentic AI workflows: it focuses on customizable autonomous AI Agents that connect to enterprise data and automatically answer questions, execute actions, and advance tasks in scenarios like sales, customer service, marketing, and commerce. Salesforce officially describes Agentforce as a "proactive, autonomous AI application," stating it allows businesses to build autonomous AI agents that can take action across business processes.
Salesforce Executive Vice President Mike Spencer stated in an interview, "Some customers are becoming more comfortable with AI, and that's helping to boost the company's performance. We are seeing a level of adoption and usage that is anchoring acceleration." For the financial quarter ended April 30, Salesforce's total revenue grew 13% to $11.13 billion. This result was boosted by $444 million in revenue from the November acquisition of data software company Informatica. According to data compiled by institutions, profit excluding certain items was an adjusted EPS of $3.88, above analysts' average expectation of approximately $3.13 per share.
Analyst Raimo Lenschow from the Wall Street financial giant Barclays PLC (BCS) wrote in a report that Agentforce's performance clearly failed to boost the overall data. "We're not sure if this is enough to drive a meaningful market reaction," Lenschow stated. Salesforce Chief Financial and Operating Officer Robin Washington said in a statement that the company still expects revenue growth to accelerate in the second half of the year, driven by AI agents. Earlier this month, the company indicated it would provide less detailed information on product-specific revenue. It will now report growth data under two main categories: Applications, and Infrastructure & Data. Management stated that, excluding the impact of currency fluctuations, Applications revenue grew 7%, while Infrastructure & Data revenue grew 23%.
'AI Disrupts Everything' Panic Sweeps Software Industry! Anthropic's Soaring Valuation Behind Repricing of Software Application Layer in Agent Era The core reason the "AI Disrupts Everything" narrative led by Anthropic has severely impacted Salesforce is the market's concern that the traditional SaaS model of "per-seat fees + manual workflows + CRM system as entry point" is being repriced by AI Agents. When agent products like Claude Code, Claude Cowork work agents, and OpenAI Agent begin directly completing tasks such as sales, customer service, coding, documentation, and data analysis, investors fear enterprises will stop paying for numerous SaaS seats and instead shift to an AI-native work layer with "pay-per-task/Token/result" models.
From a software engineering perspective, the threat from Agents like Claude Code/Claude cowork is not "replacing a CRM database," but replacing the operational layer between users and software: In the past, salespeople, customer service representatives, and operations staff needed to click, input, query, create reports, and trigger workflows within Salesforce. In the future, AI Agents might directly call APIs, read enterprise data, generate action recommendations, and automatically execute tasks. In other words, the real challenge facing Salesforce is not database replacement, but the "interaction interface, automation logic, and workflow control" being shifted upward to AI Agents.
Some academic research also shows that Agents like Claude Code possess architectural features for running commands, editing files, calling external services, and executing tasks in loops, which forms the technical foundation for Agentic AI to invade traditional enterprise software workflows. Enterprises' urgent need to improve efficiency and reduce operational costs has recently greatly accelerated the widespread adoption of two core categories of AI application software: generative AI applications and AI agents. Among these, AI agents (AI Agents) capable of autonomously executing various tedious and complex tasks are likely the ultimate AI application trend for the next decade. The emergence of AI agents signifies that artificial intelligence is evolving from an information assistance tool into a highly intelligent productivity tool. This is also why Anthropic's valuation has skyrocketed to $1 trillion, surpassing OpenAI.
The heavyweight AI agent tool launched by Anthropic was a significant trigger for the panic selling in software stocks beginning in February. More accurately, it's not that Anthropic "single-handedly disrupted all software stocks," but rather its Claude agentic AI tool made the market suddenly realize: large model companies are fundamentally disrupting the enterprise application layer from their position as "underlying model suppliers," potentially eroding the profit pools of traditional SaaS companies focused on numerous niche areas like legal, sales, marketing, data analysis, and low-code editing.
AI Disruption Narrative Splits Tech Stocks, Capital Votes with Its Feet: Hedge Funds Massively Sell Software, Pour into Semiconductors The current market is repricing AI from a "software efficiency tool" to a "disruptor of software business models + amplifier of semiconductor capital expenditure." This is why Goldman Sachs' 13F data tracking shows U.S. hedge funds and large mutual funds reached a rare consensus in Q1—massively selling software and pouring into semiconductors, pushing semiconductor long weight to a historical high. Global capital is actively betting on segments of the global AI value chain with harder logic, greater scarcity, and higher barriers to replacement, rather than traditional application software that still needs to prove its business model resilience in the Agent era. This highlights that the investment strategy of "selling software, buying semiconductors" is arguably the most perfect answer in the current stock market investment landscape.
Goldman Sachs' tracked 13F data shows institutions are significantly reducing software exposure and increasing semiconductor exposure: hedge fund semiconductor long weight rose to its highest on record, while software weight fell to its lowest since 2019; mutual fund software positioning fell to its lowest since 2012, and excluding Microsoft, mutual funds' relative over-weighting of semiconductors versus software also reached its highest since 2012. In other words, capital is flowing from the "application layer that may be reconfigured by AI Agents" to the "computing power foundation that AI Agents must rely on."
From an engineering perspective, this rotation has a clear technical basis. The stronger the capabilities of AI Agents, the greater the demand from enterprises for AI GPU/AI ASIC-led inference computing power, and for semiconductors closely related to AI computing power such as data center high-performance CPUs, DRAM/NAND/HBM memory, AI PCBs, liquid cooling systems, advanced packaging equipment, gas turbines, and data center optical interconnect systems. Conversely, the more software companies need to prove they won't be replaced by Agents, the more they need to purchase additional computing power, integrate stronger models, or use AI computing infrastructure to rebuild their product architectures. This means the flip side of "AI disrupting software" is precisely "AI strengthening chip demand." Software stocks bear the pressure of having their valuation models rewritten, while semiconductor stocks enjoy the triple support of visible capital expenditure, supply tightness, and profit revisions upward.
Morgan Stanley stated that the AI computing power arms race has entered a system-level expansion phase. The institution has significantly revised its 2026 capital expenditure expectations for U.S. tech giants upward from $433 billion a year ago to $805 billion. For 2027, capital expenditure is expected to reach $1.1 trillion, another upward revision from the previous forecast of $950 billion. Furthermore, it predicts that by 2028, nearly $3 trillion in AI-related infrastructure investment will flow through the global economy, with over 80% of the spending still ahead. Morgan Stanley's latest expectations highlight that supply chain bottlenecks at the AI computing infrastructure layer have expanded from "mass purchasing of GPUs/ASICs" to "striving to simultaneously resolve the entire AI data center delivery process chain, including data center power equipment, liquid cooling, data center CPUs, DRAM/NAND/HBM, optical communication/interconnects, high-performance network interconnects, transformers, gas turbines, and more."
Comments