The return to abundance: how AI ends the age of scarcity.
Humanity began in abundance — an ecosystem that grew our food, filled our rivers, and cleaned our air, asking only that we live simply. Scarcity is what we built on top of it. This is the case that AI and robotics drive the cost of producing our needs toward zero — and return us, at last, to where we started. Here are the models, and a dated estimate of when we arrive.
Strip away everything, and a human being needs remarkably little to live and be content: clean air, fresh water, food, warmth, shelter, and the company of others. For most of our species' history, an ecosystem supplied the first four for free. Foragers, anthropologists find, worked only 15–20 hours a week to meet their needs — the original "affluent society." We did not begin in scarcity. We invented it.
Scarcity arrived with two forces: reproduction (more mouths than a given patch of land could feed) and settlement (agriculture, which fed more people but demanded far more labor per person, and created stores of wealth worth fighting over). From that moment, the human story became a story of scarcity — innovate to produce more, and fight over what already exists. Every economy since has been a machine for managing not having enough.
The thesis of this piece is simple, and it reverses the usual doom framing. AI and robotics are the first technology capable of breaking that machine — not by rationing scarcity better, but by ending it. If machines can grow the food, mine the materials, design the products, and build them with steadily less human input, the cost of meeting a human's basic needs falls toward zero. And when needs are abundant and nearly free, humanity is released to do what it did at the start: live simply, and focus on family, community, and meaning. The "end" AI brings is not the end of humanity. It is the end of the age of scarcity.
AI & robotics push the marginal cost of producing our basic needs toward zero. As that happens, the labor a human must trade for survival collapses — and we return to the low-effort, high-abundance life we began with. The hard question is not whether we can produce abundance, but whether we choose to share it.
Everything is becoming cheaper — faster than we notice
The engine of the thesis is a law most people have never heard of but live inside: Wright's Law. For a huge range of technologies, every time cumulative production doubles, unit cost falls by a roughly constant percentage. Solar panels fall ~20% per doubling. The result, compounded over decades, is staggering — and it is accelerating as AI optimizes the design and manufacture of everything.
Here is why AI changes the trajectory rather than merely continuing it. Historically, cost curves were gated by human ingenuity and human labor — how fast we could design, build, and improve. AI & robotics attack both limits at once: AI compresses the design-and-discovery cycle (new materials, cheaper processes, better logistics), and robotics compresses the labor cost of making the physical thing. When the two inputs that never fell — human design time and human labor — begin to fall too, the marginal cost of a physical good starts to behave like the marginal cost of a digital one: it heads toward the cost of energy and raw material alone. And energy, as the chart shows, is already collapsing.
A digital good already costs almost nothing to reproduce. The endgame is when a physical good — a meal, a tool, a house — does too.
The U-curve: we are returning to where we began
Plot the number of hours a person must work to secure their basic needs across human history, and you do not get a straight line down. You get a U. Foragers needed few hours because they consumed little and nature provided. Agriculture and then industry drove that number up — more security and more stuff, but at the cost of long, hard labor. Now the curve is bending back down. AI & robotics are the far arm of the U — carrying us back toward the low-labor abundance we started in, but this time with modern comfort layered on top.
This is not a new hope — it is a very old prediction catching up to reality. In 1930, John Maynard Keynes wrote that within a century, productivity would be so high that his grandchildren would work perhaps 15 hours a week and struggle mainly with how to fill their leisure. He was early, not wrong: he simply underestimated how much of the dividend would be spent on more consumption rather than more freedom. AI & robotics are the force finally large enough to make the choice unavoidable.
Four needs, four production stacks going autonomous
Abundance is not one breakthrough; it is four production systems each crossing the same threshold — where machines do the work and the marginal cost falls to the price of energy. Every one of them is already in motion.
🌱 Food
Precision agriculture, autonomous farm robotics, vertical farms and cultivated protein. AI already lifts yield up to 40% while cutting water and inputs — the beginning of food produced with minimal human labor.
💧 Water
AI-optimized desalination and distribution, powered by collapsing solar cost. Fresh water becomes an energy problem — and energy is the input falling fastest.
⚡ Energy
Solar + storage down ~90% a decade, following Wright's Law toward near-zero marginal cost; fusion and advanced geothermal as the long tail. Energy is the master input abundance runs on.
🏭 Goods
AI design + robotic manufacturing + autonomous mining and logistics. When machines both design and build, physical goods start to price like digital ones — toward the cost of materials and energy alone.
Notice the shared dependency: every stack bottoms out at energy and raw materials. That is the whole game. Abundance is not gated by intelligence — AI is arriving fast. It is gated by the physical inputs: cheap clean energy, and the metals, water, and land the machines consume. Which is precisely why the destination is a green-economy question, not just a software one.
The road to abundance, in four phases
Here is a dated map of the destination. Treat the years as scenario markers, not predictions — the sequence is more robust than the dates. Each phase is defined by which production stack crosses the "machines do the work, cost approaches zero" line.
The zero-cost mind
Marginal cost of digital and knowledge work → ~0. Software, media, design, analysis, and tutoring become nearly free and infinitely abundant. Realized job impact is still small; the productivity payoff climbs the flat part of the J-curve.
The zero-cost hand
Robotics scales and general-purpose machines get cheap. The marginal cost of physical goods, food, and logistics falls sharply. Solar + storage dominate new energy; the first commercial fusion appears. Basic goods begin to feel abundant in adopting economies.
The post-labor threshold
Autonomous production can meet most basic needs with minimal human labor. Work becomes optional for survival — and distribution (UBI, public capital, "who owns the robots") becomes the defining political question of the era. The transition's hardest years.
The return to abundance
Food, water, energy, and everyday goods approach near-zero cost and universal availability. The cost of a simple, comfortable life collapses. Humanity is free to refocus on family, community, creativity, and meaning — the life we began with, restored on the far arm of the U.
When, exactly? Three scenarios.
The base case — practical post-scarcity for basic needs, ~2060 — is deliberately about necessities, not luxuries. Positional goods (prime land, status, originals, human attention) stay scarce forever. Abundance ends the struggle to survive, not the desire to distinguish.
Producing abundance is not the same as sharing it
Intellectual honesty requires stating clearly what could break this thesis — because the failure modes are not technical. They are human.
Why we invest at the bottom of the abundance stack
If this thesis is right, the constraint on abundance is not intelligence — it is the physical layer: clean energy, water, food, and materials. That is exactly where a green-economy firm operates. Every stack in this piece bottoms out at energy and resources, which is why we back companies at that base — renewable energy and water, precision agriculture, and the AI that makes each acre and each watt go further.
The optimistic future in this thesis is not something to wait for. It is something to build the inputs for — and to build so that abundance, when it comes, is shared. That is the whole point of growing with nature, and delivering with care.
See the portfolio → Read: The Great Reallocation →How to read this piece
This is a thesis, not a forecast. The cost-decline figures are real and sourced; the timeline and arrival dates are scenario markers derived from extrapolating current learning curves, and should be treated as illustrative. The evolutionary "U-curve" is a conceptual model. Reasonable people place the destination decades earlier or later — or argue distribution never allows it. That debate is the point.
- IRENA — Renewable Power Generation Costs: solar PV LCOE −90% and battery storage −93% (to ~$197/kWh) over 2010–2024. irena.org
- Our World in Data — Wright's Law learning curves; solar prices fall ~20% per capacity doubling. ourworldindata.org
- NHGRI — cost per human genome: ~$100M (2001) → <$1,000 today (~99.99% decline). genome.gov
- J.M. Keynes — "Economic Possibilities for our Grandchildren" (1930): the 15-hour work week prediction.
- M. Sahlins — "The Original Affluent Society" (1972): forager labor of ~15–20 hrs/week.
- Related framing — J. Rifkin, The Zero Marginal Cost Society (2014); post-scarcity economics. Counter-view: D. Acemoglu on modest near-term TFP gains & distribution.
A GreenLeafSource Research thesis · Compiled July 2026 · Speculative — a model of the destination, offered to start the debate.