The core value of a model router isn't just the tech, but the strategic leverage it gives developers. By providing access to the full market of models, it reduces dependency on any single provider, an advantage that side-project routers from other companies often miss.

Jason Calacanis identifies OpenRouter's key strategic asset as the data it collects on which AI models developers are using, switching to, and abandoning. This market intelligence is incredibly valuable to cloud providers like AWS and Google, making it a prime acquisition target for its data insights, not just its API service.

While early AI companies built moats on user data by being first-to-market with API wrappers, today's startups cannot. A defensible AI product now requires a true moat beyond a simple wrapper, as model providers can easily replicate basic functionality, making such businesses non-defensible.

Founders shouldn't over-engineer a moat on paper. True defensibility is often discovered, not designed. By focusing on shipping a high-NPS product that users love, a moat will naturally develop over time through emergent properties like proprietary data traces, brand loyalty, or deep user workflow integration.

Deliverect's founder knew from experience that POS companies couldn't build an effective delivery integration layer, contrary to what investors believed. This non-obvious, domain-specific insight was their core strategic advantage and moat.

In the AI era, defensibility comes from building a complex system of record, not just a thin wrapper on an LLM. Companies with a 'thick application layer' that offers standalone value are unattractive for model providers to replicate, whereas thin wrappers risk being absorbed by the platform they are built on.

In response to claims that OpenRouter lacks a moat, Didi Das argues that true defensibility comes from network effects and product-led growth. Once developers are integrated, the effort required to switch to a competitor creates a powerful, sticky advantage, even if the core technology is replicable.

Drawing from Verkada's decision to build its own hardware, the strategy is to intentionally tackle difficult, foundational challenges early on. While this requires more upfront investment and delays initial traction, it creates an immense competitive barrier that latecomers will struggle to overcome.

Businesses with a small take rate, like API wrappers, struggle to scale to venture-level outcomes despite processing huge volumes. A company like OpenRouter might process billions in inference to earn tens of millions. This model makes the path to $1B revenue exceptionally challenging, requiring near-monopolistic share or rapid product expansion.