How AI is changing SaaS
Given recent volatility in public SaaS valuations, we've been fielding a consistent question from our investors: How robust is the SaaS business model in the face of AI disruption? More specifically, what does this mean for the profitability and credit security of our lending model?
It's a fair and timely question that deserves a thoughtful answer.
While AI will undoubtedly have significant impacts across every aspect of business, the future of the SaaS model is more nuanced than binary narratives of disruption or survival suggest. Our analysis indicates that the SaaS business model remains fundamentally sound, but requires adaptation. Traditional per-seat pricing models face pressure from both productivity gains (fewer workers needed) and API-driven usage patterns (AI agents accessing software programmatically rather than through human interfaces). However, SaaS software itself remains essential - even agentic AI systems rely on external SaaS tools via APIs to accomplish tasks.
The critical differentiator is pricing model flexibility and market positioning: companies serving vertical markets with usage-based or platform pricing will thrive, while those dependent on per-seat pricing for horizontal knowledge worker tools face revenue challenges despite their products remaining mission-critical.
This creates a bifurcated market where some SaaS segments will see accelerated growth (vertical platforms, usage-based models) while others face margin compression (horizontal tools, pure seat-based pricing). Our lending strategy focuses on companies positioned in the former category.
The real story isn't industry collapse - it's creative destruction that expands the total market. As a16z's recent analysis argues, AI will split software into two categories: companies delivering genuine value through deep integration and modern pricing models, and those relying on vendor lock-in with outdated interfaces. The former category will grow dramatically as falling code costs enable serving previously uneconomical customer segments and automating workflows that were once too complex or expensive to address. The world remains "short software" - demand will expand as costs fall.
Recent Market Events
A recent market event illustrates these dynamics: In late January 2026, Anthropic's release of industry-specific plugins for Claude Cowork triggered a $285 billion sell-off in technology stocks, with legal information providers like Thomson Reuters and LexisNexis dropping 15-20% in a single day. However, the hardest-hit companies were information providers and databases - not traditional SaaS workflow management companies. Legal research databases compete directly with generative AI (which can answer legal questions and cite cases), while workflow management software requires agentic AI to navigate complex, integrated business processes - a far more difficult challenge. This distinction reinforces our underwriting focus: we avoid information aggregators and horizontal tools, prioritizing instead vertical SaaS companies with deep workflow integration and mission-critical functionality.
To properly assess the risks and opportunities, it's important to distinguish between two fundamentally different types of AI that are shaping the software landscape.