Andreessen says AI arrives after 50 years of stagnant productivity, making human workers a premium.
He describes a "Mexican standoff" among PMs, engineers and designers, each thinking AI lets them do the other two jobs.
Andreessen says AI could enable the one-person billion-dollar company, noting open-source equivalents emerge within a year of GPT-3.
Andreessen says AI arrives when the economy needs it
Marc Andreessen argues the AI boom arrives at a historic moment. US productivity growth has been slow for 50 years and fertility is falling, so AI and robots could ease a demographic squeeze and make human workers more valuable. He doubts mass job loss, expects tasks rather than jobs to change, and sees startups shifting in what they sell and what they are.
Takeaways
AI arrives after 50 years of stagnation. US productivity growth ran at half the 1940-1970 pace and a third of 1870-1940. Fertility is falling and many countries including the US are below replacement. Andreessen says AI and robots arrive exactly when needed making human workers a premium rather than a discount.
Mass job loss is unlikely. Andreessen says it would require productivity growth of 10 to 50 percent a year, far beyond any economy. Even then, collapsing prices would make everyone richer and safety nets cheaper. He prefers task loss to job loss because a job outlasts its tasks.
Moats are uncertain as models commoditize. Within a year of GPT-3 there were free open-source equivalents. Claude Code built Cowork in a week and a half, Copilot was lapped, and Google, Anthropic, xAI, Meta, DeepSeek and Kimi all compete. Andreessen reports one theory that the big labs have no secrets.
AGI is a basket of tasks. Rachitsky says Anthropic's co-founder described AGI as about 10 or 15 of the most valuable economic tasks, not every valuable task. Andreessen predicts machine IQ rising from 130-140 to 160, 180, 200, 250 and 300, with AI doctors and lawyers better than the best humans.
Firm raised largest fund, backs indeterminate optimism. Rachitsky says Andreessen Horowitz raised the largest fund in history. Andreessen uses Peter Thiel's framework to describe its strategy as indeterminate optimism, contrasting with Elon Musk as a determinate optimist. Founders deserve 99.9999% of the credit.
How AI changes startups
AI redefines products. Andreessen asks whether Photoshop survives Nano Banana.
AI changes headcount. He asks whether a founder with budget for 100 coders needs 100 super-empowered AI coders or only 10.
AI changes what a company is, up to the one-person billion-dollar outcome and AI businesses on the blockchain.
The outcome is unresolved. Andreessen calls it a complex adaptive system where technology, regulation, choices and capital all matter.
Andreessen's advice for individuals
Treat the task, not the job, as the atomic unit of work. Jobs persist while their tasks change, as with executives and secretaries.
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Become a super-empowered individual with depth in one domain and AI-enabled breadth in others. Scott Adams' Dilbert and Larry Summers' advice against fungibility apply.
Expect a Mexican standoff among core product roles, each believing AI can do the others' jobs. Talented people may master all of them.
He is homeschooling his 10-year-old and teaching him to use AI as a tutor. His son uses Claude and ChatGPT and does vibe coding on Replit.