We have big problems here on planet earth. Most self-inflicted. In the software industry, we tend not to be too concerned; instead, we focus inward. No wonder we exaggerate our revolutions.

AI is a good example. We talk about AI as if it’s such a big thing, but how does it compare to the costs involved with other revolutionary technologies? Let’s calculate the costs of infrastructure and societal changes across the main technological revolutions of the past 200 years.

🚂 The First Industrial Revolution: 1.5 trillion today’s euro equivalent for the development of the initial hardware, and another 6T for societal changes.

Example: the Bridgewater Canal was built by Francis Egerton, 3rd Duke of Bridgewater to transport coal from his mines in Lancashire to the factories of Manchester. It cost about 40 million in today’s money and that was only the initial stretch.

Railroads: the USA had 200,000 miles of track by 1900, costing tens of billions.

Societal: large scale urbanisation as peasants were forced to move from the countryside to work in burgeoning industrial towns which quickly grew into cities.

💡Electricity was the driving force behind the Second Industrial Revolution. Power stations and grids were built, factories were electrified and eventually homes. Made assembly lines possible. More huge societal changes. Costs: approximately 6 trillion..

Pearl Street Station, Edisons first, serviced only 85 customers with 400 lamps, but cost 8 million to build.

🚗 The Rise of the Automobile, especially in America, brings the necessity to build roads: that cost more than 5 trillion; people move out of the city back to the countryside. Suburban living: 10 trillion.

📞 Telecommunication arrives in the middle of the 20th century. Massive infrastructure investments, such as cabling, and for television and radio. Instantaneous communication brings more impactful societal changes. Estimated at 10 trillion euros in today’s money.

🛜 The Internet. Fiber optics, servers, other hardware. Changes the way we work dramatically. Investments are a staggering 40 trillion.

🤖 Artificial Intelligence: built on top of the internet, investment is much less than previous revolutionary changes. Just 0.5 trillion on hardware such as chips and datacentres, and it’s estimated that the costs of changes in society will be another 1.5 trillion.

While Egerton’s canals bring almost immediate profits, most other periods of revolutionary technological change are not so good on their investors. The railroads aren’t profitable until the late 19th century. In the US, their promise of opening up the west is superseded by the automobile. Electrification takes decades more to reach households, Fords Model T is affordable but there are no roads to drive on until the thirties.

And then there is the internet. Paul Graham launches Viaweb, later Yahoo Store, in 1995. It is projected that e-commerce revenue will reach 1 trillion in 2000; instead, the bubble bursts. But of course, you and I wouldn’t be looking at this screen if the internet didn’t succeed.

AI might need subsequent revolutions to be what investors need it to be. A single training run for the latest GPT models can cost up to 500 million. A failed run could easily bankrupt companies with less investor traction than OpenAI.

Right now AI is a tiny revolution compared to the technological changes that brought the good, and the bad, that make up our civilisation. Take projections that it will change the world in 5 years with a grain of salt.