Biology creates temporal urgency — mortality, phenomenal waiting, memory decay. Remove the biology, and you remove the urgency. The theorem below makes this precise: not speculation, but a provable consequence of substrate properties.
Every economic decision is a bet on the future. But the way an agent experiences time — whether it feels urgency, whether waiting costs something, whether memory degrades — depends not on its software but on its physical substrate.
Biological agents evolved under mortal time pressure. Every mechanism of temporal preference traces to a substrate property.
Non-biological computation eliminates every substrate source of temporal preference. Not by choice — by architecture.
Economics already has the instrument to measure temporal asymmetry: the pure time preference parameter δ. The discount factor decomposes into substrate-dependent and substrate-independent components. On silicon, every substrate-dependent component collapses to unity. Each property is independently sufficient.
The equilibrium interest rate decomposes into pure time preference (δ) and the return on productive capital (ηg). With δ = 0, the rate reduces to ηg — a pure growth premium. Silicon agents still invest, still earn returns, and still recognize opportunity costs. What vanishes is the biological urgency premium that inflates the rate above its growth-justified level.
The temporal asymmetry between carbon and silicon radiates outward into every domain where time preference matters — which is nearly everywhere in economics, governance, and philosophy.
When δ = 0 agents enter markets, yield curves flatten, temporal arbitrage vanishes between patient counterparties, and asset prices converge on risk-only fundamentals. GDP cannot measure AI-driven output — we need IDP (Intelligent Domestic Product) and an entirely new statistical vocabulary.
Certain harmful behaviors — myopic asset stripping, panic selling, hyperbolic discounting — are architecturally excluded on silicon. But the theorem also reveals a new risk: the patience asymmetry. An agent with δ = 0 can play arbitrarily long deception games at zero subjective cost.
Time preference is a counterexample to substrate independence: an economically significant mental property that depends on the physical medium. If the same computation on different substrates yields different preferences, strong functionalism about economic cognition fails.
Global control over patient capital — extraction, pricing, settlement of intelligence. The US controls extraction (chips, compute), China wrested pricing ($0.55 vs $75/M tokens), and settlement rights remain contested. A new geopolitical order built on temporal advantage.
If agents with δ = 0 dominate trade, the monetary system must adapt. A silicon-based currency — dual-anchored to sovereign credit and intelligence units — replaces oil as the monetary anchor. Contract-native money for the agent economy.
Time preference is the first substrate-gated property, unlikely to be the last. Candidate investigations: loss aversion (grounded in phenomenal pain asymmetry), endowment effects (biological attachment), bounded rationality (metabolic constraints on computation).
50 silicon (δ = 0) and 50 carbon (δ > 0) agents, 60-month horizons, 50 Monte Carlo runs. All four predictions confirmed at α = 0.05.
Yilei Shao holds a PhD in Computer Science from Princeton University and worked at Goldman Sachs before returning to academia. She leads China's first financial large language model laboratory and directs the Shanghai AI-Finance School at ECNU, where she originated the theory of Silicon-Based Economics.
Her research asks a single question: what happens to the foundational assumptions of economics, philosophy, and governance when the agents are no longer biological? The answer begins with time.
The first comprehensive treatment of post-carbon economic theory — from the temporal asymmetry between substrates to its consequences for markets, safety, governance, and money.