a16z Defends SaaS Moats
a16z is debunking the "SaaS is dead" narrative, arguing that moats like network effects, brand, and proprietary data remain highly defensible in the AI era. The firm also points to hardware, regulatory capture, and vertical-specific data as other enduring competitive advantages.
The recent "SaaSpocalypse" saw roughly one trillion dollars erased from software stock valuations between mid-January and mid-February 2026 alone. Companies like Workday, Adobe, and Salesforce saw valuations compress as product launches from AI firms like Anthropic and OpenAI fueled a narrative that AI agents could automate complex work, making traditional per-seat software licenses obsolete. Barclays analysts argue this panic is only half-justified, distinguishing between the vulnerable "application layer" of SaaS and the more defensible "system of record" layer. While AI can replicate user-facing workflows, it struggles to displace the foundational databases and business logic of companies like Salesforce and SAP, which become more critical as AI agents create more data touchpoints. a16z general partner Anish Acharya calls the "vibe code everything" story "flat wrong," noting that software and IT spend only accounts for 8-12% of enterprise budgets. He argues the true opportunity for the "innovation bazooka" of AI is to go after the other 90% of spend, not just rebuild existing systems at a slightly lower cost. The firm's thesis highlights a shift from software that assists labor to "software eating labor." Instead of selling a tool to a debt collection agency, an a16z-backed company called Salient uses AI to directly replace the collectors, improving debt recovery by 50% and taking a share of the outcome. This model moves away from per-seat pricing to outcome-based revenue. Another key moat is the "walled garden" of proprietary data. a16z gives the example of Vlex, a company with 26 years of Spanish legal data. After incorporating AI, its revenue jumped 5x because it could sell completed legal memos—a high-value output—instead of just access to its database. This doesn't mean incumbents are safe. Acharya also states that AI coding agents are dramatically lowering the switching costs between complex enterprise systems. However, he counters that many legacy giants have "hostages, not customers," with systems so deeply embedded that replacement is nearly impossible, allowing them to add AI features and charge more. The new playbook for startups, according to a16z, is to own a workflow and become the system of record. Simple AI feature "wrappers" get commoditized instantly. The focus is on building defensibility by owning unique data, distribution, or the entire workflow, creating businesses that can go from zero to $100M in revenue in under two years—a speed previously unheard of in SaaS.
