Over the last three years, hyperscale datacenter operators have massively increased their infrastructure investments, driven by the growing AI demand. Infrastructure spending has tripled, and capacity has risen by 170%, according to Synergy Research. The big players—Amazon, Microsoft, Google—among others, have been expanding at an unprecedented rate since the release of AI tools like ChatGPT.

This expansion has led to a remarkable increase in quarterly capex, nearing $142 billion, along with a 170% increase in datacenter capacity. Notably, this boom has resulted in 1,297 hyperscale facilities worldwide, nearly tripling since 2018.

Datacenter facilities have grown larger, from supporting tens of megawatts to hundreds, or even gigawatts. Meta is pushing boundaries with their multi-gigawatt clusters, solidifying the US as the dominant hub for hyperscale capacity, contributing 55% of the global operational capacity.

Despite the rapid growth, concerns linger about what might happen if the AI boom falters. Analysts worry that if AI models fail to generate expected returns, datacenter capacity could see a decline. Yet, the pipeline for future hyperscale developments stands at 770, indicating continued expansion plans.

Synergy Research predicts a further doubling of hyperscale capacity in just over three years, underscoring significant industry transformation driven by demand for AI-driven infrastructure. Experts caution, however, that a potential AI investment bubble could impact future growth.