Anthropic CEO Dario Amodei has said that 50% of white-collar entry-level jobs will be taken over by AI by 2030, but Vinod Khosla has an even more extreme prediction.
Khosla — founder of Sun Microsystems, founder of Khosla Ventures, and one of OpenAI’s earliest institutional investors — believes AI won’t just nibble at the edges of the labour market. Speaking on Fortune’s Titans and Disruptors of Industry podcast, he argued that the majority of all jobs, across virtually every profession, will be within AI’s reach by the end of this decade.
“Starting in about 2030 — four years away — 80% of all jobs, so two-thirds of all jobs, will be capable of being done by an AI. Whether you’re a physician, whether you’re a radiologist, whether you are an accountant, whether you are a chip designer, whether you are a salesperson, AI will do your job better.”
Khosla envisions a transitional phase before that full displacement arrives. “There’ll be an interim period where every professional will have four AI interns. They’re training to leverage themselves. The initial model of AI deployment will be AI interns working for somebody who’s already a senior accountant, a physician, or a chip designer.”
The economic consequences, he argues, are staggering. “What happens when all labour is free? 15 trillion of US GDP is labour — 15 trillion. That would mostly go away. That’s a hugely deflationary economy, something nobody’s planning on.”
But Khosla is careful to distinguish this from the deflationary crises of the past. “It’s a different kind of deflation than past deflationary economies. The abundance of goods and services will be very, very large. Prices will be very, very low. I would suspect by 2040, $30,000 — and maybe $10,000 — will buy much more than you can buy if you have a hundred thousand dollars income today. The level of income you need in a deflationary economy will be very different.”
Implications
The implications are hard to overstate. Khosla isn’t describing disruption at the margins — he’s describing the effective end of labour as an economic input. A world where $15 trillion in GDP “mostly goes away” would require rebuilding social contracts, tax systems, and safety nets from scratch. Policymakers have barely started that conversation.
What makes the prediction credible is that real-world indicators are already pointing in this direction. Klarna stopped hiring humans over a year ago, with its CEO stating that AI could already do every job humans do — and the company has been shrinking its workforce by 20% annually through attrition. Scale AI CEO Alexandr Wang has argued that every employee will essentially be “promoted to manager,” overseeing a pod of AI agents doing the work they once did themselves — a picture remarkably close to Khosla’s “AI interns” model.
Meanwhile, Elon Musk has predicted that goods and services will become “close to free” in an AI-driven economy — echoing Khosla’s deflationary thesis almost word for word.
The harder question isn’t whether AI will take jobs — that debate is largely settled. It’s whether the abundance Khosla describes will actually be distributed broadly enough to matter, and whether governments will act fast enough to cushion the transition. Khosla himself has warned the 2030–40 period will be “really chaotic, country by country.” Without policy to match the pace of the technology, the deflationary abundance he envisions could just as easily become a deflationary crisis.