DEPLOY LOCALLY

“Marc thinks local models, Apple Silicon, privacy, trust, and economics all point toward a major role for edge AI. Local inference allows for privacy where trust is paramount, and the economics of running these models on local hardware will eventually shift the market away from centralized bottlenecks as capacity continues to be a constraint.”

“The way I think about what's happening is basically the period we're in right now is what I call an 80-year overnight success. It's an overnight success because bam, ChatGPT hits and then O1 hits and then OpenClaw hits, and these are radical overnight transformative successes, but they're drawing on an 80-year wellspring backlog of ideas and thinking. It's an unlock of all of these decades of very serious hardcore research where scientists worked for 40 years and never saw it work, but they were fundamentally correct.”

“This time is different because of the jump from LLMs to reasoning, coding, agents, and recursive self-improvement. These breakthroughs make AI real in a way prior cycles were not, as it finally moves beyond simple pattern matching into actual utility and self-modifying software capabilities that can redefine what software even is.”

“Marc’s comparison between today’s AI capex boom and the fiber/data-center overbuild of 2000 highlights why he thinks this cycle is different because the buyers are huge cash-rich incumbents and demand is already here. Unlike the dot-com crash where infrastructure preceded demand, the current build-out is driven by companies with massive balance sheets and immediate use cases.”

“The combination of LLM plus shell plus filesystem plus markdown plus cron loop is one of the biggest software architecture breakthroughs in decades. Agents are the new Unix, where state lives in files allowing portability across models and runtimes, creating self-modifying agents that can extend themselves and change the fundamental nature of compute.”