Citigroup has revised its AI infrastructure spending forecast for tech giants, now expecting it to exceed $2.8 trillion by 2029, up from a previous estimate of $2.3 trillion. This increase is due to aggressive early investments by hyperscalers and a growing appetite from enterprises.
The AI surge, sparked by ChatGPT’s launch in late 2022, continues to drive massive capital investments and data center expansions, despite a brief setback caused by China’s more affordable DeepSeek model and ongoing concerns about U.S. tariff policies under former President Donald Trump.
According to the Wall Street brokerage, AI capital expenditures by hyperscalers are projected to hit $490 billion by the end of 2026, up from an earlier prediction of $420 billion.
Hyperscalers, such as Microsoft, Amazon, and Alphabet, have already invested billions to alleviate capacity constraints that have hindered their ability to meet the rising demand for AI.
Citi analysts anticipate that hyperscalers will discuss this increased spending during their third-quarter earnings calls, with guidance expected to surpass visible enterprise demand. Citi estimates global AI compute demand will require 55 gigawatts of new power capacity by 2030, leading to an additional $2.8 trillion in spending, including $1.4 trillion in the U.S. alone.
The brokerage notes that tech giants are no longer relying solely on profits to fund AI infrastructure projects. Given the high costs—around $50 billion per gigawatt of compute capacity—companies are turning to borrowing to keep pace.
This shift is already evident in financial reports, with spending impacting free cash flows. Investors are questioning how tech companies will finance such extensive investments, especially as traditional funding models become inadequate.
“Enterprises have clearly validated the value externally,” Citi remarked, highlighting production deployments at firms like Eli Lilly, Hitachi, and Wolters Kluwer.
