Key Takeaways

  • The rapid commoditization of AI model development means that winning in the AI era isn't about having the absolute best product, but about securing distribution.
  • Long-time tech analyst Benedict Evans compares AI models to web browser rendering engines: many can do the job, so the company that controls the user's default experience wins.
  • Incumbents like Google, Meta, and Microsoft are actively leveraging their massive existing user bases and platform ubiquity to push their AI offerings, making it difficult for new players to gain traction.
  • Even an "adequate" AI product can achieve significant market share if it's "sprayed on every service surface" of a platform, as seen with Meta's Llama in early usage surveys.

When Every Product Is 'Good Enough,' Distribution Reigns Supreme

For years, founders preached product-market fit. Build something amazing, and users will come. But in the AI era, that wisdom is breaking down. Benedict Evans, a sharp tech analyst, made a provocative claim on Lenny's Podcast: “if the product is a commodity, then the distribution is what matters.” He's not talking about the next marginal feature; he's talking about the entire playing field.

Think about it: building software is getting easier. AI models are improving rapidly and becoming interchangeable. As Lenny Rachitsky observed, “everyone's launching products... the noise in the market is just going up like crazy.” When everyone can build something that's 'fine,' the battle shifts from feature sets to access. Who can get their 'fine' product in front of the most eyeballs? That's the new moat.

Incumbents aren't sitting still. Evans pointed to Google pushing Gemini through its search and Android ecosystem. Meta, a company many in tech had "completely written off" for AI, sprayed its Llama models “on every service surface” of its platforms. What happened? Meta's AI shot up in usage surveys, positioning itself “between ChatGPT and Gemini” not because it was revolutionary, but because it was everywhere. It was "fine." Apple, with its “billion devices that can run this on edge,” holds similar power. These giants don't need the best AI; they just need one that's good enough and has a direct line to billions of users. This isn't a fair fight for a startup that has to build product and distribution from scratch.

What to Do With This

Stop obsessing over marginal product superiority if you haven't solved distribution. This week, list three unconventional ways your startup could build or acquire non-linear distribution before spending another sprint cycle on product features. Could you embed deeply into an existing, non-AI platform? Can you cultivate a community so specific and loyal that they become your 'spray surface'? Or, do you need to pivot to a niche where incumbents have no direct distribution channel, allowing your superior product a chance to breathe and grow its own user base?