UBS raised its forecast for artificial intelligence capital expenditure to nearly $1 trillion in 2026, up from a previous estimate, and projects spending will climb to approximately $1.45 trillion by 2027, according to research cited by Investing.com.
Total AI capex rose to $506 billion in 2025 and is now expected to nearly double to $998 billion this year before reaching $1.447 trillion next year.
Memory costs are the primary driver of the increase. UBS estimates memory spending will surge from $71 billion in 2025 to $367 billion in 2026 and $923 billion in 2027, accounting for roughly 90% of the nearly $1 trillion rise in total AI capex between 2025 and 2027. Memory's share of total AI spending is projected to grow from about 14% in 2025 to 37% in 2026 and 64% in 2027.
Other AI-related costs, totaling $631 billion in 2026, are forecast to decline to $525 billion in 2027, but the drop is more than offset by the memory spending boom. Higher memory costs account for about 60% of the 2026 capex increase and more than the entire net increase in 2027.
UBS cautioned that the macroeconomic impact depends on whether the spending reflects higher prices or higher volumes. Price-driven increases would add relatively little to real U.S. GDP and would instead function as a transfer of income and profits toward memory producers in Asia, positively contributing to GDP in those economies.
The estimates mark a shift in the economics of the AI investment cycle — moving away from solely the scale of infrastructure spending toward the escalating costs of components needed to support increasingly powerful computing systems.
