# Assets and Distribution - who owns the inelastic complements ← [Part IX](README.md) · [Index](../README.md) --- [Game 3](../02-games/3-firms.md) concludes that value accrues to whatever intelligence cannot manufacture. That is usually read as a guide to business strategy. It is also, and more importantly, **a prediction about wealth distribution** - because the inelastic complements are assets, and assets have owners. ## The list, reread as an ownership question | Complement | Who holds it | Ownership concentration | |---|---|---| | Energy and the right to build it | Utilities, IPPs, increasingly hyperscalers | High, and rising | | Land with power, water, zoning | Landowners, REITs, sovereigns | Very high | | Regulatory licenses and liability capacity | Incumbent professionals and firms | High, and legally protected | | Proprietary real-time flow data | Platforms, equipment makers, exchanges | Extremely high | | Distribution and default placement | A handful of platforms | Extremely high | | Trust, brand, balance sheet | Large incumbents | High | | Physical presence and skilled trades | Individual workers | **Low - the exception** | **Six of the seven are held by capital. One is held by labor**, and it is the only one whose returns flow to people rather than to portfolios. That asymmetry is the distributional story, and it is not softened by the consumer-surplus result - it is *created* by it. Consumer surplus is real and widely shared, but it is a **flow that shows up as cheaper goods, not a stock that can be borrowed against, passed on, or lived off.** Asset appreciation is a stock. The gap between those two compounds across generations in a way the flow does not. ## The uncomfortable synthesis Put [prices](prices.md) and this page together and the 2030s look like: - **Cheaper goods and services**, especially cognitive ones - genuine, large, widely distributed - **More expensive housing, energy, care, and anything physical** - regressive, salient, monthly - **Rising returns to asset ownership**, concentrated among existing owners - **Falling returns to credentialed cognitive labor**, which is what the professional middle class sells - **Rising returns to skilled physical labor**, which is the one complement individuals actually own The class that loses most is not the poorest. **It is the credentialed professional middle** - the group whose entire economic position rests on scarce cognitive skill certified by an institution, holding an asset that [depreciates on a two-year schedule](../01-substrate/inference-economics.md). That group is small in number and enormous in political influence - it staffs the media, the professions, the civil service, and the political class itself. **Predictions about the politics of this transition that ignore who is writing the coverage will be systematically wrong.** ## The skilled-trades exception is worth dwelling on Electricians, plumbers, HVAC technicians, welders, and linemen sit at an unusual intersection: their work is **physical, unstructured, licensed, liability-bearing, and in direct demand from the AI build-out itself.** Every constraint in this document protects them at once. [Robotics](../03-domains/physical/robotics/) does not reach unstructured physical work in this window. Licensure limits supply. [The energy build](../01-substrate/energy.md) raises demand directly. And the [apprenticeship gap](../06-uncertainties/apprenticeship-gap.md) that damages professional formation runs *backwards* here, because trades kept their apprenticeship systems intact while the professions let theirs erode into "junior does the routine work." > **Prediction:** real wage growth in licensed skilled trades outpaces real wage growth in credentialed cognitive professions over 2026–2035. **~70%** confidence. This is one of the most checkable claims in the document and one of the least intuitive. The bottleneck on it is training capacity, which is institutional and slow - the same shape as every other constraint here. ## The failure mode: the list has a half-life Every row in the ownership table is a claim that intelligence cannot manufacture that complement *within the window*, and [Uncertainty 7](../06-uncertainties/complement-erosion.md) red-teams exactly this. If robotics arrives early, the physical-presence row - the one row labor holds - erodes first, which would convert this page's least intuitive prediction into its worst miss. If licensure reprices under fiscal pressure, the regulatory row follows, and the professional moat goes with it. The distributional claims here inherit the half-life of the scarcities they rest on and should be read as a dated position, not a static endowment; the trades-wage prediction above is explicitly conditioned on [B12](../07-indicators/diffusion/capability.md) staying quiet. ## The sharpest line is owner against renter Within households, the divide this page predicts is not primarily labor against capital - most households hold some of both - but **housed against unhoused.** Housing is the one appreciating complement ordinary households own at scale: roughly two-thirds of households in most advanced economies own their home (OECD housing statistics, stable for decades), but ownership is concentrated sharply by age and entry keeps getting more expensive. The renter cohort experiences the inflating basket in [Prices](prices.md) with no offsetting asset gain; the owner cohort collects the gain without selling. Because ownership tracks age, **the price split lands substantially as an intergenerational transfer**, and the politics of the 2030s will run along that line at least as hard as along the employment one - housing politics is AI distribution politics whether or not anyone frames it that way. ## What could break the concentration Three mechanisms, in descending order of plausibility: 1. **Supply response in the inelastic complements.** Land and energy are inelastic *given current regulation*. Permitting reform, transmission build-out, and housing liberalization would each convert a rent into a normal return. **Nothing else on the table moves the distribution this much, and it is almost never discussed as distributional policy** - housing and permitting policy are the AI distribution policy. → [Uncertainty 2](../06-uncertainties/power-permitting.md) 2. **Broad ownership of the complements.** Sovereign wealth funds, universal capital accounts, and pension exposure to infrastructure all convert asset appreciation from a concentrating force into a distributed one. Mechanically straightforward, politically difficult, and already implemented in several countries. 3. **Taxing the rents rather than the labor.** Land value taxation is the textbook answer to exactly this configuration and has been the textbook answer for 150 years without being adopted. → [Fiscal](fiscal.md) Note that **none of these are AI policy.** They are housing, energy, tax, and pension policy. That is the recurring finding of this part: **the distributional consequences of AI are mostly determined by institutions that have nothing to do with AI**, and are being set right now by people not thinking about it at all. ### Interaction with complement erosion [Uncertainty 7](../06-uncertainties/complement-erosion.md) is the adversarial page for this one. If the scarce list has a short half-life, concentration claims here overstate late-2030s wealth inequality even while early-2030s concentration is real. The operational discipline: date-stamp which row of the complement list is doing the work in any distributional claim, and retire the claim when that row's indicator fires. Land+zoning is the residual after other rows erode; a page that needs six durable rows is more fragile than a page that needs one. **Housed vs unhoused is the household cut.** Owner-occupiers collect the land/power appreciation; renters eat the inflating basket with no offsetting asset - an intergenerational and tenure split that will brand as housing politics, not AI politics. --- **Observable:** [B7](../07-indicators/diffusion/economy.md) price split and [B12](../07-indicators/diffusion/capability.md) physical diffusion - the trades-wage prediction above holds only while unstructured physical work stays out of robotics' reach; it is tracked in the [register](../05-probabilities/register.md). **Related:** [Game 3 - Firms](../02-games/3-firms.md) · [Prices](prices.md) · [Fiscal](fiscal.md) · [Game 4 - Labor](../02-games/4-labor.md) · [Uncertainty 2](../06-uncertainties/power-permitting.md) **Next:** [Fiscal](fiscal.md)