The question keeps surfacing in tech circles and policy papers alike: if AI eliminates vast swaths of employment, shouldn’t we implement Universal Basic Income — or its more ambitious cousin, Universal High Income — to cushion the landing? The idea sounds intuitive. But spend five minutes asking who pays for it and the whole architecture starts to crack.
This is that five minutes.
The Four Funding Proposals (and Why Each Fails or Distorts)
1. Money Printing / Central Bank Financing
The most politically convenient option. The state or central bank creates money and distributes it directly to citizens. Clean in a PowerPoint, catastrophic in practice.
This is not free money. It is a reallocation of purchasing power from everyone holding the currency to whoever receives the new units first — a mechanism economists call the Cantillon effect. The AI revolution already produces a concentration of wealth in asset (corporations, equity, real estate, Bitcoin) holders. Printing money to fund UBI amplifies that concentration: assets reprice upward in nominal terms, savings in fiat erode, and the “basic income” recipients find their purchasing power tracking toward zero as prices adjust.
The state is also short its own currency — it finances deficits with debt denominated in units it can print. So money printing serves the sovereign’s balance sheet as much as it serves citizens. This is not a coincidence.
Verdict: Inflation is a hidden tax, regressive by nature. UBI-via-printer is redistribution from fiat-savers to debtors with extra steps. And there will be no fiat-savers, they are not stupid enough to stay on a sinking ship.
2. Taxing the Ultra-Productive AI Companies
This one sounds satisfying. If Google, OpenAI, Anthropic, and their successors capture trillions in productivity gains, tax them heavily and redistribute the proceeds.
The mechanism works — but only where those companies are. And here is where it breaks down for most of the world.
Consider Slovakia. Slovak citizens and businesses will consume AI services — paying in euros that get converted into dollars that flow to American corporations. The productivity gains accrue to shareholders in San Francisco, Seattle, and New York. Slovakia’s tax authority has extremely limited leverage over those companies. Corporate structures, IP holding jurisdictions, transfer pricing, and the simple fact that a software company’s “presence” in any country is largely optional — all of these mean that the country bearing the social cost of disruption is not the country capturing the tax base to pay for it.
This is not unique to Slovakia. It applies to most of Europe, all of Latin America, most of Asia. The math only works in the US, and even there it depends on regulatory capture not happening first.
Verdict: Tax incidence follows economic power, not moral intuition. Small open economies cannot tax away value that was never created domestically.
3. Taxing the UBI Recipients Themselves
Some proposals suggest taxing the productivity gains of beneficiaries — those whose labor was replaced but who now, in theory, do other things. This is circular by construction.
If the premise of UBI is that AI has made human labor largely superfluous, then there is no tax base here. You cannot fund income transfers by taxing the people receiving the transfers. This is definitionally impossible at scale. The variant where some humans still produce high-value work and cross-subsidize the rest runs directly into the brain drain and capital flight dynamics that have plagued high-tax economies for decades. The productive minority optimizes to minimize their exposure. Crypto makes this easier than it has ever been.
Verdict: You cannot bootstrap a redistribution scheme from the people being redistributed to.
4. Sovereign Wealth / Capital Accumulation
A more serious proposal: states should have been accumulating capital stakes in AI companies all along, so dividend flows fund social programs. Norway’s oil fund is the usual reference.
This has actual logic. The problem is timing and politics. Most states did not accumulate those stakes when they were cheap (the 2010s). Acquiring them now, at post-AI-boom valuations, means either massive dilution of citizen wealth through taxation or debt issuance — which circles back to problems 1 and 2. There is also a governance problem: state ownership of AI infrastructure creates incentives for political control of AI capabilities — a danger arguably worse than the inequality it was meant to solve.
It also means states should be good at stock picking – which companies to invest in? There might be a few winners while the AI index would tank. Are the bureaucrat-controlled states good at stock picking? Right…
Verdict: Good in theory, missed the window, and creates its own pathologies.
The Alternative: Prices Fall So Far That Income Becomes Less Important
This is the scenario that most UBI advocates skip past, and it deserves serious treatment.
The Austrian tradition holds that the correct response to productivity gains is price deflation — goods and services become cheaper, the real purchasing power of whatever income you have increases, and over time the standard of living rises across the board without any redistribution mechanism required.
The historical parallel is manufacturing. In 1900, a car cost more than a house. Today a functional car costs less than three months of rent in most cities. AI could do the same to knowledge work, software, education, medical diagnosis, legal research, and creative production.
If the cost of a software-built business drops from €500k to €5k… if a medical second opinion drops from €300 to €0.30… if a personalized educational curriculum drops from €40k/year to €40/year… then the denominator changes. You don’t need €80k/year to have a decent life. You might need €20k. And €20k is achievable through a much wider range of activities — including things people actually want to do.
This is not utopian. It requires that productivity gains actually flow through to consumer prices — which historically they do, when markets remain competitive. The failure mode is monopoly capture: a handful of AI companies controlling key capabilities and extracting rents rather than passing savings to consumers. Which is precisely where antitrust and open-source matter enormously.
It’s worth realizing, by the way, that the European Union is doing the exact opposite — it supports rent extraction through AI regulations. These, as in other areas, will solve nothing — those terribly dangerous AI models that are supposed to destroy the world will destroy it even without a stamp from Brussels. What AI regulations will cause is regulatory capture.
What This Means in Practice
The UBI debate is mostly a political debate dressed as an economic one. The economic question — where does the real purchasing power come from? — has no clean answer under any of the four funding proposals, especially outside the handful of countries that house the AI corporations generating the surplus.
The deflationary abundance scenario is not guaranteed, but it is more mechanically coherent. It requires:
- Open source AI staying genuinely open — so no single entity captures the capability stack
- Protocol-layer infrastructure that is not rent-extractable — Lightning, Nostr, open APIs
- Hard money that holds its purchasing power as prices fall (Bitcoin in deflation is extremely favorable)
- Competitive pressure on AI services — which is currently intense but could consolidate
The risk is not that AI eliminates income. The risk is that AI’s productivity gains get captured by a few actors who then extract rents instead of passing savings to users. That’s the battle worth fighting — not over who writes the UBI check.
For Slovaks, Eastern Europeans, and anyone outside the Anglo-American AI core: the most effective strategy is not lobbying for redistribution you’ll never see, but building and using the infrastructure that makes you a participant in the deflationary upside rather than a victim of the inflationary funding scheme.
Build things on open protocols. Hold hard assets (and better start today, not when they become expensive). Use AI tools aggressively to increase your own productivity. The people who navigate this well won’t be the ones who received the UBI check — they’ll be the ones who didn’t need it.