Field Note
The Railway Bubble of 1847
How an echo from the past explains the AI bubble today.
By the autumn of 1845, railways had captured the popular imagination in Britain.
This was not a faster horse. It was the disruptive innovation of the age. Before, travel was based on the vagaries of horse, mud, and weather. The Liverpool and Manchester Railway had already proved that steam could compress distance itself. Coal. Cotton. Mail. Newspapers. Fish. Strawberries. Clerks. Soldiers. Gossip. All of it could move differently now. Railways were the operating system of the new era.
Liverpool and Manchester, London and Birmingham, Grand Junction, and York and North Midland were all reported at 10 percent dividend rates during the boom period. Railway share prices roughly doubled between 1843 and the autumn of 1845. Parliament was flooded with railway bills. Newspapers carried railway advertisements, prices, warnings, route disputes, committee names, engineer reports, and rumors. Everybody wanted in.
And new creative financial instruments were created to meet that demand. The railway scrip was one of those inventions. A subscriber did not need to pay the full nominal value of a railway share upfront. A deposit could secure an allotment. The scrip certificate could then be traded before the railway was fully authorized, fully capitalized, fully built, or operationally proven. Future calls on capital would come later. The immediate object was tradable now.
Gareth Campbell and John Turner's work on Railway Mania investors showed the breadth and quality of participation. Merchants, gentlemen, legal professionals, MPs, railway directors, and local investors were on the subscription lists. Charles Darwin invested in the railways too. His records later show dividends from Great Western Railway shares, and in August 1847 forty guaranteed shares in the Leeds and Bradford Railway were bought with money from Emma Darwin's trust fund. Measured against Anne Bronte's governess salary, roughly normalized to a modern $60,000 a year, the Bronte sisters - Charlotte, Emily, and Anne - had the majority of their life savings from Aunt Branwell's legacy, the equivalent of about $750,000, tied up in York and North Midland shares.
Normally, when price outruns economic reality, the system should begin to tremble. Prices fall. Capital retreats. The error starts correcting. But the scrip delayed that tremor. It unlocked more buyers and more capital into the system. Premiums continued to rise. More lines could be financed. Demand for railway exposure started to look like validation of railway economics.
Think of an earthquake fault. Small slips release pressure. A locked fault holds its shape. It looks stable until too much force accumulates, and then the movement is not a tremor. It is an earthquake. Belief in the transformative power of railways was the counter-pressure that held against the increasing divergence between railway prices and the economic reality on the ground.
The crash came in 1847. The immediate cause was not one thing. It was a liquidity squeeze that hit a railway system already committed to huge capital calls. In April 1847, money-market pressure intensified as the Bank's banking reserve fell. By autumn, credit was tight, commercial failures were spreading, and the system had to find cash at exactly the moment railway shareholders were being asked to meet calls. Investors, including a broad range of the middle class, were devastated. Charlotte had wanted to sell at the height of the bubble. She could not persuade Emily and Anne. When the crash came, roughly $450,000 of that money vanished.
Those who argue there is no AI bubble today will point to the insatiable demand for chips, for data centers, and for inference. They will point to AI, rightly so, as one of the most transformative technological shifts of our age.
That misses the point.
The modern version of the railway scrip is the circular financing among AI labs, chipmakers, hyperscalers, and cloud providers. Hyperscalers bear enormous capital costs to build AI infrastructure. AI firms use that infrastructure and help boost cloud revenue. Chipmakers finance or support customers who buy the chips. Cloud providers book the revenue. AI firms show scale. Infrastructure providers show record revenue. Participants see a wall of demand.
But behind that are creative financial instruments - private credit vehicles, BDCs, hyperscaler-backed lease structures, GPU-backed debt instruments. Corporate bonds sustained by depreciation schedules don't match the actual value of last-generation GPUs when the new one hits the market. Just like the railway scrip, they add more capital to the system and extend the divergence between price and economic reality.
And more and more potential energy is stored for the snap that will inevitably occur.
Calculation Notes
Anne Bronte earned about 40 pounds per year as a governess in 1845. The family railway stake is estimated around 500 pounds. Charlotte wrote in 1849 that York and North Midland shares originally priced at 50 pounds had fallen to 20 pounds, implying a 60 percent loss from par. Normalizing 40 pounds per year to a modern $60,000 annual income makes 500 pounds equivalent to 12.5 years of income, or $750,000. A 60 percent loss equals about $450,000 on that normalized basis.
Sources
- Gareth Campbell and John Turner, Dispelling the Myth of the Naive Investor during the British Railway Mania.
- Darwin Correspondence Project, DCP-LETT-1042.
- Orlando, York and North Midland Railway Company.
- Andrew Odlyzko, railway mania, dividends, and accounting context.
- Hansard, London and York Railway scrip inquiry, August 5, 1845.