UBS Forecasts AI Capex to Near $1 Trillion in 2026, Driven by Soaring Memory Costs

Global capital expenditure (capex) on artificial intelligence is on track to approach $1 trillion in 2026 and climb to approximately $1.4 trillion in 2027, driven primarily by a dramatic spike in memory costs, according to a new forecast from UBS.

Forecast Details

UBS projects total AI capex will reach $998 billion in 2026, a near-doubling from the $506 billion estimated for 2025. The investment bank anticipates a further increase to $1.447 trillion in 2027.

The core driver behind this growth is a sharp escalation in spending on memory components. UBS estimates that memory-related capex will surge from $71 billion in 2025 to $367 billion in 2026, before reaching $923 billion in 2027. In contrast, other AI-related expenditures are expected to fall from $631 billion in 2026 to $525 billion in 2027.

A Structural Shift in AI Spending

The rising cost of memory is fundamentally altering the composition of AI investment. According to the report, increased memory costs will account for approximately 60% of the total AI capex growth in 2026. By 2027, the growth in memory spending is projected to exceed the net increase in total AI capex due to declines in other component spending.

Ad

This trend marks a significant structural shift:

• Memory's share of total AI capex is forecast to jump from 14% in 2025 to 37% in 2026, and is expected to reach 64% by 2027.
• Over the 2025-2027 period, UBS calculates that rising memory expenditure will be responsible for about 90% of the nearly $1 trillion in AI capex growth.

Macroeconomic Implications

The economic impact of this spending surge depends on whether it is driven by higher prices or increased volume, UBS noted. If rising prices are the main factor, the contribution to U.S. real gross domestic product (GDP) may be limited.

Instead, the bank suggests this expenditure will largely translate into higher revenues and profits for memory producers, which are predominantly based in Asia, providing a positive contribution to their respective economies. This highlights a deeper shift in the AI investment cycle, where growth is increasingly driven not just by infrastructure expansion but by the escalating costs of critical components for high-performance computing.