Goldman Sachs analysts project that five major US hyperscalers—Amazon, Alphabet, Microsoft, Oracle, and Meta—will invest approximately $1.2 trillion in AI infrastructure in 2027. This estimate reflects a 50–54% increase compared to the $800 billion spent in 2026, exceeding the Wall Street consensus of $1.1 trillion.
The firm estimates that cumulative spending on AI infrastructure from 2026 through 2031 could reach $7.6 trillion. While spending grew by nearly 100% in 2026, the growth rate is expected to moderate to 54% in 2027 and 12% in 2028, with total annual expenditure projected to hit $2 trillion by 2028.
To achieve a break-even point on these capital expenditures, the five companies will require approximately $300 billion in annual AI-related revenue. Analysts warn that increased reliance on debt financing to fund these projects could create balance sheet risks if revenue growth fails to meet these targets.
Goldman Sachs identified three primary bottlenecks that could impact this investment trajectory: energy supply, labor availability, and shortages of memory chips. Data centers are currently straining power grids, while demand for high-bandwidth memory continues to outpace the production capacities of manufacturers like SK Hynix and Samsung.
