For those who follow the financial markets even a little, it is no secret that the race in the field of artificial intelligence is currently being fought out among the largest technology companies. However, AI is not a free revolution. To develop ever more efficient models, Big Tech needs hundreds of billions of dollars for data centres, servers, chips, cooling and energy. It is astonishing, but according to an analysis by the IEA (International Energy Agency), data centres are expected to consume around half of the electricity produced in the US between 2025 and 2030.
This is beginning to be reflected in the cash flow of the major companies in the AI sector. The chart below shows the growing gap between the capital expenditure (CapEx) of the Big 4 – namely Amazon, Alphabet, Meta and Microsoft – and their available cash (FCF – Free Cash Flow).
Capex of Amazon, Alphabet, Meta and Microsoft (dark blue) vs. their free cash flow (red) in billions of dollars, including forecasts for the coming years.
From 2023 onwards, these companies’ expenditure will rise steadily and, according to analysts, will be six times higher by 2030 than it was eight years ago. 2026 will be the first real test of this strategy. It is predicted that surplus cash will virtually disappear and debt will rise. In a year’s time, we will find out whether it was a brilliant gamble – or the most expensive mistake in the history of technology.
Another risk for Big Tech is competition. Although tests clearly show that Chinese models still lag far behind the American market leaders, whilst the US remains the leader where scale and computing power matter, the technological advantage is not yet a market advantage – especially when the cheaper Chinese alternative is good enough for most business applications.
