Unshackling Energy: Four Waves That Built the Modern World

In 18th-century Britain, power still came from muscle, animals, water, and wind — all bound by geography and season. The Newcomen engine had pumped water from mines since 1712, but it was ruinously coal-hungry and could do only one thing. James Watt, an instrument-repairman at the University of Glasgow, was fixing a Newcomen model in 1765 when it struck him: condense the steam in a separate condenser, and the cylinder need not be repeatedly heated and cooled — a huge fuel saving.

Watt patented the separate condenser in 1769, but a gulf of capital and precision machining lay between principle and production. In 1775 he partnered with the Birmingham industrialist Matthew Boulton; only John Wilkinson's precision boring machine — invented to bore cannon — could machine a cylinder accurate enough. In 1781 Watt used sun-and-planet gears to turn reciprocating motion into rotary motion — upgrading the engine from "mine pump" to universal power able to drive any machine. When the patent expired in 1800, high-pressure engines erupted.

Whether a general-purpose technology lands rarely turns on how advanced it is, but on whether using it to replace the current way pays off. So with AI — adoption speed is set by relative cost, not model capability.

Once steam was fixed inside the factory, the next step was to make it move. Trevithick built the first rail locomotive in 1804, but the track could not bear the weight; transport still crawled along canals and by cart, slow and fragmented. George Stephenson, a self-taught colliery mechanic, bet on locomotive haulage.

The 1825 Stockton–Darlington line mostly hauled coal; the real pivot was the 1830 Liverpool & Manchester Railway — the first locomotive-only trunk line linking two great cities for passengers and freight. In the 1829 Rainhill Trials, the Stephensons' "Rocket" won at 30 mph, setting the locomotive standard. On opening day, September 15, the MP William Huskisson stepped down and was killed by the oncoming Rocket — the most famous railway accident in history, yet no brake on the "Railway Mania" that swept the country.

What truly transforms an economy is rarely point efficiency (10% faster), but "wiring islands into a network": the internet, the shipping container, payment rails all did this. Value lives in the links, not the nodes.

The First Industrial Revolution was steam + cotton + iron. After 1870 a second wave surged: cheap steel (the Bessemer converter, 1856), chemicals, the internal-combustion engine — and at its core, electricity, a universal energy that could be transmitted over distance and distributed instantly. In 1831 Faraday discovered electromagnetic induction, the basis of generation.

In 1882 Edison built the first central station on Pearl Street, New York — but using direct current (DC), which reached barely a mile. The "War of Currents" erupted: Edison's DC against Westinghouse and Tesla's alternating current (AC), which could transmit at high voltage over distance. The 1893 Chicago World's Fair and the 1895 Niagara hydro plant sealed AC's victory. Yet the deeper insight came from Paul David: after electricity arrived, factory productivity stagnated for decades — owners merely swapped the steam engine for one big motor, keeping the old central line-shaft layout. Only in the 1920s, with "unit drive" (a small motor per machine) and shop floors rearranged by process flow, did productivity explode.

When productivity falls rather than rises after adopting a powerful new technology, it doesn't mean the technology is useless — it means the organization hasn't learned to use it yet. The early days of ERP, and of AI, look just like this.

Steam, electricity, digital — three waves of general-purpose technology. The 1971 Intel 4004 microprocessor ignited the digital revolution (details in Days 7–10). But this time, historians recognized the curve.

In 1987 the economist Robert Solow quipped: "You can see the computer age everywhere but in the productivity statistics." That is the "Solow Paradox" — IT investment soared through the 1970s–90s while total-factor productivity stalled. Only in the late 1990s, as the internet spread and firms rebuilt their processes, did productivity briefly recover. As early as 1990, Paul David had precisely predicted this using the historical analogy of "electricity's 40-year lag." Carlota Perez, Technological Revolutions and Financial Capital (2002) systematized it into long waves: every revolution passes through an "installation period (bubble, mismatch) — turning-point crisis — deployment period (diffused dividend)."

With AI, the real dividend may arrive after organizations learn to restructure, not the moment a model ships. Early "no effect" is neither a scam nor a failure — it is restructuring not yet complete.