In 1800, roughly 90% of humanity lived in extreme poverty. Life expectancy was under 40 years everywhere on earth. Nine in ten people could not read. Today, 847 million people still live in extreme poverty — and that number, as terrible as it is, represents the sharpest reduction in human suffering in all of recorded history. This letter traces the full arc: where we came from, where progress has stalled, and what the next wave of technology may do to it.
Not investment advice. Not a recommendation to buy or sell. Research and long-horizon thinking only. Consult a qualified financial advisor before making any investment decision. Figures cited are sourced from the World Bank Poverty and Inequality Platform (March 2026 update), Our World in Data, the UN SDG reporting database, the Dallas Fed, and BCG's 2026 global scenarios analysis, current as of writing.
In 1800, the world was almost uniformly poor by any standard we recognise today. Every country on earth had a life expectancy at or below 40 years — not because any particular government was failing, but because this was simply the human condition for all of prior history. Extreme poverty, defined as living on less than the equivalent of $3 per day in today's terms, was the near-universal experience of humanity. Nine in ten adults could not read. Child mortality ran at approximately 40% — four in ten children born did not survive to age five. These were not the statistics of a struggling country. They were the statistics of the entire planet.
What happened next is the most underreported story in economics. The Industrial Revolution, beginning in Britain in the late eighteenth century and spreading outward across the nineteenth and twentieth, broke the Malthusian trap that had kept living standards flat for millennia. For the first time in human history, the economy began producing more per person each decade than the last. The effects compounded. By 1930, global literacy had reached roughly one in three adults. By 1950, global extreme poverty had fallen to approximately 63% — still devastating, but moving. By 2015, the World Bank's older $1.90 per day measure had the global extreme poverty rate at roughly 9.5%. By 2021, using the same old line, it had fallen further still to 3.7%.
"A child born today in the country with the lowest life expectancy on earth will likely outlive an English or American child born in 1800. That is what two centuries of human progress looks like." — Our World in Data
The story of global living standards in 2026 is not a single story. It is two stories running simultaneously, so different from each other that they barely feel like the same era. In most of Asia, much of Latin America, and increasingly in parts of Africa, the arc of the previous two centuries has continued: urban migration, trade integration, and technology adoption have lifted hundreds of millions in living memory. But in Sub-Saharan Africa, and specifically in fragile and conflict-affected states, the improvement that defined the twentieth century has largely stopped, and in some places reversed.
The March 2026 World Bank update puts 847 million people in extreme poverty. Sub-Saharan Africa accounts for 71% of the global extreme poor despite being only 16% of world population. Within Sub-Saharan Africa, Nigeria and the Democratic Republic of the Congo alone are projected to host one-quarter of the world's extreme poor by 2030. Children are dramatically overrepresented: roughly 412 million children under 17 live in households earning less than $3 per day, more than half of all those in extreme poverty despite being only 30% of the global population. And the pace of change has dramatically slowed: the World Bank projects only 69 million people will escape extreme poverty between 2024 and 2030, compared to 150 million who did so in the six years between 2013 and 2019. The "lost decade" framing the World Bank has adopted for the 2020s is not rhetorical.
Every previous wave of living-standards improvement was driven by a general-purpose technology — the steam engine, electrification, the internal combustion engine, the green revolution in agriculture, the internet. Each took decades from invention to broad welfare impact, and each left behind the populations least able to access the enabling infrastructure. The question worth asking in 2026 is whether AI constitutes the next such wave, and if so, what makes it different from the waves that came before.
The Dallas Fed's 2025 analysis of US GDP per capita since 1870 — growing at approximately 1.9% annually through two world wars, the Great Depression, and every prior technological revolution — puts the current AI transition in sober historical perspective: productivity growth has always been the single most important determinant of living-standards improvement, and no prior technology shock derailed that trend for long, even the ones that seemed at the time as transformative as AI seems today. The BCG 2026 "AI Abundance" scenario models labor productivity growth in high-income countries at 5.7% — nearly three times today's 2% rate — sufficient to support aging populations with expanded social safety nets through 2050. The "AI Concentration" risk scenario runs in the opposite direction: global GDP tripling by 2050, but the richest 1% holding nearly half of all wealth, a share not seen since the industrial societies of the early 1900s.
AI diagnostic tools are beginning to reach populations that have never had consistent access to specialist medicine — low-cost AI-assisted TB screening, malaria detection, and maternal health monitoring are in active deployment in parts of Sub-Saharan Africa and South Asia. The bottleneck historically has been specialist scarcity; AI compresses the specialist-to-patient ratio in ways that have no precedent in previous technology waves.
AI-driven precision agriculture — yield prediction, pest detection, soil analysis via satellite — is showing documented 15-25% productivity gains in early deployments across smallholder farms in Asia and Africa. For the 500+ million smallholder farmers who remain the backbone of food production in the poorest countries, this is the most direct living-standards lever available.
AI tutoring systems that adapt to individual learning pace are demonstrating genuine catch-up gains in contexts where the alternative is severely under-resourced classrooms, not high-quality alternatives. Unlike every prior education technology (TV, radio, early internet), these tools can provide meaningful two-way interaction without a trained teacher in the room.
AI credit-scoring using non-traditional data — mobile phone usage patterns, purchase history, location data — is extending access to financial services to populations that lack the formal credit histories that traditional banking requires. M-Pesa's trajectory in East Africa is the model; AI-powered lending is currently expanding that trajectory into South Asia and West Africa.
The historical record is genuinely, unmistakably positive — and is also genuinely at risk of being misread as automatic. The poverty reduction of 1990-2022 was not an accident of technology alone; it required sustained political stability, trade openness, and institutional investment in health and education, primarily in Asia. The regions where progress has stalled in 2026 — fragile, conflict-affected states in Sub-Saharan Africa — are stalled precisely because those conditions are absent, and no technology resolves a civil war or fills an institutional vacuum on its own. AI reaching these populations will require the same enabling conditions that trade and industrialisation required, and those conditions are political, not technical.
The BCG concentration risk scenario deserves as much attention as the abundance scenario. The previous industrial revolution that lifted living standards globally also produced the Gilded Age, peak inequality, and the social conditions that drove the political upheavals of the early twentieth century — before the benefits diffused broadly enough to produce the mid-century prosperity that most people in wealthy countries take as the historical baseline. If AI follows the same pattern, the welfare gains are real but they arrive decades after the concentration gains, and the intervening period is politically unstable in ways that can set progress back sharply. The 2020s "lost decade" the World Bank describes is not unrelated to the distributional tensions the previous technology wave produced and left unresolved.
The honest summary of two centuries of global living standards is this: humanity escaped a trap it had been in for its entire recorded history, and it did so faster and more completely than anyone who lived through 1800 would have thought possible. The 847 million people still in extreme poverty in 2026 are not evidence that the arc failed — they are evidence that the arc is not yet finished, and that it bends more slowly where the enabling conditions of stability, institutions, and access are absent. The question for the next quarter century is not whether AI will accelerate that arc — it will, where the conditions permit — but whether the political and governance work of extending those conditions to the people progress has not yet reached will keep pace with the technology that is waiting to serve them.
Long-horizon thinking on capital, technology, and the forces shaping the next decade of wealth creation. Written from first principles. Not consensus. Not noise.