Will the AI Data Centre Boom Become a Nine Trillion Dollar Bust
Financial Times analysis questions the sustainability of massive infrastructure spending on AI
The global race to build AI infrastructure has seen tech giants, sovereign wealth funds, and private equity firms pour unprecedented sums into data centre construction. But the FT analysis questions whether demand for AI compute will ever match the supply being built.
The piece draws parallels with previous technology investment bubbles, noting that current spending projections assume sustained exponential growth in AI workloads that may not materialise. Power consumption concerns, water usage, and regulatory pushback could further constrain the buildout.
Several major projects have already faced delays or cost overruns, and some investors are beginning to quietly reduce their exposure to pure-play data centre investments.
Analysis
Why This Matters
Nine trillion dollars is not a rounding error. If even a fraction of planned data centre capacity goes unused, the financial fallout could ripple through capital markets, affecting pension funds, sovereign wealth funds, and the broader tech sector.
Background
The AI infrastructure boom accelerated in 2024-2025 as companies raced to secure GPU capacity. Hyperscalers like Microsoft, Google, and Amazon announced hundreds of billions in capex, while new entrants from the Gulf states and Asia added to the frenzy.
Key Perspectives
Bulls argue that AI demand is being underestimated and that data centres are the new critical infrastructure. Bears point to utilisation rates that are already softening and question whether enterprise AI adoption will scale fast enough.
What to Watch
Utilisation rates at newly commissioned facilities, and whether any major planned projects get shelved or scaled back in the coming quarters.