Dario Zanette (@dariozanette.mastodon.uno.ap.brid.gy)

The impossible maths of the AI boom Non è un’assunzione eroica, è un’assunzione priva di senso pratico. > … > > One aspect of today’s boom is already much larger than the TMT bubble ever was. In 2025, US businesses invested almost $1.5tn in IT equipment and software. At the peak of the TMT bubble, it was $466bn or $829bn when adjusted for inflation. > … > The developers of large language models such as OpenAI and Anthropic are preparing for blockbuster initial public offerings later this year to benefit from investor optimism about their growth. Meanwhile, the hyperscalers Microsoft, Alphabet, Amazon, Meta and Oracle plan to invest hundreds of billions in the next five years in data centres to provide the computing power to run these models. > And this is where the maths of the AI boom becomes challenging. For each of these hyperscalers, I collected the consensus estimates of analysts for the capital expenditures and revenues between 2025 and 2030. > In these five years, capital investments are expected to rise by 20 per cent a year, a growth rate never seen before in this industry. Meanwhile, revenues are expected to grow 15 per cent annually. If we make the heroic assumption that there are no costs, then the additional revenue is the profit these companies are expected to make from their additional investments in AI data centres. Yet, even under these extremely optimistic assumptions, I calculate the implied return on investment is highly negative for all of them except Amazon. > These numbers show that if the hyperscalers continue on the current trajectory, the AI boom will become a story of one of the largest destructions of shareholder value in history. But there are two ways out of this corner. > One is that the rise of AI will generate much more revenue for these companies than currently expected. Unfortunately, that comes with its own mathematical challenges. If the hyperscalers want to generate, say, a 10 per cent return on investment, they would have to find an additional $2tn to $5tn in revenue a year. A tall order for a group of companies that currently generates revenues of just $1.5tn per year. > The other option is that the planned investment in data centres, computer chips and other areas never materialises — maybe as equity investors turn more cautious on the sector or if debt funding for data centres becomes harder to get. > … > > Viewed through this lens, the IPO of these AI companies is probably nothing more than a major transfer of investment risk from the current owners to retail investors, pension funds and others who are willing to buy the hype. Source: _The impossible maths of the AI boom_ If you like this post, please consider sharing it. https://blog.quintarelli.it/2026/05/the-impossible-maths-of-the-ai-boom/