What Happened
TechCrunch surveyed venture capitalists on March 1, 2026, about what they are no longer funding in the AI SaaS category. The consistent pattern across investor responses: firms are passing on companies that amount to thin wrappers around foundation models without proprietary data, network effects, or workflow integrations deep enough to create switching costs. The bar that was sufficient to raise a seed round in 2023 and 2024 - a compelling demo with GPT-4 and some growth traction - is no longer clearing the threshold at most firms.
Why It Matters
The VC perspective matters because it shapes which AI startups get funded, which affects what products reach market and at what price points. The categories still attracting capital, based on investor commentary, tend to involve vertical-specific data advantages in legal, medical, or financial domains; deep workflow integrations that create genuine lock-in; or infrastructure that sits below multiple applications in the AI stack.
Pure-play chatbot interfaces for general tasks and writing assistants without proprietary data are the categories that have lost the most VC enthusiasm. The reason is structural: foundation model providers - OpenAI, Anthropic, Google - keep improving their native interfaces and lowering their API prices. A product that was differentiated on top of GPT-3.5 in 2022 often isn't differentiated on top of GPT-4o in 2025.
For founders building AI tools, the investor message is direct: access to AI isn't a business model, it's a distribution channel. The product question is what you build that the model providers can't or won't build themselves, and whether you can build it fast enough that the window doesn't close before you reach sustainable revenue.
Existing companies built on the wrapper model are facing the same pressure from the other direction - not from VCs cutting off their funding, but from their own customers asking why they should pay for a layer on top of models they can access directly.
Our Take
This isn't a surprise, but the public acknowledgment from investors matters for founders still planning in this space. The AI SaaS market is undergoing the same selection that happened to cloud SaaS around 2012-2015: the generic plays get commoditized by platform providers, and the vertical specialists with real distribution and data advantages survive. The companies that have built proprietary workflows, user-generated training data, or deep integrations with adjacent systems are in a structurally different position than those that haven't.