Power

The numbers are eye-watering. Over recent quarters, the largest technology companies have collectively committed close to $600 billion to artificial intelligence. Most of it is flowing into data centres, fuelling what is being billed as a once-in-a-generation investment cycle. And it seems far from over. According to McKinsey, a further $6 trillion could be spent by 2030.

Much of this looks defensive. No one wants to be the platform that fell behind because it hesitated. But building vast data-centre capacity does not automatically confer pricing power or attractive returns on capital. The killer statistic sits quietly in the background: data centres already consume electricity on a scale comparable to mid-sized countries, and AI workloads are expected to multiply that demand.

Then there is the ever-increasing cost of talent. In 1996, Alan Shearer moved from Blackburn Rovers to Newcastle United for £15 million, a world-record fee at the time. Today, that sum would barely register. The same inflationary logic now applies to AI talent, and no self-respecting superstar moves for anything less than a fortune, including stock options please. Unlike footballers, however, they do not sell tickets or shirts; they simply raise fixed costs.

For now, markets are giving Big Tech the benefit of the doubt. Balance sheets are strong, earnings are intact, and who is going to argue with putting data centres in space? However, given the recent negative reactions in stock prices from the likes of Alphabet, Meta and even Amazon, it seems investors are becoming less concerned about who wins the AI race and more about who ultimately foots the bill.