Tech
The link between bulldozers and AI •
Before being a political symbol, data centres are a property business
The AI race is propelling not only Nvidia but also Caterpillar, lorry drivers, construction firms and logistics companies. Behind ChatGPT lie bulldozers, cables, water and energy: so ‘digital gold’ requires very tangible infrastructure

“During a gold rush, sell shovels and pickaxes”: this old saying, attributed to Mark Twain, became widespread during the Californian gold rush of the second half of the nineteenth century, when the promise of easy riches attracted thousands of aspiring and improvised gold diggers. Very few of them actually succeeded in getting rich; all of them, however, bought the equipment they needed to give it a go. And so, shovels and pickaxes.
The idea that every speculative bubble conceals an opportunity for growth in less high-profile sectors is borne out once again today, in the race for artificial intelligence. At the heart of it all, of course, are cutting-edge products such as Nvidia’s GPUs – graphics processing units originally designed to enhance graphics in video games but which have ended up being used as key components in the development of artificial intelligence models.
And then there are the data centres – the large data processing centres that house the computing power needed to run ChatGPT, Claude and the like. These are enormous, expensive and increasingly controversial structures, due to their consumption of water and electricity, to the extent that they have sparked an almost bipartisan opposition to their construction in the United States. Even before they become a political symbol, however, data centres are a property business: large-scale buildings that need to be designed and built (and cabled, of course), bringing prosperity to a range of sectors that we do not normally associate with Nvidia’s latest-generation chips. Construction, transport, goods handling: the shovels and pickaxes of artificial intelligence are, almost literally, shovels and pickaxes.
Suffice it to say that one of the shares linked to artificial intelligence has nothing to do with IT: it is that of Caterpillar, the world’s largest manufacturer of construction and mining machinery, which, as such, is benefiting greatly from the construction of data centres. Over the past year, the company’s share price has risen by around 90 per cent, despite a sharp fall from its July highs, when it reached $1,000 per share (it currently stands at around $810). The cause of the fall is said to be the growing opposition to data centres, which risks hindering the construction of these buildings, with serious consequences for Caterpillar.
This is nothing new, let’s be clear. For some time now, economists have been measuring the vitality of the global economy partly on the basis of orders received by companies such as Caterpillar: after all, the more construction there is, the more GDP tends to rise. However, the link between bulldozers and artificial intelligence is a new development, confirming how a software service such as ChatGPT is having an increasingly tangible and physical impact.
Opposition to data centres is based on this too, of course. It is not enough for the big names in the AI sector to promise ever more powerful systems capable of replacing the work of millions of people; to do so, they also need space, water and electricity.
On the other hand, lorry drivers and, more generally, those working in logistics have reason to celebrate. In fact, two-thirds of the data centres currently under construction in the United States are located in rural areas, where building materials need to be transported and where new job opportunities are being created for regions that would otherwise be economically depressed. According to Janelle Griffith, a logistics expert interviewed by CNBC, “data centres are not the end of the supply chain, but its centre”, and they are said to have beneficial effects on many professional groups, including, in particular, the aforementioned lorry drivers and warehouse staff. Not everyone agrees, however.
Some point out that the demand for movement and transport in this sector does not last forever: once a data centre has been built and cabled, it requires nothing further, unlike a warehouse or distribution centre, which continues to communicate with the rest of the logistics network. Rural areas therefore risk reverting to the previous situation, with the sole addition of a large data centre on their doorstep.
However, these remain potential future issues. At present, demand for transport remains extremely high, partly because once one data centre is completed, another is ready to be worked on. The business of diggers and pickaxes continues to boom, so long as there are those convinced they can strike gold.