The question sounds absurd until you follow the money.
Payroll taxes fund most of the safety net — social security, unemployment insurance, healthcare. That system assumes a human is being paid a wage somewhere in the chain. Replace the human with a robot, and the wage disappears. So does the tax that rode along with it.
This isn’t new. Bill Gates argued in 2017 that governments should tax companies for their use of robots, both to slow automation’s spread and to fund other kinds of employment. The EU considered the same idea and rejected it. South Korea instead cut tax incentives for automated equipment — a tax by subtraction rather than addition.
What’s changed is who’s saying it now. In April 2026, OpenAI released a policy document proposing robot taxes, a shift of the tax burden from payroll to capital, and a 32-hour workweek pilot at full pay — warning that AI-driven growth could hollow out the tax base itself. The company profiting most from automation is the one flagging the hole it leaves behind.
Andrew Yang’s version is blunter: the former U.S. presidential candidate who first ran on a Universal Basic Income platform now argues the U.S. should tax the task, not the worker — a per-activity levy paid to the state whenever a robot performs work a human used to do.
Economists push back from the other side. Some argue that taxing robots is really a capital tax in disguise — the same mistake as taxing steel during the industrial revolution: you end up discouraging the investment that produces the productivity gains in the first place. Others point out a subtler problem — there’s no clean line between a self-checkout kiosk, a claims-processing algorithm, and an industrial welding arm, so any tax needs boundaries that are arbitrary by design and that companies will spend real effort routing around.
None of these ideas has become law anywhere yet.
The numbers, as far as they go
4.66 million industrial robots were operating in factories worldwide in 2024 — up 9% on the year before. China alone runs about 2 million of them, more than half of all installations globally.
The industry’s own estimate: one robot replaces roughly 1.6 factory workers. Multiply that across 4.66 million robots, and the arithmetic points to millions of wage packets that no longer exist to be taxed — no payroll tax, no income tax, no social security contribution, from work a robot now does instead.
But here’s where the picture goes soft. A widely cited McKinsey study projects 400 to 800 million jobs lost worldwide by 2030 — a projection, not a count, with 800 million the extreme end of the range. In the US specifically, confirmed AI-related job losses total around 55,000 through 2025 — smaller than the roughly 120,000 AI-linked jobs created in 2024 alone.
Which means the tax-base argument runs ahead of the evidence. Nobody is tracking, in real time, what happens to the specific tax revenue attached to an automated job. Older research on 1990s–2000s factory automation shows displaced workers mostly moved into other, often lower-paid service work — meaning less tax, but not zero. There’s no equivalent dataset for the current wave.
So the “hollowed-out tax base” argument is plausible arithmetic, not measured fact. We know the robots. We don’t yet know, with any precision, what the humans they replaced are earning — or paying — now.
And where does UBI’s money come from?
The robotics industry itself doesn’t expect this to slow down. Humanoid robots are barely present in workplaces today, but the sector’s own forecasts put annual shipments at 115,000 units by 2027 — the inflection point analysts have been pointing to. If that many more robots enter workplaces on top of the industrial robots already running, the question of what the displaced humans live on stops being theoretical.
UBI is one answer to that — and tellingly, it isn’t coming from labour unions or governments first. It’s the AI companies themselves proposing it. OpenAI’s blueprint puts a public wealth fund and a UBI-style safety net on the table, alongside the robot tax. The people building the technology are also the ones naming the parachute.
Ask “where does the money come from,” though, and you get four different answers, depending on who’s proposing it.
Andrew Yang, the former U.S. presidential candidate who ran on a Universal Basic Income platform, proposed a concrete funding mechanism: a 10% VAT — a value-added tax charged at every stage a product moves through, from raw material to finished good on the shelf. The revenue from that tax would fund a $1,000-a-month payment to every American adult over 18.
OpenAI goes further upstream: a public wealth fund, where the state holds an ownership stake in AI’s gains directly, the way Alaska’s Permanent Fund pays residents a dividend from oil revenue. Instead of taxing income after the fact, the public owns a share of the asset producing it.
The UK Green Party’s 2019 plan was blunter still: higher earners pay more tax than they receive back. Not a new revenue source — a redistribution, dressed up as universal.
And the robot tax fits here too — if automation is what’s displacing the income UBI is meant to replace, taxing automation directly closes the loop, rather than raising VAT or income tax on everyone else to cover it.
None of these are free money. Every version is really a different answer to the same underlying question: who currently holds the wealth that automation is creating, and how much of it are they willing to give up.