Wall Street has treated artificial intelligence as the most bullish trade for two years, a growth engine that turbocharges earnings and promises a productivity windfall. But the Federal Reserve sees the same numbers as a fresh source of demand in an economy still fighting to drag inflation back to 2%.
The Signal
Goldman Sachs expects AI-related capital spending to approach $800 billion in 2026, lifting its full-year business investment forecast to 7.8% and adding roughly 3.3 percentage points to capital-expenditure growth. TrendForce, tracking the nine largest cloud providers, places their combined 2026 outlay near $830 billion, a jump of about 79% over the previous year. A significant slice of that increase reflects rising prices rather than added capacity: Microsoft attributes some $25 billion of its $190 billion budget to costlier memory and components. This trend is not isolated; Alphabet, Amazon, and Meta have all raised their spending guidance, citing shortages of advanced chips and the need for infrastructure to train large language models. The scale of these numbers dwarfs previous tech investment cycles, rivaling only the railroad boom of the 19th century or the internet buildout of the 1990s.
Where does the $800 billion actually go? It takes the shape of land, steel, transformers, copper wiring, gigawatts of fresh generation capacity, industrial-scale cooling, and incredibly skilled trades. Goldman describes a wave across servers, semiconductors, memory, power infrastructure, data centers, software, and research budgets, with a longer-range model tracing annual AI capex from around this year toward by 2031. The global supply chain is already showing bottlenecks: lead times for high-voltage transformers have doubled, and copper prices are up 15% year-over-year. Local governments in regions like Northern Virginia and Singapore have imposed moratoriums on new data centers due to grid strain. This is not just corporate spending; it is a reshaping of the global economy that the Fed cannot ignore.
