Analysis
The AI industry needs to generate roughly $6 trillion in annual revenue by 2031 just to pay for the data centers being built today, according to Bain & Company's Global Technology Report, published Tuesday. That's the size of the bill; existing AI products, Bain estimates, could bring in only $1.2 trillion to $1.8 trillion of it.
The Gap
Annual spending on AI infrastructure could reach $1.5 trillion by 2031 on its own, per Bain, meaning the roughly $4.2 trillion revenue shortfall -- the gap between what's being spent and what current AI products can plausibly earn back -- would have to come from entirely new categories of products, services and business models that don't exist yet. Bain's David Crawford, chairman of the firm's global technology practice, frames it as a GDP-level problem: funding the buildout sustainably would require adding roughly 1 percentage point to annual global GDP growth.
Who's Actually Spending
Microsoft, Google, Amazon, Meta and Oracle could collectively spend as much as $780 billion on capital expenditure in 2026 alone -- nearly five times their combined level just three years earlier, according to the same report. That capex is already tracked on our funding tracker and cuts against every AI infrastructure financing story this week, from GMI Cloud's $668 million raise to the widening yields on Meta-tied data-center debt: all of it is a bet that the revenue gap Bain is describing closes on schedule.