Nearly three decades ago, James Dale Davidson and William Rees-Mogg published The Sovereign Individual: Mastering the Transition to the Information Age. Written in 1997, the book argued that information technology would shift power from the nation-state to the individual: knowledge, capital and eventually money would become mobile, letting skilled people choose where they lived, worked and paid taxes.[1]
Much of this was perceptive. Work has partly detached from place — a consultant or engineer can now serve clients across continents without leaving home — and talented individuals and firms increasingly shop among jurisdictions for favourable taxation and quality of life. The authors were also right about returns to skill: a single algorithm can serve millions of users, and AI is pushing this further. The IMF estimates close to 40 per cent of jobs globally carry some AI exposure, rising to 60 per cent in advanced economies; the ILO’s more conservative, generative-AI-specific measure puts it at one in four jobs worldwide, over a third in high-income countries.[2][3] Both institutions expect transformation, not wholesale replacement, in the near term.
Where the authors were less convincing was in foreseeing the state’s eclipse. Technology has empowered governments as much as individuals. States now hold digital identities, vast databases and automated systems of taxation and enforcement — AI use among OECD tax administrations has risen from 9 per cent in 2016 to 69 per cent in 2023, with a quarter more implementing it.[4] The Information Age has not transferred sovereignty from government to citizen; it has strengthened both. That contest — between the sovereignty of the individual and the sovereignty of the state — runs beneath everything that follows.
Artificial intelligence now presents a transformation potentially greater than the internet revolution the authors examined. I would venture eight propositions: four about what AI does to the individual’s leverage, a fifth about the rise and taming of cryptocurrency — the individual’s most literal monetary leverage — three about why the state’s leverage endures, and an eighth about what neither can touch.
I. ECONOMIC GROWTH WILL INCREASINGLY BECOME DETACHED FROM EMPLOYMENT
For two centuries, economic progress created new employment even as machines destroyed old jobs. AI may break that pattern. It could let firms produce more while employing fewer people, with the initial pressure falling on occupations built around manipulating information — clerical workers, accountants, junior lawyers, analysts, customer-service staff and layers of middle management — precisely the roles the ILO index identifies as most exposed.[3]
This does not mean these professions vanish; it means one AI-assisted worker may do what once took several. We could see jobless prosperity: rising output and income alongside weak job creation. The great economic question then becomes not how wealth is produced, but how a society distributes purchasing power, dignity and status when human labour is no longer needed in the old quantities — a question, as Proposition VI argues, only the state can answer.
II. THE SKILLED HAND MAY TEMPORARILY BECOME MORE VALUABLE THAN THE EDUCATED MIND
For generations, professional occupations were considered safer than manual trades. AI could reverse that, temporarily. A computer can draft a contract or write software fairly easily; replacing an electrician rewiring an old house, or a nurse caring for a difficult patient, remains far harder — the physical world is untidy in ways no manual anticipates.
The market is already pricing this in. American manufacturers project a shortfall of roughly 2.1 million skilled-trades workers by 2030, and trades wages and job satisfaction now compare favourably with many white-collar careers.[5][6] For an important interval, dexterity and practical judgement may command a premium precisely because cognitive routine is easier to automate than physical adaptability — and vocational education, long treated as inferior to a university degree, may enjoy a renaissance.
III. THE INTERMEDIATE AGE COULD PRODUCE ITS OWN GOLDEN GENERATION OF WORKERS
Between today’s AI systems and tomorrow’s capable general-purpose robots lies an interval that deserves more attention. Machines may possess extraordinary intelligence while still needing humans to act in the physical world. The valuable worker will be neither the traditional labourer nor the traditional professional, but both at once: the electrician using AI diagnostics, the farmer running autonomous equipment, the nurse working alongside medical AI.
That interval is the present moment. Even the most advanced humanoid robots remain largely at proof-of-concept and small-batch stage, with global shipments estimated at only 16,000–18,000 units in 2025.[7] The premium today belongs to people who can connect machine intelligence with physical reality. Learning to use AI will be essential for the young, but it may not be enough on its own — a practical skill or physical competence could prove more resilient during this transitional window, one of history’s great, if temporary, opportunities.
IV. BUT THE ROBOT WILL EVENTUALLY COME FOR THE SKILLED HAND
That sanctuary will not last indefinitely. Robotics was never limited by mechanics — motors, wheels and arms have long existed. The harder problem has been intelligence: reading an unpredictable environment, judging force, learning from mistakes. AI is progressively solving that, and the capital behind it is serious: humanoid robotics start-ups drew roughly $4.3 billion of the $8.5 billion invested in robotics in 2025, with Goldman Sachs projecting a $38 billion humanoid market by 2035.[8][7]
Once machines can learn physical tasks from demonstration, and knowledge from one robot can transfer instantly to thousands of others, the economics of manual work will change. A human apprentice needs years to become proficient; a network of machines can learn collectively. After AI challenges the educated mind, embodied AI may challenge the skilled hand — a squeeze on individual sovereignty from both directions at once.
V. CRYPTOCURRENCY WILL RISE, THEN MEET THE STATE IT WAS BUILT TO ESCAPE
Davidson and Rees-Mogg predicted that money itself would become mobile, and cryptocurrency is the clearest fulfilment of that prophecy — digital, borderless, resistant to a single government’s control. In the near and intermediate term, it will likely remain relevant and grow further, both as a speculative asset and, increasingly, as regulated payment infrastructure: stablecoin market capitalisation has more than doubled since 2023, and 2026 has become the year enforcement replaced legislation, with the US GENIUS Act, the EU’s MiCA and the UK’s new cryptoasset regime all becoming operational within the same twelve-month window.[12][13]
But that same maturing is the tell. Over 92 per cent of jurisdictions worldwide have tightened crypto rules in some form, and regulation of this kind rarely stops at licensing and disclosure.[14] Reserve mandates, audit requirements and AML/KYC obligations are steadily converting a currency built to bypass the state into one the state supervises at every issuance point — and a state that can license a stablecoin issuer can just as easily unlicense one. The infrastructure question compounds the legal one. Bitcoin mining and AI training now compete directly for the same scarce grid connections, land and cooling capacity, with a single AI compute deal now claiming multiple gigawatts that miners once counted on; several major miners have already begun converting their facilities to AI hosting because it pays better than hashing.[15][16]
There is a third pressure, subtler than regulation or energy, and it comes from the code itself. Cryptocurrency’s promise of sovereignty rests on the assumption that its underlying algorithms are sound; increasingly capable frontier models are eroding that assumption. Anthropic’s own safety researchers have shown that models including Claude Opus and GPT-5-class systems can autonomously discover and exploit vulnerabilities in blockchain smart contracts, producing exploits worth hundreds of millions of dollars in controlled benchmarks and even uncovering genuine zero-day flaws in live contracts.[19][20] The pattern is already playing out beyond the laboratory: a security researcher used an AI coding agent to catch a critical bug in the Zcash network that had gone undetected for four years, while CertiK recorded over a billion dollars in crypto losses in the first half of 2026 alone, with AI-assisted tools increasingly cited as the means of discovery.[21] The same capability cuts both ways — the ethical hacker auditing a protocol before launch and the criminal draining it after are now drawing on the same class of model. As these systems grow more capable still, the sifting-out of vulnerabilities, and the fortunes that ride on them, may happen at a pace no human audit team can match.
If regulation squeezes crypto’s legal space, AI’s hunger for power and chips squeezes its physical one, and frontier models probe its mathematical foundations for weaknesses, the long-run trajectory may not be extinction, but something closer to captivity: a technology built for sovereignty from the state ending up licensed, audited, energy-rationed and algorithmically besieged all at once. It would be a fitting, if ironic, coda to the book that first predicted its rise.
VI. POLITICS WILL REMAIN SUPREME
Here I part company with futurism that foresees the state withering away. Technology determines what can be done; markets determine what is worthwhile; but politics determines what society permits, taxes, distributes and protects. If AI and robotics eventually generate enormous output with little human labour, questions of ownership become unavoidable. Who owns the algorithms, the data, the computing infrastructure — and what share, if any, goes to citizens no longer producing directly? These are political questions, not engineering ones.
States possess something even the largest technology companies do not: sovereignty — the power to tax, regulate, and enforce, which they are already deploying more effectively through AI, not less, as the OECD’s catalogue of over two hundred government AI use cases shows.[4][9] A society needing less human labour may become more political, not less, since distribution grows more contentious as production grows easier. The sovereign individual will still need a passport, a court system and a currency — all in the state’s gift.
Davidson and Rees-Mogg’s heirs did not stop at money. In The Network State (2022), the technologist Balaji Srinivasan proposed a literal sequel to their thesis: found a community “in the cloud” first, build its economy and culture, then acquire dispersed physical territory and, eventually, seek diplomatic recognition as a sovereign state in its own right. In 2024 he took the first concrete step, opening Network School — a paying membership community for “techno-optimists” — in Malaysia’s Forest City. It lasted barely two years. In July 2026, following an immigration investigation and viral allegations about which passports its members held, the Johor state government revoked its business licence outright and ordered all activities to cease within a day; Srinivasan relocated the project to Kazakhstan within the week.[22] A single municipal council, exercising ordinary licensing power, ended in an afternoon what a widely read manifesto had proposed as the first rung of a new sovereignty. The network state, it turns out, still needs a landlord.
VII. RELIGION, IDENTITY AND THE SEARCH FOR MEANING WILL SURVIVE THE ALGORITHM
Human beings are not simply economic units maximising income; we seek belonging. Religion, family, nation, language, caste and inherited memory have survived earlier technological revolutions because they meet needs technology does not remove — and AI could make these attachments more important, not less.
Work has always given people more than wages: routine, status, hierarchy, and the satisfaction of being needed. If employment no longer supplies these, people will seek them in faith, family, community or new digital tribes — with both constructive and dangerous possibilities, since identity can offer solidarity or become a vehicle for grievance and exclusion. Our technology may grow universal while our identities stay intensely particular. AI may globalise intelligence; it cannot globalise the human soul — the one sovereignty neither algorithm nor state has yet annexed.
VIII. CHARACTER WILL OUTLAST COMPUTATION
A final category resists automation altogether. Courage, resilience, endurance, emotional intelligence, empathy, verbal communication, sound judgement and intuition are not skills but qualities of character, forged through lived adversity — a district officer deciding at 2 a.m. whether to order a lathi-charge, a surgeon reading fear in a patient’s eyes, a negotiator sensing the moment an adversary will bend. These are acts of the whole person, not information-processing tasks.
Employers are already pricing this. The World Economic Forum’s Future of Jobs Report 2025, surveying over a thousand employers across fifty-five economies, finds resilience, leadership and social influence among the fastest-rising core skills, with empathy and self-awareness in the top ten.[18] AI can draft fluent sentences, but the spoken word — persuading a hostile crowd, calming a grieving family, holding a room through presence — draws on timing and trust that text generation does not touch. It can simulate empathy but cannot be held accountable for a decision, or summon the nerve to make an unpopular call and live with it. Intuition, built from decades of pattern recognition under real stakes, remains human because it is inseparable from having something to lose.
If Propositions I–IV describe a contest between machine and mind, Proposition V a contest between the individual and the state over money, and Proposition VI a contest between citizen and state over ownership, this eighth is not a contest at all. Character is not a leverage point in a negotiation with technology; it is the ground on which every other negotiation stands. A society may automate its clerks and eventually its electricians. It will not automate the person who must decide, absorb the consequence, and carry on.
PEERING AHEAD, NOT PROPHESYING
These eight propositions are not prophecies. Technology does not advance along one predetermined track, and societies do not respond identically to the same invention. Governments regulate, markets adapt, cultures resist, and unexpected events repeatedly overturn the futurist’s straight line. Nor should we claim telescopic eyes capable of seeing decades ahead.
What we can do is extrapolate from forces already visible: capable AI, the growing exposure of cognitive work to automation, advancing robotics, the enduring authority of political institutions, and humanity’s persistent attachment to faith, family and community. Davidson and Rees-Mogg got some things spectacularly right and others significantly wrong; their value lies not in a map of the future but in showing how seriously to think about technology while it is still young. Our task is similar — not because our eyesight is telescopic, but because we stand on the shoulders of giants, from where we may still misjudge the distance or direction. But it is wiser to look ahead, and debate what we see, than to wait until the future has all but arrived.