The speaker, Benedict Evans, argues that foundation models lack strong network effects and are on a path to becoming commodities. This will lead to intense price competition and margin compression for model providers like OpenAI and Anthropic, making the market structure more analogous to cloud infrastructure (AWS) than a winner-take-all platform (Windows).
The current state of AI is frequently compared to previous technology shifts, particularly the internet in 1997. This analogy highlights the uncertainty, the immaturity of the technology, and the high probability that the current leaders may not be the ultimate winners, just as early search engines like Yahoo were eventually displaced.
As the underlying AI technology becomes more accessible and commoditized, the primary competitive advantage shifts to distribution. Incumbents like Google and Meta can leverage their massive user bases and existing products to deploy AI features, making it difficult for new startups to compete for attention and market share.
AI's adoption and impact are not uniform across all sectors. For software developers, AI tools are already transformative, akin to the introduction of VisiCalc for accountants. However, for other professions like law, adoption is slower and more cautious due to concerns about reliability and hallucinations.
Keep pulling the thread on Benedict Evans.