Editorials

The software-as-a-service industry faces an existential question: will artificial intelligence erode the competitive moats that justified premium valuations, or will it deepen them? Three companies, Adobe, Microsoft, and Salesforce—reveal starkly different trajectories in how incumbent SaaS providers position themselves against an agentic future that is already here.

Classical SaaS competitive advantages center on high switching costs, network effects, and data lock-in. But agentic AI introduces a novel dynamic: software that can reason, decide, and act autonomously. This threatens the foundation of SaaS moats in a critical way. When AI agents can span multiple tools, workflows, and integrations, the stickiness of any single application diminishes. A company using 10 disparate cloud apps today might soon deploy a unified agentic system that reduces that to three—or consolidates everything under one intelligent layer.

Meanwhile, open-source and frontier AI models (GPT-4, Claude, open-weights alternatives) commoditize the core intelligence that SaaS companies once owned exclusively. The real winner may not be the company with the best product, but the one that owns the orchestration layer, the data integration, or the trust infrastructure that makes agents actually usable at enterprise scale. This shifts the moat conversation entirely.

Microsoft's position is deceptively strong, yet increasingly expensive to justify. The company commands a P/E of 22.6x (as of April 2026) and an EV/Sales multiple of 8.9x—above historical averages and well above most peers. Yet the moat here is arguably becoming deeper, not weaker.

The Azure moat. Microsoft owns enterprise identity, office productivity, and cloud infrastructure in a bundled stack. Copilot integration across Office 365, Windows, and Azure creates a compounding advantage: once an enterprise is locked into Microsoft's ecosystem, adding AI capabilities into that ecosystem becomes the path of least resistance. The cost of rip-and-replace is prohibitive. Azure's $72.9 billion in FY2025 revenue (from Intelligent Cloud segment) and 15.3x EV/EBITDA valuation reflects this reality: Azure is the enterprise AI infrastructure play.

The challenge: justifying AI growth at scale. Microsoft's market cap of $2.8 trillion assumes continued double-digit cloud growth into perpetuity. Current fundamentals show operating income of $50B+ in Productivity and Business Processes and $28.4B in Intelligent Cloud, but much of the recent stock appreciation prices in a scenario where AI margins are higher than traditional software. If that assumption fails—if AI services turn out to have lower gross margins or if competitive intensity increases—Microsoft has less room to disappoint.

Valuation vs. history. At 8.9x EV/Sales, Microsoft trades above its long-term median. It's no longer a value play; it's a quality/growth premium. The moat is real, but it's already well-reflected in price.

Salesforce entered the AI era in a precarious position—and is making aggressive moves to escape it. The company trades at a more modest 20.7x P/E and 3.85x EV/Sales. Those multiples reflect skepticism: investors are not yet convinced that Salesforce's historical moat (dominance in CRM and business process workflows) will survive the agent transition.

The historical moat—and its vulnerability. Salesforce's lock-in was formidable: sales teams, customer service teams, and marketing departments built entire workflows around Salesforce. Switching costs were high because data, customization, and user training were deeply embedded. But the platform's complexity, the need for extensive Salesforce professional services, and the rising cost of ownership created a vulnerability that AI naturally exploits. A good agentic CRM could theoretically handle 70% of Salesforce workflows with simpler, cheaper infrastructure. The question is whether Salesforce is still Salesforce if it becomes just another cloud application in a portfolio, rather than the central nervous system of enterprise go-to-market.

The Agentforce gambit. Salesforce's response—rebranding around "Agentforce 360" and positioning itself as the orchestration layer for agentic workflows—is strategically sound but unproven. By consolidating sales, service, marketing, commerce, and data management around a single agentic platform, Salesforce is betting that it can be the "AI-native CRM." The integration of Slack as the conversational interface, plus the acquisition of Informatica for data management, signals an attempt to own the full stack: agents + data + integration + trust.

However, execution risk is high. Salesforce's customer base is not known for adopting new paradigms quickly. The company's relatively flat growth and periodic restructuring suggest internal challenges. And Salesforce's willingness to acquire aggressively (Tableau, MuleSoft, Slack) hints at difficulty building competitive moats organically.

Valuation and the bet. At 20.7x P/E and $124 billion in market cap, Salesforce is cheaper than Microsoft but not cheap. The P/E is justified only if the Agentforce transition succeeds—if the company can convince customers that moving 10-20% of their workflow onto agents (rather than replacing Salesforce entirely) will justifies ongoing premium pricing. If Salesforce becomes just a module in a broader agentic stack, the stock has further to fall.

Adobe may be the only SaaS incumbent that benefits—rather than suffers—from generative AI at scale. The stock trades at 11.5x P/E and 3.08x EV/Sales, a valuation discount to both Microsoft and Salesforce that appears unwarranted given the company's positioning.

The alignment of moats with AI. Adobe's core moat is user skill and habit in Photoshop, Illustrator, Premiere Pro, and other creative tools. These are not "use-to-switch-to-Figma" applications; they are foundational to millions of creative professionals' workflows. The barrier is not just switching cost but the skills and psychological investment of the user. A 15-year Photoshop expert has no incentive to switch to a competitor's Generative Fill tool.

Generative AI amplifies, rather than undermines, this moat. Adobe's Firefly model, integrated natively into Photoshop and other Creative Cloud apps, makes your existing skills more powerful. You don't leave Photoshop for AI; you use AI inside Photoshop. This is fundamentally different from CRM or productivity software, where an AI agent could theoretically replace large portions of the workflow.

The data and trust angle. Adobe has invested heavily in a proprietary generative AI layer trained on licensed and public-domain content, with explicit IP indemnification for enterprise customers. This is credible differentiation: a freelancer or enterprise that worries about AI copyright liability uses Adobe's licensed models, not DALL-E or Midjourney. That trust position is defensible and valuable.

The subscription stickiness question. Adobe's subscription base (Digital Media revenue of $13.8 billion in FY2025, plus $4.3 billion in Digital Experience) is large and growing. The company achieved this through three decades of building irreplaceable tools. Generative AI doesn't change that; it deepens it. A user who pays $50–100/month for Creative Cloud gets their existing tools plus native AI capabilities. There's no reason to leave.

Valuation—genuine opportunity. At 11.5x P/E and 3.08x EV/Sales, Adobe is valued as if it faces structural headwinds. But the evidence suggests the opposite: the company's moat is actually strengthening. Gross margins are strong (89% in Q2 FY2026), and the business is generating $5.1 billion in operating cash (Q1–Q2 annualized). The discount relative to Microsoft and Salesforce looks like misplaced skepticism. Either Adobe's Firefly-driven transition will succeed and justify higher multiples, or the market is wrong about the risk to creative professional software.

The narrative that "AI will destroy SaaS" misses a crucial distinction: the threat is not to software or subscriptions, but to undifferentiated, feature-parity platforms.

Categories under real pressure:

• Generic productivity tools: Email, document collaboration, basic CRM. These are vulnerable because agents can automate many workflows. Companies like Notion face risk here, though Notion's lock-in via custom templates and embedded databases provides some defense.
• Data visualization and analytics: Tableau and Looker face commoditization pressure as AI-generated dashboards and natural language query interfaces erode the value of manual report creation.
• Mid-market enterprise software: Any SaaS that is "useful but not irreplaceable" risks being subsumed into a broader agentic layer.

Categories with durable moats:

• Specialized verticalized software: Industry-specific tools (legal tech, medical imaging, financial modeling) that combine domain expertise with user habit.
• Infrastructure and AI platforms: Azure, GitHub, Databricks. The harder the problem to solve, the stickier the winner.
• Creative and design tools: Adobe, Figma (if it stays competitive), JetBrains (developer tools). These combine craftsman-like skill barriers with native AI integration.
• Customer data and trust layers: Okta, Twilio, Stripe. These own the critical junction points between enterprise systems and external services.

Salesforce's challenge is that it sits awkwardly in the middle: not specialized enough to claim irreplaceability (like legal or medical SaaS), not infrastructure-level enough to be foundational (like Azure or GitHub). It is trying to pivot into that territory with Agentforce, but it's a harder road than it looks.

The agentic future will not treat all SaaS companies equally. The winners will be those whose moat—whether skill-based (Adobe), platform-based (Microsoft), or vertical-specific—becomes harder to dislodge, not easier, once AI is integrated.

Adobe's moat is deepening because generative AI is best used inside the best creative tools, not instead of them. Microsoft's moat is holding because the ecosystem cost of leaving Office, Windows, and Azure together is higher than any single competitor can overcome. Salesforce's moat is under stress because CRM workflows, while specialized, are not as irreplaceable as they once were—an AI agent could potentially handle most of the work.

The valuation divergence reflects this reality:

• Microsoft at 22.6x P/E: Expensive, but the moat is real and strengthening.
• Adobe at 11.5x P/E: Cheap, and the moat is overlooked by the market.
• Salesforce at 20.7x P/E: Fair value, but contingent on execution of a high-risk pivot.

For value investors, the lesson is clear: the question is not whether SaaS survives AI, but which SaaS companies' competitive advantages are complemented by AI rather than undercut by it. That distinction is far more profitable to get right than any bet on the industry as a whole.