The Numbers

The Food System Model

Computational Model — Version 2

All figures are first-order estimates derived from peer-reviewed sources and clearly identified original analysis. Every number that requires further research is flagged. Peer review is actively solicited.

Part One: Population and Caloric Demand

The Curve

The United Nations World Population Prospects 2024 revision projects global population peaking at approximately 10.3 billion in the mid-2080s, then gradually declining to around 10.2 billion by 2100. This represents a significant downward revision from projections made a decade earlier, which anticipated 10.9 billion by 2100 with no peak in sight. The earlier and lower peak is driven by faster-than-expected fertility declines across sub-Saharan Africa and sustained low fertility in East Asia, particularly China, whose population is projected to fall from 1.4 billion today to roughly 633 million by 2100.

Across the TC's first 200 years (the primary construction and transition period), the population trajectory looks roughly like this: 8.5 billion at Year 0 (2026), rising to approximately 10.3 billion at peak (roughly Year 60), then declining gradually. Some models (Randers 2012, IHME 2020) project earlier and lower peaks in the 2040s–2060s at 8.8–9.7 billion if education and family planning targets are met. The TC itself would accelerate the decline through universal education, universal provision, and the well-documented inverse correlation between security and fertility.

Design Population

Design target: capacity for 12 billion people. This is not a population projection. It is a deliberate engineering margin. The food system should be built to feed 12 billion people at 2,500 kcal per person per day with full nutritional diversity. At peak population of 10.3 billion, this provides a 16% surplus. As population declines, the surplus grows. The surplus is not waste. Surplus plant biomass continues to sequester carbon. Surplus food supports genetic conservation, seed banking, and ecological resilience. Every calorie of surplus is a calorie of insurance and a gram of carbon removed from the atmosphere.

Total Caloric Demand at Design Capacity

12 billion people × 2,500 kcal/day × 365 days = 10.95 × 10¹⁵ kcal/year. This is the number the food system must deliver. All zone calculations below sum to this total or exceed it.

Nutritional requirements beyond raw calories: approximately 60–70 grams of protein per person per day (FAO/WHO), adequate fat intake of 20–35% of calories (44–97 grams/day), plus essential vitamins, minerals, and fiber. The four-zone system must deliver a complete diet, not merely energy.

Part Two: The Four-Zone Food Architecture

The food system operates across four integrated zones. Each zone serves a distinct function. Together they provide redundancy: if any single zone underperforms, the others compensate. This is not optimization for minimum sufficiency. It is deliberate over-engineering for civilizational resilience.

Zone 1: Underground Controlled Environment Agriculture (CEA)

Function: Primary calorie engine. Climate-independent. Latitude-independent. Year-round production. The backbone of the food system.

Why underground. Three reasons. First, every square meter of underground growing space is a square meter of surface land returned to living ecosystem. The food system must not compete with the minimum nature requirement (80–90% of habitable land returned to a natural state). Second, underground facilities are thermally buffered by surrounding rock mass, dramatically reducing heating and cooling energy. Third, at depth, geothermal energy is closer — a facility one kilometer underground needs one kilometer less drilling to reach superhot rock.

Crop Matrix and Yield Derivation

The following yields are derived from peer-reviewed data, scaled to annual production in a continuous-harvest controlled environment with optimized lighting, nutrients, and atmospheric CO₂. All yields are per square meter of floor space per growing tier per year.

Blended Yield Derivation

An 8-tier underground facility with a diversified crop matrix produces the following blended yield per square meter of floor space per year.

The derived blended yield range is 105,000–208,000 kcal/m²/yr. The moderate estimate of approximately 147,000 kcal/m²/yr is the working figure.

Zone 1 Floor Area Derivation

If Zone 1 provides 60% of total caloric demand at design capacity: 0.60 × 10.95 × 10¹⁵ = 6.57 × 10¹⁵ kcal/yr.

Working estimate: 32,000–63,000 km² of underground floor space. For reference, 45,000 km² is roughly the area of Denmark.

Zone 2: Surface Agriculture Under Agrivoltaics

Function: Grain production, orchards, perennial crops. Dual-use land generating both food and energy. Distributed near settlements.

Surface agriculture handles the crops that thrive in sunlight and are less efficient underground: field grains (wheat, rice, oats, barley, corn), full-sized orchards, and perennial polyculture crops. These are grown under agrivoltaic arrays that generate electricity while providing partial shade that can actually improve yields for certain crops.

Agrivoltaic energy quantification. At an average power density of 0.6 MWp/ha (60 MWp/km²), agrivoltaic panels covering 30–40% of Zone 2 surface area generate substantial electricity. With Zone 2 at 2–4 million km² and 30% agrivoltaic coverage, installed capacity is 36,000–72,000 GWp. At a global average capacity factor of 17%, this yields 6.1–12.2 TW of continuous power. This alone could power the entire underground food system's LED lighting load.

Caloric contribution: average global grain yield is approximately 4 tonnes/hectare. At 3,200–3,400 kcal/kg for cereals, this is approximately 1,300 kcal/m²/yr of surface area. Across 2–4 million km², Zone 2 contributes 2.6–5.2 × 10¹⁵ kcal/yr — roughly 24–48% of design capacity.

Zone 2 area: approximately 2–4 million km² (approximately 1.5–3% of habitable land).

Zone 3: Food Forests

Function: Nutritional diversity, community connection, genetic conservation. The transition gradient between settlement and restoration. Perennial polyculture requiring minimal maintenance once established.

Food forests are multi-story perennial polycultures: canopy trees (nut and fruit), understory trees (dwarf fruit, nitrogen-fixers), shrubs (berries), herbaceous layer (perennial vegetables, herbs), ground cover, vine layer, and root layer. Research from Scotland (Crawford 2017) found mature food forests can supply caloric needs for 6–10 people per hectare. Danish and Norwegian studies found similar ranges. Tropical food forests significantly exceed these figures.

Caloric contribution: roughly 1–3% of total demand. The value of Zone 3 is not caloric volume. It is nutritional diversity, community-scale food sovereignty, genetic conservation of thousands of cultivars, and the ecological transition gradient from human settlement to restoration ecosystem. Food forests take 8–15 years to reach full productivity, aligning with the TC transition timeline. Area: 200,000–500,000 km².

Zone 4: Wild Harvest

Function: Protein diversification, ecosystem management, cultural practice, invasive species control. Integrated with restored ecosystems (80–90% of habitable land).

In a world with 80–90% restored ecosystem, wild harvest is not marginal. It is a substantial food source. This includes large and small game, managed marine harvest, freshwater fish from restored rivers and lakes, wild-harvested plants, and the systematic culling of invasive species such as feral pigs.

The transition from domestic cattle to wild ungulate herds occurs over decades. AI-assisted drone herding can manage wild populations humanely, using behavioral algorithms to direct animals through sorting infrastructure for selective harvest. Females, juveniles, and breeding stock are released; animals selected for harvest are directed to processing. This replaces industrial feedlots with ecosystem-integrated protein production that simultaneously maintains grassland health through grazing pressure.

Wild harvest grows over centuries as ecosystems mature. The food system does not degrade over time. It improves.

Part Three: Total System Balance

At moderate estimates, the four-zone system meets 100% of the 12-billion design capacity. At conservative estimates, it meets 86% — requiring either additional Zone 2 surface area or improved Zone 1 yields to close the gap. At optimistic estimates, the system produces a 13% surplus above design capacity.

Note on surplus as carbon sequestration: Every kilogram of surplus plant biomass represents approximately 0.45 kg of carbon removed from atmospheric CO₂. A 10% food surplus at full system capacity represents roughly 1–2 billion tonnes of additional biomass per year — a meaningful contribution to the sequestration stack. Surplus is a feature, not a flaw.

Part Four: What-If Scenarios

Northern Zone Design: Dwarf Fruit Trees and Grain Underground

Settlements above approximately 55°N latitude face severely reduced growing seasons for surface agriculture. For these populations, Zone 1 underground capacity must compensate for limited Zone 2 and Zone 3 output.

Indoor wheat: The Asseng et al. (2020) PNAS study demonstrated wheat yields of 700 t/ha/yr (measured) to 1,940 t/ha/yr (modeled optimum) in a 10-layer facility with optimized lighting at 2,000 μmol/m²/s, 24-hour photoperiod, elevated CO₂ at 1,200 ppm, and 70-day growth cycles. This is 220–600 times conventional outdoor yields. The energy cost is substantial, but with geothermal-powered electricity at effectively zero fuel cost, the economic barrier that currently makes indoor grain impractical disappears.

Dwarf fruit trees underground: Dwarf apple, pear, cherry, and citrus varieties typically reach 2–3 meters in height and can be grown under high-ceiling (4–5m) indoor tiers with specialized LED spectrum including far-red wavelengths for flowering induction. Commercial trials have demonstrated economically viable indoor strawberry production. For northern zones where surface fruit production is limited, the energy trade-off is justified.

All-Underground Scenario: Surface Agriculture Lost

Scenario: 50 years from now, cascading climate disruption renders large-scale surface agriculture unviable. All food production must move underground. This scenario is speculative and explicitly less rigorous than the primary four-zone model. It is included because responsible engineering requires knowing what the fallback position looks like.

Assessment: An all-underground food system is physically and energetically feasible within the TC framework. It is not the primary plan. It is the insurance policy. Nothing in physics, biology, or engineering prevents it. The barrier is scale of construction, which the TC's robotic construction capacity over 50–100 years can address.

Part Five: Infrastructure Requirements

Concrete Volume and Carbon Mineralization

Underground growing facilities require structural concrete for chambers, tier supports, corridors, and utility infrastructure. The TC uses carbon-cured concrete incorporating CO₂ mineralization and geopolymer binders that reduce cement-related emissions by up to 80%.

Volume derivation: A single underground growing tier requires approximately 0.15–0.25 m³ of concrete per m² of floor space. An 8-tier facility requires approximately 1.5–2.5 m³/m² of ground footprint including corridors and utilities. At the moderate floor area estimate of 45,000 km²: approximately 90 billion m³ of concrete, or approximately 216 billion tonnes. This is constructed over 100–200 years using automated robotic manufacturing.

Carbon mineralization potential: Advanced carbonation-cured concrete can achieve 100–200 kg CO₂/m³ through accelerated carbonation. At 90 billion m³ with advanced carbonation at 150 kg CO₂/m³: approximately 13.5 GtCO₂ permanently mineralized in food infrastructure alone. Distributed across 150 years of construction: approximately 90 MtCO₂/yr.

1,000-year infrastructure philosophy: The TC builds infrastructure to last a millennium. Underground growing bins are poured concrete, not plastic trays. Structural elements are designed for geological timescales. A concrete growing tier installed in Year 10 is still operating in Year 1,010.

LED Lighting System

Current commercial LED horticultural fixtures achieve 2.7–3.2 μmol/J photosynthetic photon efficacy. The theoretical maximum for LEDs is 4.6–5.1 μmol/J. LED lifespan: current high-quality horticultural LEDs maintain greater than 90% output for 36,000–50,000 hours (approximately 4–6 years of continuous operation). Under the 1,000-year infrastructure philosophy, LED modules are designed as replaceable cartridges in permanent concrete fixtures. Annual LED replacement covers approximately 9,000 km² — a large but manageable robotic manufacturing and installation task.

Energy Demand Derivation

Revised estimate: Underground food energy demand of 4–8 TW sustained is defensible when accounting for variable lighting intensity by tier (tubers at 100–150 W/m², greens at 150–200, grain at 300–400), duty cycles, and the smaller total floor area from higher yields. Potentially offset by the 6–12 TW generated by Zone 2 agrivoltaics.

Part Six: Geographic Distribution

The food system does not concentrate in a single region. It distributes roughly in proportion to where people live, with latitude-dependent adjustments to the zone mix.

Principle: People move short distances, not long ones. The TC does not relocate populations from coasts to interior or from north to south. Settlements shift modestly — perhaps 50–100 miles inland from flooding coastlines. The food system follows the population, not the reverse.

Part Seven: Honest Gaps and Research Needs

This model is a first-order derivation. The following gaps are named explicitly.

Gap 1: Staple grain yields underground are modeled, not commercially demonstrated at scale. The Asseng et al. PNAS study used crop simulation models validated against small-scale experimental data. Commercial-scale indoor grain production does not yet exist. Confidence: Reasonable Inference for moderate estimates, Informed Speculation for optimistic.

Gap 2: Underground fruit tree cultivation at scale is untested. Dwarf fruit trees under LED are grown in research settings and high-end commercial operations. No one has operated a multi-hectare underground orchard. Confidence: Informed Speculation.

Gap 3: Energy budget requires formal derivation. The 4–8 TW estimate is derived from average lighting loads and total area. A crop-by-crop, tier-by-tier energy model would produce a tighter range.

Gap 4: Concrete volume estimate is order-of-magnitude. Actual structural requirements depend on depth, geology, seismic zone, and construction method.

Gap 5: Wild harvest caloric contribution is highly uncertain. Pre-colonial carrying capacity estimates are contested, and restored ecosystems will not replicate pre-colonial conditions exactly.

Gap 6: Nutritional completeness has not been formally verified. A formal nutritional audit against WHO/FAO dietary reference intakes for all essential vitamins, minerals, and amino acids has not been performed.

Gap 7: Food forest yield data is limited. Most food forest research comes from small-scale temperate projects with 3–7 years of yield data.

Gap 8: Agrivoltaic crop interaction data is early-stage. Crop-specific yield responses to partial shading vary significantly by species, climate, and panel configuration.

What this model establishes: the food system proposed by the Trust Collective is physically, biologically, and energetically feasible. Every component rests on demonstrated technology or published research. No step requires a breakthrough. What it requires is scale — and the TC's robotic construction capacity across a 100–200 year timeline is designed to deliver exactly that scale.

What this model does not establish: optimized engineering specifications for underground growing facilities. That is the work of the institution's research program, not a framework document. This model shows that the math closes. The engineering is solvable. The specific solutions are proposals, open to revision.

The thread grows brighter with every person who chooses it.

From the Trust Collective Project  |  April 2026