The speaker posits that enterprise AI adoption is not following a typical S-curve but an 'L-curve,' signifying a near-vertical, explosive growth phase. With penetration currently below 1% among knowledge workers, the market is at the very beginning of a massive, rapid expansion.
The market for foundational AI models has rapidly consolidated from over 60 players to a three-company oligopoly, mirroring the evolution of the public cloud. Leaders like Anthropic and OpenAI are establishing durable moats through superior performance, scale, brand, and ecosystem lock-in.
The explosive growth in AI workloads (growing 10x annually) has created a massive, multi-year shortage of compute capacity and related hardware like advanced PCBs, DRAM, and fiber. This intense demand is 'decommoditizing' the hardware industry, creating significant pricing power for key suppliers.
The speaker's firm has divested almost all holdings in traditional application software, viewing the category as a 'horse and buggy' in the age of AI. CIOs are deprioritizing legacy software spend in favor of AI tokens and infrastructure, which offer a faster ROI, creating a major headwind for incumbents.
AI-powered coding tools represent a 'true unlock' for productivity, with a total addressable market estimated at half a trillion dollars. The rapid improvement in these tools, exemplified by Andrej Karpathy's shift to writing code almost exclusively in English, demonstrates a paradigm shift in software development.
Keep pulling the thread on Alex Sacerdote.