Software Stocks: A Value Investor's Guide to the AI Sell-Off

The market's reaction to AI disruption has created a stark disconnect with the financial reality of leading software companies. This year, the sector has stumbled out of the gate, posting its weakest opening stretch in years. A basket of SaaS stocks tracked by Morgan Stanley has fallen roughly 15% so far this year, following an 11% drop in 2025. That marks the worst start to a year for the group since 2022. The immediate trigger was renewed anxiety after AI startup Anthropic unveiled a new tool, Claude Cowork, on January 12. This research preview, designed to automate complex workplace tasks, reignited fears that AI agents could eventually render entire software categories obsolete.

The fear is palpable. Tools like Anthropic's Claude Code are seen as a direct threat to legacy software, offering fully autonomous coding that could eliminate the need for expensive licenses. This has sent shockwaves through the sector, with stocks like Intuit and Adobe plunging sharply last week. Yet, for a disciplined investor, the core question is whether this fear is pricing in a distant, uncertain future or a present-day reality. The financial metrics of established leaders tell a different story.

Take Workday, a leader in enterprise software. Its fiscal third-quarter results, reported last month, showed subscription revenues up 14.6% year-over-year and total revenue growth of 12.6%. The company is not just growing; it is compounding profitability, with operating income and non-GAAP earnings per share both rising significantly. This is the kind of durable, high-margin growth that builds intrinsic value over decades. The market's reaction suggests it is discounting this proven performance in favor of a hypothetical disruption.

The AppLovin case study drives this point home. The company's stock is currently trading nearly 22% below its 52-week high. That sell-off is hard to reconcile with the explosive growth of its AI-powered Axon-2 platform, which grew 68% year over year last quarter. This segment is expanding gross margins and profitability, providing a clear growth catalyst. The market is punishing the stock for sector-wide fears, even as one of its key growth engines accelerates.

The tension here is classic value investing material. The market is pricing in a potential future where AI agents displace software, creating a wide moat for the new entrants. But the financials of today's leaders show they are not passive victims. They are adapting, innovating, and compounding value in the present. For a patient investor, the sell-off may represent an overreaction to noise, creating a potential opportunity to buy quality at a discount. The real test will be whether these companies can maintain their moats and convert their strong financials into long-term compounding.

Assessing the Moat: Can AI Agents Truly Displace SaaS?

The fear that AI agents can automate entire workflows and eliminate the need for software licenses is understandable. The vision of a tool like Anthropic's Claude Code writing, testing, and debugging entire software stacks autonomously is a direct threat to the SaaS model. Yet, for a value investor, the critical question is not about the theoretical potential, but the practical durability of the economic moat. The evidence suggests that the current wave of AI agents faces significant hurdles that will likely slow their displacement of established software for years to come.

First, these tools are still in their infancy. Anthropic's latest offering, Cowork, is explicitly a research preview, available only to a limited set of subscribers with a waitlist for others. It is not a finished product but an early experiment in making agentic AI more accessible. This early stage is crucial. The market is pricing in a distant future where these tools are ubiquitous and flawless, but the reality is a beta tool with known risks. As Anthropic itself warns, the approach of giving a tool vague instructions can be potentially dangerous, risking prompt injection or accidental file deletion. This security and reliability gap is a major barrier to enterprise adoption.

Second, the user adoption curve is steeper than the hype suggests. While tools like Cowork are designed to be less intimidating for non-technical users, the experience of testing them reveals a pattern of overpromising and underdelivering. A reporter who has tested many such agents notes they often struggle to complete even basic tasks. The fact that Cowork is a "nice surprise" in this context highlights how poor the track record has been. For a business to replace a proven, integrated software suite with a new, unreliable agent, the cost of failure is simply too high. The need for clear, unambiguous instructions is a fundamental friction that limits broad utility.

This is where the response from established players becomes telling. UiPath, a leader in automation, is not waiting for disruption. It is actively building enterprise-grade governance and security into its agentic platform as a direct counter to the chaos of early agents. The company's recent expansion of its platform includes built-in security and governance to accelerate ROI and simplify deployment. This is a classic moat-building move: taking the disruptive technology and embedding it within a trusted, compliant, and scalable framework. UiPath is not being replaced; it is evolving to own the agentic automation space.

Finally, consider the case of Salesforce. The company faces legitimate disruption fears, as noted in the sector-wide sell-off. Yet its moat is not just in its CRM software; it is in its vast ecosystem, deep customer relationships, and the immense switching costs for enterprises. The threat from AI agents is real, but it is a threat to specific workflow components, not the entire enterprise relationship. Salesforce's scale and integration give it a formidable advantage in adapting these new tools into its platform, rather than being displaced by them.

The bottom line is that AI agents are a long-term, not an immediate, threat. The technical hurdles, user adoption friction, and the established players' response all point to a gradual evolution, not a sudden obsolescence. For a value investor, this means the wide moats of leading SaaS companies are not yet under siege. They are being tested, but the financial reality of compounding profitability and customer lock-in remains intact. The sell-off may be overestimating the speed of disruption while underestimating the resilience of the moat.

Valuation and the Margin of Safety

The sell-off has compressed valuations to a point where a margin of safety may finally be reappearing. The market's fear-driven reaction has created a disconnect between price and the underlying financial strength of many leaders. Consider Workday, a company with a wide moat in enterprise HCM and finance. Despite reporting a solid fiscal third quarter with subscription revenues up 14.6% year-over-year and a significant expansion in profitability, its stock is not immune to sector-wide anxiety. Yet, the price action suggests the market is pricing in a distant disruption scenario rather than today's compounding reality. The stock's forward price-to-sales multiple sits below 5x, with a forward price-to-earnings ratio of 18. These are not the valuations of a company facing imminent obsolescence; they are the marks of a durable business trading at a discount.

Analysts are beginning to see constructive longer-term outlooks, a shift that could provide the catalyst for a valuation reset. Barclays recently stated that software stocks are poised to finally catch a break in 2026. This view is grounded in the expectation that the sector's fundamental growth-driven by recurring revenue and high margins-will eventually outweigh the noise of technological disruption fears. The key for the patient investor is to distinguish between companies with wide, durable moats and those more exposed to commoditization. Workday's moat is built on the largest and cleanest HR and finance database, creating immense switching costs and a trusted platform for AI agents. This is a moat that can be strengthened, not eroded, by the very technology that scares the market.

The case study of Workday's Q3 results shows the financial engine is still powerful. The company's non-GAAP operating income margin expanded to 28.5% in the quarter, a clear sign of operational leverage and pricing power. This profitability fuels its $6.84 billion cash position and enables strategic moves like its recent acquisition of Paradox. For a value investor, this is the essence of intrinsic value: a business that is not just growing, but growing profitably and compounding its capital efficiently. The current sell-off, while painful, may be offering a rare opportunity to buy such a business at a price that does not fully reflect its durable competitive advantages.

The bottom line is one of patience versus panic. The market is pricing in a future of rapid, disruptive change, but the financials of today's leaders show a different story of steady, high-margin growth. The margin of safety here lies in the gap between that fear and the reality of expanding profits and strong balance sheets. For those willing to wait, the current setup offers a chance to acquire quality at a discount, provided they focus on the companies with the widest moats.

Catalysts and Watchpoints for the Thesis

For the patient investor, the current setup hinges on a few key watchpoints. The thesis that the sell-off is overdone depends on whether AI agents remain a distant threat or begin to materially accelerate disruption. The coming months will provide the evidence.

First, monitor the commercial adoption and technical reliability of tools like Anthropic's Cowork. The market is pricing in a future where these agents are ubiquitous and flawless. The reality is a research preview with known limitations. The early user experience is telling: a reporter who has tested many such agents notes they often struggle to complete even basic tasks. The fact that Cowork is a "nice surprise" in this context highlights the poor track record. For enterprise adoption, the cost of failure is too high. The watchpoint is whether these tools move from beta experiments to reliable, integrated parts of business workflows, or remain niche, unreliable helpers.

Second, watch for signs that AI disruption is materially accelerating, such as a sustained shift in enterprise software spending budgets. The sector-wide sell-off is driven by fears that AI companies like OpenAI and Anthropic will emerge as direct competitors. Yet, the financial reality of today's leaders shows they are adapting. The case study of UiPath is a direct response to this threat. The company is not waiting to be replaced; it is actively building enterprise-grade governance and security into its agentic platform. Its recent expansion includes built-in security and governance to accelerate ROI and simplify deployment. This is a moat-building move, aiming to secure its enterprise moat by embedding the disruptive technology within a trusted framework. The watchpoint is whether other established players follow suit, or if the market sees a clear bifurcation where only the agile incumbents thrive.

Finally, the next earnings reports will provide updated financial performance and management commentary. AppLovin's Q4 results, scheduled for February 11, 2026, are a prime example. The company's stock is down over 20% from its highs, yet its AI-powered Axon-2 platform grew 68% year-over-year last quarter. The earnings call will reveal if that growth is sustainable and whether management sees any near-term pressure from AI agents. More broadly, the results from leaders like Workday and Salesforce will show if their high-margin, recurring revenue models are holding firm. A sustained beat on profitability and guidance would confirm the financial engine is still powerful, while a notable deceleration in growth or margin could signal the disruption fears are materializing faster than expected.

The bottom line is that the margin of safety depends on these catalysts. The current price may be a discount to intrinsic value, but the gap will narrow only if the evidence shows AI agents are not yet a credible substitute for proven enterprise software. The coming quarters will test that assumption.