Sequoia Partner Grady Booch: Entrepreneurs Need to Build Moats and Distribution Capabilities Against AI Labs
Grady Booch, a partner at Sequoia, defined the current phase at the AI Ascent event as the "AGI moment" and suggested that startups should prioritize building three elements: "Moats, Affordances, Distribution" when competing with large AI labs. He emphasized that product strategy must be driven by customer needs rather than technology.
This judgment is based on the rapid commoditization of AI capabilities. Several English media outlets and research institutions have pointed out that the capabilities of large models are being standardized quickly, with labs like OpenAI, Google DeepMind, and Anthropic continuously lowering the barriers to access these capabilities, thereby narrowing the competitive space for startups that rely solely on model capabilities. Sequoia previously defined the potential market size in the AI sector as being in the trillions of dollars and emphasized that distribution and application layers are key to value capture.
Additionally, several founders and investors in AI companies have recently reiterated on English podcasts and social media that true sustainable advantages are shifting from "model performance" to "user relationships, data loops, and channel control," echoing Sequoia's framework.
Source: Public Information
ABAB AI Insight
This statement's core is not about whether "AGI is coming," but about the reallocation of capital in value capture positions. The foundational model layer is undergoing a typical technology diffusion path: rapid performance improvement, continuous cost reduction, and supply expanding from a few to many. At this stage, model capabilities are gradually shifting from "scarce assets" to "infrastructure," similar to the early evolution of cloud computing towards general computing power.
As underlying capabilities become homogenized, profits and pricing power naturally shift to the application and distribution layers. The so-called Moats and Distribution essentially represent the competition for "user entry points" and "demand control," consistent with the historical competition logic of operating systems, search engines, and mobile app stores: it is not about who has the strongest technology, but who controls users and traffic.
"Customer-back" points to another structural change—AI is transitioning from a technology-driven industry to a demand-driven industry. In the early stages, engineering capabilities defined product boundaries, but in the oversupply stage, what is truly scarce is the understanding of specific scenarios and the ability to build data loops. This means that the barriers to entry for startups have not lowered; instead, they have shifted from "training models" to "organizing real-world demands."
On a deeper level, this is a typical redistribution of the value chain: capital-intensive foundational model companies bear high costs and uncertainties, while the application layer captures cash flow and user relationships through a lighter asset structure. Sequoia's framework essentially guides entrepreneurs to avoid direct competition with giants in capital and computing power, instead occupying a more stable structural position.