Evan Spiegel's foundational lesson from running Snap: individual software features do not create durable competitive advantage, because competitors can copy them almost as fast as they ship. Snap learned this 15 years ago; Spiegel argues the whole tech industry is now relearning it because AI makes replicating features even cheaper and faster.
Because features are trivially copyable, Snap shifted its defense away from patents and network effects toward things that are hard to replicate: ecosystem-as-moat (platforms connecting creators and developers) and Owning Core Infrastructure as Competitive Moat (a vertically integrated AR hardware stack). Spiegel: "It's certainly better than making stuff that people don't want to copy."
See product-market-fit-and-gtm-fit-interplay for the complementary claim that distribution, not the product itself, is consumer tech's harder problem.
Spiegel says Bobby Murphy started Snap's Gen AI lab roughly ten years before this interview, tied to a real-time, on-device face-swap lens — years ahead of generative AI's mainstream boom. Despite that early technical lead, Spiegel notes that whole businesses later built around face-swap technology "all went away," reinforcing the thesis that a software feature or model capability, however early or impressive, doesn't by itself create a durable business; distribution and ecosystem lock-in do.