UBS Lifts AI Spending Forecast to $1.4 Trillion by 2027, Cites Soaring Memory Costs

Investment in artificial intelligence infrastructure is set to nearly double this year and climb significantly higher in 2027, driven by a dramatic surge in memory costs, according to a new forecast from UBS. The bank now projects total AI capital expenditure (capex) will reach $998 billion in 2026 and grow to $1.447 trillion by 2027.

Forecast at a Glance

UBS's updated estimates represent a substantial increase from the $506 billion in AI capex recorded in 2025. The bank's projections show a rapid escalation in spending over the next two years.

• 2025: $506 billion
• 2026 Forecast: $998 billion
• 2027 Forecast: $1.447 trillion

The primary driver for this upward revision is the rising price of memory components, which are essential for powering advanced AI systems.

Memory Costs Reshape AI Spending

Ad

The analysis from UBS indicates that spending on memory is expected to skyrocket from $71 billion in 2025 to $367 billion in 2026 and an astonishing $923 billion in 2027. In contrast, spending on other AI-related components is forecast to decline from $631 billion in 2026 to $525 billion the following year.

This dynamic means higher memory costs will account for approximately 60% of the increase in AI capex this year and will represent more than the entire net increase in 2027. UBS calculates that roughly 90% of the nearly $1 trillion capex increase between 2025 and 2027 will stem from memory spending alone. Consequently, memory's share of total AI capex is projected to expand from just 14% in 2025 to 37% in 2026 and a majority 64% in 2027.

Economic and Market Implications

This shift carries significant macroeconomic consequences. UBS noted that if the spending increase is driven by higher prices rather than greater volumes, it would contribute little to real U.S. gross domestic product (GDP).

Instead, the bank suggests this trend would represent a massive transfer of income and profits toward memory producers, which are predominantly located in Asia. This would, in turn, provide a positive contribution to GDP in those economies. The forecast highlights a fundamental change in the AI investment cycle, shifting focus from the sheer scale of infrastructure to the rapidly rising cost of critical components.