FT analysis questions whether projected $9 trillion in AI infrastructure spending will generate returns, raising overbuilding fears.
The Financial Times published a major analysis questioning whether the unprecedented wave of AI data centre investment — estimated at up to $9 trillion globally — is economically justified. The piece examines whether AI revenue growth can keep pace with infrastructure commitments already made by hyperscalers and sovereign investors. It highlights risks of overcapacity, stranded assets, and demand projections that may be based on speculative AI adoption curves rather than proven enterprise spend.
If the data centre overbuild thesis is correct, GPU and cloud compute prices will stay suppressed or fall further as hyperscalers fight for utilisation. That's structurally good for developers: cheaper inference, more aggressive spot pricing, and cloud providers desperate to lock in workloads with credits and discounts. The risk scenario — where credit dries up and build-outs halt — is not imminent and doesn't affect current API availability.
Check your current cloud provider's committed-use discount tiers this week — if overbuild continues, negotiating a 1-year reserved instance or committed spend agreement now locks in low rates before any pricing floor correction.
Go to calculator.aws and configure your current primary compute instance type and region
Tags