The global tech industry is in the midst of a $3 trillion spending spree on AI datacenters, creating a sharp divide between market optimists and financial skeptics. One side sees an industrial revolution, while the other sees a dangerous, debt-fueled bubble.
The “boom” narrative is driven by staggering numbers. Nvidia is a $5tn company, Alphabet (Google) just hit a $100bn revenue quarter, and OpenAI is valued at $500bn. Tech’s “hyperscalers” (Amazon, Meta, Google, Microsoft) are spending $750bn in two years to build the “central nervous system” for AI. In communities like Newport, Wales, a new Microsoft datacenter is hailed as the “economy of the future.”
The “bubble” narrative is equally compelling. Analysts warn of a $1.5tn funding gap being filled by “private credit”—a form of shadow banking. Gil Luria of DA Davidson calls this the “speculative” part of the boom, financing “unproven” assets and creating “structural risk to the overall global economy.”
This skepticism is fueled by data. An MIT study found 95% of businesses piloting generative AI are getting zero returns. The Uptime Institute, a datacenter rating agency, warns that many announced projects are “speculative” and “will never be built.” Alibaba’s chair has also warned of a “bubble” in projects launching without customers.
The industry is caught between the optimism of ChatGPT’s 800 million weekly users and the warnings of financial experts. The $3 trillion question is whether the “lofty revenue expectations” for AI will materialize before the debt-fueled “exuberance” backfires.