Big Tech's AI Infrastructure Spending Nears $200 Billion a Quarter
AI infrastructure spending is projected to total about $10 trillion and average roughly 3.6% of GDP annually through the early next decade. Amazon, Alphabet, Microsoft, Meta, and Oracle are spending tens of billions quarterly on data centers, with hyperscalers reinvesting about four-fifths of cash flow. The article notes pressure at Meta, where free cash flow fell sharply, and Oracle, whose shares have dropped 52% over the past year amid financing stress.
The projected $10 trillion AI buildout would average about 3.6% of GDP yearly into the early 2030s. Private data-center construction through July already exceeded all other private construction combined. Amazon, Alphabet, Microsoft, Meta, and Oracle are each committing tens of billions quarterly.
Much demand is contracted: Microsoft’s remaining performance obligations climbed 84% to $678B, Oracle’s stood at $664B, and AWS’s backlog hit $496B. Yet financing strains are visible: Meta’s quarterly free cash flow dropped to $784M from $8.55B, while Oracle’s shares fell 52% over the past year.
The buildout could reshape job markets, energy demand, and local infrastructure around data centers, affecting host communities and workers with AI-related skills. Investors, including retirement savers, may be exposed through equities and credit if returns lag spending. Consumers could gain from new services, but may also face cost shifts as companies seek to recover capital. Whether these effects prove broadly beneficial remains uncertain and depends on whether projected demand materializes.