The core driver of the global artificial intelligence (AI) investment cycle is undergoing a fundamental transformation. UBS’s latest research shows that the sharp surge in memory prices has replaced infrastructure scale expansion as the primary force propelling AI capital expenditure growth, with global AI capital expenditure in 2026 projected to approach the $1 trillion mark.

According to UBS’s latest estimates, global AI capital expenditure will total $998 billion in 2026, nearly doubling from $506 billion in 2025. By 2027, that figure will climb further to $1.447 trillion. Notably, this significant upward revision does not stem from a synchronized expansion in overall investment scale, but rather is the direct result of rapidly rising memory costs.

Memory Moves from Supporting Role to Center Stage

Memory is rapidly becoming the single largest source of AI capital expenditure growth. UBS estimates that global memory spending will leap from $71 billion in 2025 to $367 billion in 2026, and expand further to $923 billion by 2027.

In stark contrast, non-memory AI capital expenditure is trending in the opposite direction. UBS projects that such spending will be approximately $631 billion in 2026, but will decline to $525 billion by 2027.

From a structural perspective, memory’s share of AI capital expenditure is undergoing a fundamental reshaping. In 2025, memory spending accounted for approximately 14% of total AI capital expenditure; UBS expects that ratio to rise to 37% in 2026 and leap further to 64% by 2027.

YearTotal AI CapexMemory SpendingNon-Memory SpendingMemory Share2025$506 billion$71 billion$435 billion14%2026$998 billion$367 billion$631 billion37%2027$1.447 trillion$923 billion$525 billion64%

Note: The 2025 non-memory spending figure is derived by subtracting memory spending from total AI capital expenditure; the UBS report does not separately list a non-memory spending forecast for that year.

Structural Breakdown of Incremental Sources

UBS’s calculations reveal the underlying logic of this AI investment expansion. In the year-over-year increase in 2026 AI capital expenditure, rising memory costs contribute approximately 60%; by 2027, the incremental increase in memory spending actually exceeds the overall net increase in AI capital expenditure, as spending on other components declines during the same period.

Taken together, UBS estimates that of the nearly $1 trillion increase in AI capital expenditure between 2025 and 2027, approximately 90% will come from rising memory expenditure.

This data indicates that the economic drivers of the AI investment cycle are undergoing a profound shift—from simple infrastructure scale expansion toward rising costs of the critical components needed to support increasingly powerful computing systems.

Macroeconomic Impact: Capital Flows Toward Asia

UBS specifically highlights that the nature of memory spending—whether it is price-driven or volume-driven—has fundamentally different macroeconomic implications.

If the incremental spending primarily reflects memory price increases, the contribution to U.S. real GDP will be quite limited. In this scenario, capital flows will largely manifest as a transfer of profits from AI infrastructure investors to memory producers. Given that the world’s major memory producers are concentrated in Asia, those economies will receive more direct positive GDP contributions.

This assessment carries important implications for investor asset allocation. As the AI investment cycle deepens, the Asian memory supply chain may benefit to a greater degree than the market previously anticipated. Market participants note that the sustained upward trend in memory prices could reshape the profit distribution landscape across the AI supply chain, requiring investors to reassess the valuation positioning of related stocks.

From a longer-term perspective, the economic logic of AI investment is no longer a simple linear relationship of “more input, more computing power.” When memory costs become the dominant variable, economies with memory production capacity and technological advantages will capture greater economic benefits from this AI wave. This also means that the global landscape of AI investment is gradually shifting from a U.S.-centric structure toward a dual-track structure with balanced emphasis on both the United States and Asia.