Combined 2026-2027 capital expenditure projections for Alphabet, Microsoft, Amazon, Meta, and Oracle have risen roughly 66%, or about $750 billion, since the start of 2026, putting the total near $1.9 trillion. Google and Amazon together are expected to spend over $1 trillion combined; Microsoft's estimate stands at $373 billion, Meta's at $333 billion, and Oracle's at $174 billion. Jensen Huang told CNBC that a 1-gigawatt NVIDIA AI factory requires $50 to $60 billion in capital expenditure but can generate about $50 billion in annual rental revenue, implying capital payback within roughly one year. Spending is directed at GPUs, custom chips, servers, memory, networking, data centers, power, and cooling.
Bain & Company projects the AI industry must generate $6 trillion in annual revenue by 2031 to fund approximately $1.5 trillion in annual capital spending, treating capex as roughly 25% of industry revenue. Known revenue sources fall far short: consumer subscriptions and ads could supply $200 billion to $400 billion, enterprise productivity gains $1 trillion to $1.4 trillion, physical AI categories such as simulations, digital twins, and robotics up to $900 billion, autonomous vehicles and industrial automation around $400 billion, and search and ad integration $100 to $200 billion. After accounting for all identified streams, Bain still sees a $4.2 trillion gap attributed to what it calls 'new products and uses that don't exist today'. That $6 trillion target is three times the bar Bain set a year ago, which was $2 trillion by 2030 with an $800 billion shortfall. Sustainably funding the buildout would require adding approximately 1% to the annual global GDP growth rate.
Sequoia's David Cahn calculated a roughly $600 billion annual gap between hyperscaler AI infrastructure spending and actual AI ecosystem revenue, identified in 2025 and continuing to widen into 2026. Goldman Sachs estimates hyperscalers may need roughly $300 billion in annual AI revenue to justify the current level of investment. McKinsey warned that while per-token AI prices are falling, total AI bills are rising as agentic applications require much more compute, meaning government and enterprise costs that were near-zero will increase substantially. Credit rating agency Egan-Jones published a note titled 'It's Over,' stating that 'the complete disruption of the economy is all but certain'.
IMF Managing Director Kristalina Georgieva stated that AI investment as a share of GDP will likely exceed the railroad buildout era, and identified AI infrastructure spending as a factor keeping global inflation above target. A source cited by Bloomberg said current AI infrastructure investment is on track to surpass railroads, electrification, and highways as the largest infrastructure buildout in US history.