Descriptions:
SemiAnalysis founder and CEO Dylan Patel joins Latent Space’s “In-Context Cooking” format—recreating chicken fried rice while dissecting the AI industry—for an unconventional but substantive conversation on the economics and competitive dynamics of the current AI moment. Patel, who built one of the most widely-read independent research operations covering chips and AI infrastructure, traces his path from anonymous hardware forum moderator and beekeeper in rural Georgia to running a 60-person firm whose analysis is read by hedge funds, hyperscalers, and AI labs alike.
The core AI argument Patel makes is that the industry has crossed into genuine economic transformation territory. He points to Claude Code adoption data—GitHub commits increasingly AI-assisted, roughly a third of SemiAnalysis’s own engineers, financial analysts, and researchers now using Claude Code as a primary tool—as evidence of escape velocity. He contextualizes Google’s $180 billion and Amazon’s $200 billion capex announcements against falling stock prices, arguing the market is misreading infrastructure investment in transformative technology. The $2 trillion in annual global software developer wages is framed as a proxy for AI’s near-term addressable economic impact.
Patel also surveys the recent release wave including Claude Code, Maltbook, Kimmy K2.5 Swarms, and Codex 5.3, positioning the past several weeks as a compressed inflection point for practical AI deployment. While the cooking-show format keeps the tone light, Patel’s industry access and track record make this a useful signal read on where enterprise AI adoption actually stands.
📺 Source: Latent Space · Published February 26, 2026
🏷️ Format: Interview
