The meta-technology

Artificial intelligence is not simply another technology on the list. It is the technology that accelerates every other technology on the list. It is the meta-technology: the one that makes all other exponentials steeper.

The mechanism is recursive self-improvement. Better AI designs better chips. Better chips provide more compute. More compute trains better AI. This loop now generates visible generational leaps within months rather than years.

What AI accelerates

Energy: AI optimises solar cell design, battery chemistry, grid management, and fusion plasma control. Every percentage point of efficiency gained cascades through the entire energy cost curve.

Materials: AI models predict the properties of novel materials before they are synthesised. GNoME identified over two million stable crystal structures; the entire history of human materials science had found fewer than fifty thousand.

Biotech: AlphaFold solved protein folding. AI-driven drug discovery, synthetic biology, and genomic medicine are advancing at extraordinary pace.

Robotics: Foundation models give robots the ability to generalise, moving them from rigid automation toward flexible, adaptive physical labour.

The risk

The same recursive capability that makes AI transformative also makes it dangerous. The alignment problem is not theoretical; it is a practical engineering challenge that must be solved in parallel with capability development. FOME takes AI safety seriously precisely because the abundance thesis depends on AI going well.

The ur-input

Almost every form of scarcity, traced back far enough, is an energy problem in disguise. Water? Desalination is energy-intensive. Food? Vertical farming needs lighting. Materials? Extraction is energy-expensive. Compute? Fundamentally an energy cost. Energy is the ur-input: make it cheap enough and you dissolve a dozen other problems that were secretly energy problems all along.

The solar cost curve

Solar PV costs have fallen roughly 99% since the 1970s and continue dropping 10–15% per year. Every doubling of installed capacity drives a predictable cost reduction. Solar is now the cheapest electricity source in history, and we are nowhere near the floor. The curve does not flatten because each generation of panels is manufactured using energy from the previous generation, and AI is accelerating materials discovery for next-generation cells.

Batteries and fusion

Lithium-ion costs have fallen 97%. Sodium-ion, iron-air, and solid-state chemistries are entering production. Fusion is the backstop, not the prerequisite. If it arrives in the 2030s, extraordinary. If not, solar plus storage already gets us there.

What cheap energy unlocks

Desalination: unlimited fresh water for every coastal nation. Carbon capture: planetary-scale emissions reversal. Vertical farming: food production decoupled from land. Manufacturing: costs collapse when energy, robots, and AI converge. Transport: EVs, autonomous freight, eventually electric aviation; all energy-cost stories.

The tightest loop

AI designs better chips. Better chips provide more compute. More compute trains more capable AI. This loop now turns in months. When DeepMind used AI to design chip layouts that outperformed human engineers, the loop closed: the intelligence running on the chips was improving the chips it runs on.

Why compute matters

Every domain on the map is a compute problem at some level. Drug discovery, materials prediction, climate modelling, robotic perception, grid optimisation: all require more compute. Cheaper, more powerful compute accelerates everything AI touches, which is increasingly everything.

The energy dependency

Compute is ultimately an energy story. Without cheap energy, compute hits a cost ceiling. With cheap energy, the ceiling lifts. This is why energy and compute sit side by side at the foundation.

Physical intelligence

Foundation models are giving robots the ability to generalise across tasks. Robots trained on vast datasets of physical interactions are learning to manipulate objects they have never seen, navigate spaces they have never mapped, and recover from failures they have never encountered.

Labour cost approaches zero

When a robot can do physical work, the cost becomes energy, maintenance, and amortised hardware. All three are falling exponentially. Every form of repetitive or dangerous physical labour will be automated within a generation.

Robots powered by AI, consuming cheap energy, build the solar panels, assemble the chips, and manufacture the next generation of robots. The loop closes.

Genome sequencing fell from $3 billion to under $200 in two decades. CRISPR, synthetic biology, and AI-driven protein design have moved biology from observation to engineering.

Food: Precision fermentation produces dairy proteins without animals. Cellular agriculture grows meat from cells. Cost curves are falling steeply.

Medicine: AI-driven drug discovery compresses pharmaceutical timelines from decades to years. Personalised medicine, calibrated to individual genomic profiles, becomes standard practice.

Materials: Engineered organisms produce biofuels, biodegradable plastics, industrial enzymes, and novel materials. Biology becomes a manufacturing platform.

The hidden accelerator. Better photovoltaic materials make solar cheaper. Better semiconductors make chips faster. Better battery materials make storage denser. Nearly every cost curve has a materials bottleneck, and AI is dissolving them.

GNoME predicted over 2.2 million stable crystal structures. The entire previous history of materials science had found fewer than fifty thousand.

Energy: perovskites, solid-state batteries, superconductors. Compute: new semiconductors, photonic substrates. Construction: self-healing concrete, carbon-negative building materials. Recycling: better catalysts make it profitable to recover almost anything from waste streams.

The least mature acceleration technology but potentially the most transformative. Classical computers face fundamental limits simulating quantum systems. Quantum computers operate natively in that domain.

Drug discovery: quantum-accurate molecular simulation. Materials: first-principles property prediction. Optimisation: exponential speedups for combinatorial problems. Cryptography: breaks current encryption, drives post-quantum security.

When AI, robotics, and cheap energy converge, the marginal cost of one additional unit becomes negligible. This is already visible in digital goods. It is now extending to physical goods.

The Migrating Boundary: the line between "scarce" and "abundant" moves through categories in sequence. Digital goods went first. Now the frontier advances into manufacturing, food, construction, and services.

An economy built on scarcity does not transition smoothly to abundance. Price signals, profit motives, and labour markets all assume scarcity. As costs collapse, these assumptions break. This is not a bug; it is the signal that the economic operating system needs upgrading.

Healthcare: As a GP, I watch this with particular intensity. AI diagnostics match specialists in radiology, dermatology, pathology, ophthalmology. A patient in rural sub-Saharan Africa could access the same diagnostic intelligence as one in central London.

Education: Personalised AI tutoring for every child on Earth. Early evidence shows outcomes comparable to one-on-one human tutoring, which was never scalable before.

Legal and admin: Contract drafting, compliance, tax, benefits. Expert-level service at near-zero cost, accessible to everyone.

Conventional agriculture: 50% of habitable land, 70% of water, 25% of emissions. Three technologies decouple food from all three constraints.

Vertical farming: 100× yields per square metre. 95% less water. No pesticides. No weather risk. The constraint was energy cost; as energy falls, the economics flip.

Precision fermentation: engineered microbes producing dairy, egg proteins, collagen. Molecularly identical to animal products.

Cellular agriculture: from $300,000 per burger in 2013 to approaching price parity. AI accelerates growth media and bioreactor optimisation.

When energy is near-free, everything it touches changes. These are scarcity problems that were secretly energy problems all along.

Fresh water: desalination at planetary scale. Carbon capture: $400–600/tonne today, mostly energy cost; cheap energy makes reversal viable. Heating and cooling: heat pumps plus cheap electricity equals universal climate comfort. Transport: EVs, autonomous freight, electric aviation. Recycling: recovery becomes profitable for almost any material.

Universal Basic Income is the bridge. Universal High Income is the destination. As production costs vanish, the purchasing power of any given income rises without limit. A modest UBI in a near-zero-cost world delivers what today would count as luxury.

As each category crosses the Migrating Boundary, effective purchasing power increases. Food, energy, healthcare, education, legal services: each becomes near-free. Eventually the "basic" income buys everything anyone wants.

Funding is a transitional problem. Automation taxes, carbon dividends, sovereign wealth funds during the transition. Once marginal costs hit zero, "funding" becomes as quaint as "funding" access to sunlight.

"Fiat goods" is a term coined in FOME. Just as fiat money has value by collective agreement, fiat goods are scarce because we maintain systems that keep them scarce, even when the technology to make them abundant already exists.

Housing: building costs are falling; prices rise because the scarcity is in planning permission and zoning. Education: marginal cost of an online lecture is zero; tuition rises because the scarcity sold is the credential. IP: digital files copy at zero cost; artificial scarcity is imposed through copyright. Healthcare: treatments exist; access is rationed by systems designed around scarcity assumptions.

The danger: abundance arrives technologically while scarcity is preserved socially. A world where the means exist to feed, house, heal, and educate everyone, but artificial barriers prevent it.

"But what will people do?" Sounds practical; it is philosophical. Work has been the primary source of purpose, status, and self-worth. What replaces it?

FOME argues our relationship with work is "moral software" installed by scarcity. Cultures that valorised labour had survival advantage. These values are adaptive in scarcity; maladaptive in abundance.

Retiring this software does not mean abandoning purpose. It means expanding what counts as purposeful: care, creativity, curiosity, community, exploration, contemplation, play. These have always been among the most valuable human activities; scarcity economics simply could not afford to recognise them.

Every other node describes a technology or consequence. This one describes how humanity decides what to do with all of it. Without this, abundance defaults to whoever controls the technology. This is the most important node on the map.

The FOME architecture

The Book diagnoses how scarcity built our moral software and what is at stake.

The Salon is a global conversation series: scientists, economists, philosophers, artists, citizens, and leaders exploring the questions together.

The App is an AI-mediated deliberation platform that synthesises millions of contributions into genuine democratic mandate. Not a poll. Not a petition. A structured expression of what humanity collectively wants.

The Concerts are large-scale public events where that mandate is presented to world leaders in a setting that makes commitment visible and accountability inescapable.