In 1847 Britain was putting roughly 7% of everything its economy produced into railways. Our new desk’s first issue follows that story from the railways to AI, and asks where today’s boom has got to.
In 1847, British railway investment peaked at about £44 million, roughly 7% of GDP, the economy’s total output. The railway construction workforce was about twice the size of the British Army, and the boom ended in a bust.
America then repeated the story on a bigger scale. In 1893 alone, more than 27,000 miles of US railroad, capitalised at almost $2 billion (about $75 billion in 2026 dollars), went into receivership. Yet the railways became the backbone of both economies. The technology won. Many of the people who paid for it did not.
Cars came next. The number of US car producers peaked at 272 in 1909 and fell to nine by 1941. The Model T launched in 1908 at $850, about $31,900 in 2026 dollars; by the early 1920s a two-seat version cost as little as $260, about $5,000 in 2026 dollars. Drivers did very well. Most of the makers did not last.
Air travel repeated the pattern after deregulation in 1978. Passengers won; airlines paid: Airlines for America, the industry body, counts 198 US airline bankruptcy filings between May 1979 and February 2016 in an unofficial compilation.
In the five years after 1996, telecom carriers put more than $500 billion into fibre, switches and wireless networks, roughly $1 trillion in 2026 dollars. No more than 2% of North American long-distance capacity was in use. Bandwidth became almost free, and shareholders lost roughly $2 trillion (about $3.8 trillion in 2026 dollars). Chips followed a similar arc: the main US chip-stock index fell 80% from March 2000 to September 2002 and took 17.6 years to regain its 2000 peak.
Each time, a scarce and expensive new technology pulled in a flood of money, capacity outran demand, prices collapsed, and most of the companies fell away.
Where does AI sit in that story? Some of it fits. The cost of getting the same level of performance from an AI model has been getting between nine and 900 times cheaper a year, depending on the test. US spending on computer equipment and software reached 5.35% of GDP in the third quarter of 2026: above the 4.46% peak of the dot-com boom. The builders are financing part of it with debt and equity; Meta alone priced $37.5 billion of new stock, ordinary and preferred, before options for more were taken up. And Nvidia, the AI build-out’s main supplier, lost about 30% of its value in six straight weekly falls to 13 November.
Other parts have not happened. Our first issue rates the shakeout as only “partly (early)”: start-ups are being absorbed, but there has been no wave of failures among funded leaders. Consolidation is “partly” too: yes in infrastructure, not yet among the model makers. Users gain on price, the supplier earns most, and the builders are mixed. The main US chip-stock index’s 30% fall from its June high is a correction so far, not a bust on the scale of 2000–02. And AI may yet differ: its main spenders are some of the most profitable companies in history, where the railways and telecom were funded by leveraged newcomers. We do not know how this chapter ends.
That is why we have started Blindspot. It is a new MoneyTurkey desk, separate from our daily missed-markets desk, for the big structural shifts that markets tend to miss: the blind spots. Each long-form issue tests one shift against history and hard data, not hype or tips. The aim is an honest map of where we are in a cycle, so readers can understand the risk in what they own. Blindspot gives no investment advice.
Read the first issue: The Commoditisation of High Tech.
