In one sentence
AI could create unprecedented abundance, but without a new distribution system it will also intensify inequality. Altman’s proposed answer is to tax the assets that capture AI-driven gains—companies and land—and distribute the proceeds as universal citizen ownership and income.
Overview
Written in March 2021, the essay presents a forecast and a policy proposal. Altman expects AI systems to perform more cognitive and physical work, producing a recursive acceleration of technological progress. As labor becomes cheaper, the prices of many goods and services could fall dramatically—his expanded version of Moore’s Law. However, ownership of productive companies and scarce land would capture most of the resulting wealth. Progressive income taxation, he argues, will become less effective as labor contributes less economic value. The proposed replacement is an American Equity Fund: companies above a valuation threshold would transfer 2.5% of their market value annually in shares, while private land would incur a 2.5% annual tax in dollars. Adults would receive recurring distributions in cash and company shares.
Core ideas
Abundance is not the same as equality
AI may make goods and services much cheaper, increasing everyone’s purchasing power. But cheaper production does not automatically give people ownership of the systems producing that abundance. Without redistribution, capital owners could become vastly richer while workers lose bargaining power.
“Moore’s Law for everything”
The central economic intuition is that technological progress can improve living standards by lowering prices, not only by raising incomes. Altman imagines AI and robotics reducing the cost of housing, education, healthcare, food, and other essentials—though this depends on bottlenecks such as land, regulation, physical materials, and infrastructure also becoming manageable.
Tax capital rather than labor
If AI performs much of the economically valuable work, taxing wages becomes a weaker way to fund social welfare. Altman therefore focuses on the two assets he expects to capture most value: AI-enabled companies and land, whose supply is fixed and whose value partly reflects surrounding public investment.
The American Equity Fund
The fund would receive annual transfers equal to 2.5% of qualifying companies’ market value, paid in shares, plus 2.5% of privately held land value, paid in cash. Adults would receive annual distributions, gaining both purchasing power and a direct stake in national economic growth.
Ownership aligns incentives
Altman prefers equity transfers to profit taxes because share ownership connects citizens’ fortunes to company and national performance. His broader political claim is that shared ownership could make people more supportive of growth while reducing conflict over redistribution.
Implementation is politically and administratively difficult
The essay acknowledges avoidance and valuation problems: firms could move offshore or remain private; land assessments could be disputed; excessive taxation could reduce investment; and recipients might borrow against future payments. It suggests revenue tests, delayed taxation for private firms, constitutional limits, and restrictions on pledging future distributions.
Practical takeaways
- When evaluating AI policy, separate two questions: how much abundance technology creates and who owns the productive assets that create it.
- Price reductions can function like income increases, but only for goods and services whose supply technology can actually expand. Scarce land and regulated services may not follow the same path.
- Automation strengthens the case for broad asset ownership, but the design of taxes matters: valuation, avoidance, investment incentives, and political durability are central issues.
- A universal dividend is not the same as universal prosperity. Its real effect depends on whether distributions outpace inflation and whether essentials—especially housing, healthcare, and education—become more affordable.
- The proposal is best read as a conversation starter and strategic framing device, not as a fully specified fiscal plan.
Caveats and counterpoints
- The essay’s technological timeline is highly speculative. Its predictions about AI capabilities, recursive improvement, and near-zero labor costs are asserted more confidently than demonstrated.
- The claim that housing, healthcare, and education will become dramatically cheaper may fail where the binding constraint is scarce land, professional licensing, institutional capacity, or political allocation rather than labor.
- A 2.5% annual tax on company value could create substantial dilution, valuation volatility, avoidance incentives, and pressure to prioritize distributions over reinvestment.
- Land-value taxation is conceptually distinct from taxing the full value of buildings and improvements; the essay admits that accurately separating land from improvements may be difficult.
- The proposal assumes that growth will remain strong enough to offset transitional disruption and that citizens will accept significant changes to taxation and corporate ownership.
- The essay treats growth and shared ownership as broadly mutually reinforcing, but it gives less attention to power concentration, market competition, environmental limits, and governance of highly capable AI.
Questions worth revisiting
- If AI makes labor cheap but land and infrastructure remain scarce, which costs actually fall—and which become more expensive?
- How would the fund value private companies, complex financial assets, and land in rapidly changing markets?
- Would distributing company shares create meaningful citizen influence, or mainly passive exposure to corporate performance?
- What prevents AI-generated wealth from accruing to a small number of firms with monopoly power?
- How should the system handle people unable to wait for long-term growth or who live in regions with sharply different land values and costs of living?
Return to this when…
Return to this essay when thinking about AI-driven automation, universal basic income versus universal basic ownership, land-value taxation, or the difference between technological abundance and equitable access. Its most useful enduring question is: if machines produce most value, what institutional mechanism ensures that people own part of the machine economy?