Vinod Khosla has a message for Washington: the technology he spent decades funding is about to make most American workers obsolete, and the only way to preserve capitalism is to stop taxing their wages entirely.
Khosla, who placed the first institutional money into OpenAI in 2019 — a $50 million investment at a $1 billion valuation, now worth an estimated $11.5 billion — has spent the past several weeks on a media tour arguing that the United States should eliminate federal income tax for anyone earning under $100,000 per year. The offset, he says, is straightforward: tax capital gains at the same rate as ordinary income. His calculation removes roughly 125 million Americans from the tax rolls and, he claims, stays revenue-neutral.
The argument rests on a stark premise. "Starting in about 2030, 80 percent of all jobs will be capable of being done by an AI," Khosla told Fortune. That represents roughly $15 trillion in U.S. labor GDP that stands to be disrupted. Worker anxiety about AI displacement has nearly doubled since 2024, from 28 percent to 40 percent. "You can't leave 80 percent of the population behind," Khosla said. "They will revoke capitalism if that happens."
It is a remarkable position from a man whose fortune was built by investing in the very technology he now says requires a fundamental restructuring of how the federal government raises money. Khosla frames this not as redistribution or charity but as political economy — the maintenance premium for keeping the system intact. If the bottom 80 percent of earners gets left behind by automation, he argues, the social contract breaks. The tax code should be restructured to reflect the new reality: in an AI age, capital is favored over labor, and the tax code should stop pretending otherwise.
The proposal has specific mechanics. Khosla's plan would zero out federal income tax for roughly 123 million to 125 million Americans earning under $100,000 per year. The revenue replacement comes from equalizing capital gains tax rates with ordinary income tax rates — currently, long-term capital gains max out at 20 percent while top ordinary income rates sit at 37 percent. His rationale: the top 1 percent of earners, those making $650,000 or more annually, account for roughly 45 percent of all capital gains. Those making more than $10 million a year pay about 40 percent of all capital gains taxes.
The Financial Times first reported Khosla's latest comments — behind a paywall — and Fortune and CityAM confirmed the substance independently.
The proposal landed at the Hill & Valley Forum, where Khosla appeared alongside Senator Maria Cantwell of Washington. Cantwell did not embrace the idea. "Congress is a little better at the near term," she said, a diplomatic demurral about a decade-away restructuring of the tax code.
The policy world has pointed reasons for hesitation. The Congressional Budget Office's fiscal outlook for 2026 shows a deficit of $1.9 trillion — meaning the federal government is already spending considerably more than it collects. The revenue-neutral claim sits uneasily against that backdrop. The Brookings Institution found that taxing automation directly would lead to slower adoption and less economic growth — not because the technology is bad, but because policy-induced friction slows deployment.
There is also a structural problem with the capital gains offset that Khosla's math does not fully address. The real preferential treatment in capital gains taxation is not the rate itself but the mechanism: the ability to buy an asset, hold it for decades, and defer — or through death, entirely escape — the taxable event. Inequality.org noted that equalizing the nominal rate with ordinary income "would leave in place the gaping buy-hold-for-decades-sell loophole," meaning preferential treatment would continue despite rate equality. The Washington Center for Equitable Growth has documented that while real capital gains in 2021 totaled almost $6 trillion — 39.2 percent of national income — much of that wealth existed on paper indefinitely without ever triggering a taxable realization. An IRS and academic study found households accumulated approximately $116 trillion in total capital gains from 1954 to 2021; less than a fifth of that was ever reported as a realized gain.
Khosla has been consistent in his framing: this is about political stability for the technology sector, not altruism. He has said the single biggest issue in the 2028 presidential election will be fear of AI. That is also, not coincidentally, the environment that produces the regulatory conditions — favorable or hostile — in which his portfolio companies operate.
The counterarguments are not merely technical. Deutsche Bank's Research Institute, using its own AI modeling tool, forecast that while 92 million jobs will be eliminated by 2030, 170 million new roles will be created — a net positive in employment terms, if an enormously disruptive one in composition. Actual 2025 hiring data provides a different signal: U.S. employers added just 181,000 jobs across the full year — the BLS annual revision, published in February 2026, revised the earlier headline figure downward to 181,000. Whether that pace reflects a maturing labor market or the early stirrings of AI-related displacement is a question nobody can answer yet. The Tax Policy Center, a nonpartisan think tank, estimates 40 percent of households, or roughly 76 million tax units, will not pay federal income tax for 2025 anyway — suggesting the zero-income-tax population Khosla wants to formalize is already a substantial reality.
What Khosla is selling is not a tax plan so much as a political argument dressed in arithmetic. The technology sector's most prominent prophet of automation is now arguing that automation requires a complete restructuring of the social contract — one where workers stop paying income tax and the rich pay more on their investment returns. Whether the numbers hold up, whether Congress would ever act, and whether 80 percent of jobs will actually be AI-capable by 2030 are separate questions. The premise itself — that an AI investor should be taken seriously as a redistributive philosopher — is the story.
