Big Tech To Spend $1.2 Trillion On AI Infrastructure In 2027

Goldman Sachs has run the numbers. Amazon, Alphabet, Microsoft, Oracle and Meta will spend $1.2 trillion on AI infrastructure in 2027. That is more than 50% above this year's forecast of roughly $800 billion. It is also above the Wall Street consensus of $1.1 trillion. Ryan Hammond, an analyst at the bank, compares this investment cycle to the railroad building boom of the 19th century.

I would not read this jump in capex as a sign that model subscriptions will get cheaper tomorrow. It is a bet by the industry that demand for agents will keep growing for years. Goldman itself notes that big tech needs about $300 billion in annual AI revenue to make these expenditures pay off. That kind of money simply is not there yet. At the same time, the pace of spending growth is already slowing. It goes from nearly 100% in 2026 to 54% in 2027 and then 12% in 2028, by the bank's own math.