In 1865, British economist William Jevons formulated a paradox that is still in force today. Discovered that When steam machines improved their efficiency and began consuming less coal per unit energythe result was not a fall in the consumption of coal, but a mass increase. The cost reduction led to its adoption in more sectors, expanding the market instead of contracting it.
Now, Satya Nadella, CEO of Microsoft, has brought this idea to the debate about artificial intelligence. I mentioned it recently in a tweet, when With the arrival of more efficient models such as Deepseekmany wonder: if the computing becomes cheaper, will the demand for chips and servers in the cloud collapsed? Or, on the contrary, will we see an explosion in the use of AI that further triggers the demand for technological infrastructure?
To understand it better, let’s return to the nineteenth century and the rise of the railroads.
Railways and artificial intelligence: history is repeated
At the dawn of the railroad, building roads and locomotives was a expensive and complicated process. Only great merchants and governments could afford this luxury. However, when rail technology became more efficient and cheap, something unexpected happened: instead of reducing investment in trains, the world lived a railway explosion. Disconnected cities were integrated into global markets and trade flourished at unimaginable levels.
Artificial intelligence is at a similar time. Today, training AI models is very expensive. It requires thousands of advanced chips such as Nvidia, huge data centers and a huge energy infrastructure. If the efficiency of the models improves and the costs fall, the superficial logic would say that the investment in hardware should be reduced.
However, Jevons’s paradox suggests otherwise: The accessibility of AI will trigger its use, multiplying the need for chips, servers and energy.
The efficiency trap: more accessibility, more consumption
When the first train lines were built, only the richest merchants could move long distance merchandise. But when the costs went down, The railroad ceased to be a luxury and became a need. The same goes for AI.
Today, training an avant -garde artificial intelligence model can cost hundreds of millions of dollars. But if Deepseek and other advances reduce these costs, access to AI will democratize. Smaller companies, local governments and even individual developers will be able to use advanced models to optimize their operations, improve medical care, automate industries and transform education.
The result will be an explosion in the computer demand. As well as railway efficiency multiplied the need for steel, coal and distribution networks, efficiency in IA will multiply the demand for chips, servers and energy.
The Nvidia and Microsoft case: Paradox winners
If the Jevons paradox is true, the biggest beneficiaries will be the companies that control the AI infrastructure. Nvidia, who designs the necessary GPUs to train models, will not see a drop in the demand for its chips. On the contrary, as more companies adopt, they will need more hardware.
Microsoft, with its Azure platform, is also in a key position. Cloud computing is the backbone of artificial intelligence. If access to AI expands, more companies will depend on servers in the cloud to process data and execute advanced models.
Thus, although the efficiency in IA reduces the cost per computing unitthe total number of users and applications will increase so much that the general demand for infrastructure will not fall, but will grow exponentially.
The future: omnipresent and new challenges
If the history of the railroad tells us something, it is that artificial intelligence is far from reaching its roof. Accessibility will bring challenges such as energy consumptionthe regulation of data use and the concentration of power in a few technological companies.
But if there is something clear, it is that AI, like the train at the time, is about to become a fundamental element of society. It will not be a luxury, but a necessity. And, as Satya Nadella pointed out, Jevons’s paradox teaches us that efficiency does not reduce the use of a technology: it accelerates it to limits that we do not even imagine today.
#collapse #investment #future #artificial #intelligence #law #nineteenth #century
