Universal basic capital and the AI economy

If you've read about the mechanics of universal basic capital, the question that follows is whether ownership (as a concept, as a policy) is actually up to the task. Income transfers are the intuitive answer to economic displacement: people have less money, so you give them money. UBC makes a more ambitious claim: that the right response to a world where AI generates most of the value isn't supplementing diminished wages, but restructuring who participates in the economy's ownership. That's a compelling idea. It's also a harder one to make real.

The Ownership Argument, Seriously Considered

There's a distinction worth sitting with between receiving a payment and owning a stake. They can produce the same cash flow on paper, but they mean different things.

A transfer payment, whether a negative income tax, a UBI, or any welfare program, is a claim on someone else's productivity. You receive it because policy says you should. It can be increased, decreased, or eliminated by whoever controls the policy. Your relationship to it is passive: you wait for it to arrive.

An ownership stake is a different kind of thing. A shareholder has a legal claim on a portion of whatever value a company generates, independent of any particular government's generosity. They can vote on how the company is governed. In principle, their interest compounds over time as the asset grows. Their relationship to the economy isn't as a beneficiary but as a participant.

This distinction matters more as AI becomes more economically central. As scarcity economics suggests, what commands lasting value in an AI economy is ownership of productive assets, compute, data, the models built on top of them. If those assets generate most of the economy's wealth, the question of who owns them isn't just a distributional question; it's a question about who has genuine participation in the economy versus who is merely tolerated at its edges.

UBC's core promise is to extend that participation broadly. That promise is worth taking seriously even before examining whether it's achievable.

Where the Promise Gets Complicated

The first complication is access. Most of the companies most likely to generate AI-driven wealth are not publicly traded. OpenAI, Anthropic, xAI, and their peers are private companies with private shareholders. A broadly diversified sovereign wealth fund would hold publicly listed companies, capturing some AI-related growth through the large tech companies that are public, but would have limited direct exposure to the private companies where much of the concentrated return may be.

This isn't fatal to the UBC argument, but it changes what UBC can realistically deliver in the near term. A fund tracking global public equities will capture perhaps 30–40% of AI-driven wealth, not the concentrated upside of the specific companies driving the transformation. The citizens who benefit most from AI won't be the ones holding index funds; they'll be the existing shareholders of the handful of companies that win the most.

Meaningful access to private AI companies would require either forced equity distribution (companies above a certain valuation required to issue shares to a public fund) or waiting for those companies to go public. Forced equity distribution is politically explosive. Waiting means the window of maximum value creation may pass before the fund can participate.

The second complication, discussed in the canonical post, is dividends. Owning shares in companies that don't pay dividends and can't be sold produces no cash flow. Sovereign wealth fund models handle this better than direct share distribution, since the fund can actively manage its portfolio and generate cash. But this introduces a new layer of complexity: now you're depending on a government-managed fund to make good investment decisions over decades, which is a form of institutional trust that has historically been hard to maintain.

The third complication is timing. UBC is a long-horizon solution. Norway built its petroleum fund over four decades. Alaska's fund has compounded since 1977. Building a fund large enough to pay meaningful dividends to a large population takes time, time that people being displaced by AI today don't have. The structural durability of UBC, its greatest advantage over transfer programs, comes precisely from the slow accumulation of assets. In a world where AI disruption might happen faster than political systems can adapt, that's a problem.

The Governance Question

Behind every sovereign wealth fund is a governance structure, and governance structures fail. They get captured by political interests. They make bad investment decisions under pressure. They become vehicles for patronage. They get raided by governments facing fiscal crises.

Norway's Government Pension Fund is often cited as the model to emulate, and for good reason: it has maintained rigorous governance for decades, kept political interference at bay, and grown to over $1.7 trillion with disciplined management. But Norway is a small, high-trust, oil-rich country with strong institutional norms and a long tradition of consensus politics. Replicating that governance discipline in a larger, more politically divided country is not a given.

The governance question isn't an argument against UBC. It's an argument for taking the design of governance structures as seriously as the economics. A poorly governed sovereign wealth fund may be worse than no fund at all, it creates the appearance of shared ownership while delivering the reality of politically managed handouts. Getting the governance right is as important as getting the funding right.

What "Meaningful Ownership" Actually Requires

It's worth being precise about what makes ownership meaningful rather than nominal. Three things matter: liquidity (can you actually access the value?), scale (is the stake large enough to affect your life?), and representation (does your ownership give you any voice in how the asset is managed?).

Most UBC proposals score poorly on at least one of these dimensions. Direct share grants tend to be illiquid and small. Sovereign wealth fund dividends are liquid and distributed but typically too small to be economically significant, Alaska's $1,000–2,000 per year is meaningful for low-income families but not a replacement for a job. And neither model gives individual citizens meaningful voice in how AI companies are run.

This matters because ownership without voice is a passive relationship to the economy, not an active one. The deeper promise of UBC (that it turns citizens from beneficiaries into participants) requires some degree of actual influence over the assets they nominally own. Whether that's achievable at scale, or whether broad ownership necessarily becomes diffuse and voiceless, is one of the genuinely unresolved questions in the UBC literature.

The Most Realistic Path

Given all of this, what does a realistic UBC approach look like in an AI economy? A few features seem likely to be necessary rather than optional:

Fund-based rather than direct share distribution. The dividend and liquidity problems with direct shares are severe enough that a fund structure (where professional managers hold diversified assets and distribute cash returns) is almost certainly more viable in practice.

Seeded by AI-specific revenue. The funding problem is real, but an AI economy offers a natural funding source: the profits generated by AI systems. A tax on AI-generated profits, structured specifically to fund a long-horizon investment vehicle, has a cleaner political logic than general tax revenue. It connects the fund's beneficiaries directly to the activity generating the wealth.

Built now, not during the crisis. A fund started before widespread AI displacement has decades of compounding available to it. A fund started in response to a crisis starts behind and is built under pressure, which is when governance failures are most likely. The case for building UBC structures proactively (while the political environment is still relatively calm) is stronger than it might appear.

Paired with near-term income support. UBC alone cannot solve the near-term income problems that AI displacement may create. A negative income tax or similar mechanism handles the immediate floor; UBC builds the long-term foundation. Neither is sufficient without the other.

The Deeper Question

Underneath the policy mechanics is a question about what kind of society we want to build around AI. There are roughly two visions on offer.

One vision is an AI economy that generates enormous wealth, most of which accrues to a small number of asset owners, with transfer programs providing a floor for everyone else. People are taken care of (they have income) but the economy's productive core is not something most people participate in as owners. They are consumers and beneficiaries of an economy they don't shape.

The other vision is an AI economy where the gains from AI are broadly shared not just through redistribution but through ownership, where the compounding returns of AI-driven growth flow to a wide range of people who have genuine stakes in the system. This is harder to build and slower to deliver. But it creates a fundamentally different relationship between people and the economy: one of participation rather than dependency.

UBC is the mechanism most directly aimed at the second vision. Whether it can actually deliver on that vision, in the specific conditions of an AI economy with private companies, illiquid assets, and fast-moving disruption, is genuinely uncertain. What is less uncertain is that the choice between these two visions will largely be made in the next ten to fifteen years, through the ownership structures we build or fail to build now, through the tax policies we design or neglect, through the political will we muster or defer.

Being intentional about that choice (thinking carefully about it before we are forced into it by circumstances) is precisely the kind of relationship with AI that is worth cultivating. Not fear, not passive acceptance, but genuine engagement with the question of what we want the world to look like and what it will take to build it.

Frequently Asked Questions