Article · Essay
After AGI, where you live becomes a choice again
If artificial general intelligence cuts the link between job and address, water, climate, nature and land prices start to weigh more, and it is worth thinking about extreme scenarios too.
Join our WhatsApp community (in Portuguese) Be the first to see new cities, articles and videos.
This piece mixes data and speculation. The figures come from Onde Viver Bem, as updated on 18 November 2026, and every score quoted is indicative. Anything said about the future is a hypothesis and is marked as such.
When work stops tying you down
For many people, work and income strongly limit where they can live. Large job markets tend to be concentrated in the main urban centres, and everything else follows: high rent, traffic, worse air, nature far away.
There is no single definition of artificial general intelligence, or AGI; broadly, the term tends to describe systems with wide, adaptable intellectual abilities, comparable to human ones across a large range of tasks. Nobody knows whether it will arrive, when, or how its gains would be shared. What follows is speculation about one specific scenario: AGI arrives, output grows a great deal, and part of that gain reaches people as a high universal basic income.
In that scenario, the question "where is my job?" loses its grip. A different one takes its place: what does this place offer on its own? Water that doesn't run out. A climate that makes you want to go outside. The sea or the mountains nearby. Things to do. Safe streets. Land prices that fit your income.
There is a possible side effect, and it is speculative too. If a shift like this changed demand for certain cities, their relative prices could change as well.
What the ranking measures
The Onde Viver Bem ranking starts from that premise. It compares 175 cities in the global ranking, with separate sections for Brazil and the rest of the world. The score combines 13 variables with fixed weights: safety, beach or mountains, water, parks, sunshine, restaurants, entertainment, bars, mild climate, health, clean air, airport and culture. Cost of living and the price per square metre are not part of the score; the land price per m² is shown beside it, for comparison. Jobs and salaries are left out too.
That visibly changes the order. Vigo, in Galicia, has an indicative score of 7.86, and Annecy, in the French Alps, 7.65. New York, one of the largest job markets in the world, has 7.15, and Paris 6.89. In a score that does not measure jobs, the weight of a large labour market does not show.
Water enters the score through national indicators from the World Bank and the WHO/UNICEF JMP, the same for every city in a country. Local reality can be much worse: in Porto Velho, Brazil's SINISA 2024 data puts water service coverage at 30.74%. The city page shows that figure as complementary information; it is not part of the score.
The score vs. price-per-m² chart puts the two side by side: quality of life on one axis, the price of a square metre on the other. It is a useful tool for finding good places that have not yet become expensive. And the simulator lets you change how much each variable matters and see which cities fit you.
The scores carry uncertainty. When a figure is estimated, the city page says so and, where possible, shows a range. Porto Velho has an indicative score of 4.38, with a range from 4.06 to 4.40. Small gaps between cities that sit close together in the ranking don't mean much.
Extreme scenarios
The same question, what a place offers on its own, works for thinking about the worst case. There are high-impact scenarios with very different probabilities, which this site does not attempt to estimate:
- prolonged drought, regional water stress or episodes of extreme heat;
- armed conflict involving the country or the region;
- a pandemic more severe than the recent ones;
- a prolonged infrastructure failure: power, communications, supply chains;
- failures or misuse of AI systems affecting essential services (speculation).
The site has no data that would let us put a probability on any of them, and this piece will not make one up. Some resilience factors come up in several of these scenarios: local water, land that can grow food, low density, few natural hazards and less direct exposure to conflict.
The bunker ranking
That is what the bunker ranking is for. It does not compare cities. It compares 29 candidate rural regions chosen by the editors (8 in Brazil and 21 elsewhere in the world), where it would make sense to set up a retreat with some degree of self-sufficiency.
The score combines six pillars with indicative weights. Among them are "Surface water and growing climate" and "Measured natural hazards and climate". A seventh pillar, rural land price, has no score, because there is no comparable cross-country source. The index also looks at isolation, mapped seismic faults, fire detections measured by satellite from 2015 to 2025, distance from operating nuclear reactors and whether there has been recent armed conflict in the country. So as not to hinge on a single choice of weights, they are varied by ±30% in a Monte Carlo simulation. The result is a range of positions, not a fixed rank.
In the World section, Inland Uruguay (Tacuarembó, Rivera and Durazno) comes first, with an indicative score of 8.71, and stays first in every Monte Carlo run used in this version. In the Brazil section, the top three are the Southern Espinhaço (Diamantina and Serro, Minas Gerais) at 7.30, the Campos Gerais (Paraná) at 7.26 and the Serra Catarinense at 7.21, all indicative. The Espinhaço does not have a robust lead: across the simulations it lands between 1st and 3rd, and the Campos Gerais between 1st and 7th.
The site also shows the price of farmland. In Sweden, according to the Swedish Board of Agriculture (Jordbruksverket), arable land averaged SEK 48,300 per hectare in 2025 in the statistical region that includes Värmland and SEK 30,800 in the one that includes Jämtland. At the ECB reference rate of 13 November 2026 (about SEK 10.05 per US dollar), that is US$ 0.48 and US$ 0.31 per m². These values are derived from the official regional statistic, which covers a larger area than the ranking region, and converted to US$/m²; that is why the site labels them estimated. Cheap land, however, is not a cheap retreat. Building, maintaining and getting there all cost money.
The index has stated limits. There is no probabilistic seismic hazard, only mapped faults and historical seismicity. Where no fault has been mapped, the score treats that as a gap, not as a guarantee. Landslides are a gap in places such as the Espinhaço escarpments, and radon is a gap in the Limousin. Borders and rules on land purchases by foreigners have not been checked in every region. The ranking helps you ask the right questions. It does not replace a visit, a geologist or a local lawyer.
Practical questions to consider
What follows are general considerations, not financial or legal advice.
Normal times and extreme scenarios are separate questions. Most of life happens in normal times. A main city is usually chosen for what it offers day to day; the extreme case can be treated separately, not as a test for everything.
Accessibility is a relevant dimension. A fallback only serves that purpose if it can be reached when flights stop. A region reachable overland is a different proposition from one on the other side of the world.
Depending on a single jurisdiction can narrow options in some scenarios. Citizenship, residence, taxation and powers of attorney have legal consequences and need to be assessed case by case, by someone who knows the local law.
Water and energy often feature in contingency plans. At home, that can involve stored and filtered water and solar power with a battery. In a fallback region, surface water and rainfall are characteristics that one of the ranking's pillars measures.
Accessible documents. A valid passport and paper and digital copies of documents and property records often appear on preparedness lists.
The cost of a second home goes beyond the land. Rural land is usually cheap next to urban prices per square metre, but the bill includes building, taxes, upkeep, access and travel time.
Protection has a cost. Preparing for extreme scenarios takes money and time, and that cost also weighs on life in the more common ones.
For people who already work from anywhere, some of these questions already arise today.
Frequently asked questions
Will AGI really happen?
Nobody knows. This article treats AGI and a high universal basic income as a hypothesis, to think through what would change in choosing where to live.
Are the ranking scores precise?
They are indicative. When a figure is estimated, the page says so and, where possible, shows a range. In the bunker ranking, the weights vary by ±30% and each position is shown as a range.
Does the price per square metre count towards the score?
No. The city score uses 13 variables; cost of living and the price per m² are left out. The land price per m² is shown beside the score, for comparison.
Do I need a bunker?
The ranking does not establish that anyone needs a bunker. It is an exploratory tool for comparing territorial characteristics in contingency scenarios.
Is the bunker ranking a buying recommendation?
No. It compares regions using public data and has stated limits, such as borders and rules for foreign buyers that have not been checked. Any decision calls for a visit and local professional advice.
Compare as well
Cities mentioned in this article. The score is indicative; land price is not part of it.
Score 7.86 / 10 (indicative)
Score 7.65 / 10 (indicative)
Score 7.15 / 10 (indicative)
Score 6.89 / 10 (indicative)
Score 4.38 / 10 (indicative)
Share this page
Card for Instagram and TikTok
Download the image and post it to your feed, story or video. To send people to the site, use the link sticker in your story or the link in your bio (links in captions are not clickable).
Download card (PNG 1080×1350)Global ranking Score × m² chart Bunker ranking Simulator Methodology
