UBS raises AI capital expenditure forecast: approaching $1 trillion by 2026, with 90% of the increase coming from memory price hikes.

UBS has significantly raised its forecast for AI capital expenditures, with the sharp rise in memory costs being the core driver of this round of spending expansion. Its impact is profoundly changing the structure and economic logic of the AI investment cycle.

According to UBS's latest estimates, global AI capital expenditure will reach $998 billion in 2026, nearly double the $506 billion in 2025. This figure is projected to rise further to $1.447 trillion in 2027. This significant upward revision is primarily due to the rapid increase in memory prices, rather than an overall expansion in infrastructure investment.

The explosive growth in memory spending signifies a structural shift in the landscape of AI investment beneficiaries. UBS points out that if the spending growth is primarily driven by price increases, its impact on US real GDP will be relatively limited, with revenue and profits shifting more towards Asian memory manufacturers, thus contributing positively to the GDP of these economies.

Memory expenditure: From supporting role to leading role

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

Meanwhile, other AI-related spending is trending in a completely different direction. UBS projects that non-memory AI capital expenditures will reach approximately $631 billion in 2026, but will fall back to $525 billion by 2027.

In terms of structural proportions, memory spending is undergoing a fundamental reshaping in AI capital expenditures. In 2025, memory spending accounted for approximately 14% of total AI capital expenditures; UBS projects this proportion to rise to 37% in 2026 and further jump to 64% in 2027.

Breakdown of sources of incremental expenditure

UBS's calculations reveal the underlying logic behind this round of AI investment expansion. In 2026, rising memory costs contributed approximately 60% to the year-on-year increase in AI capital expenditure; however, in 2027, the increase in memory expenditure even exceeded the overall net increase in AI capital expenditure, as spending on other components declined during the same period.

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

This data indicates that the economic drivers of the AI investment cycle are undergoing a profound shift—from simply expanding the scale of infrastructure to gradually moving towards the rising costs of key components needed to support increasingly powerful computing systems.

Macroeconomic impact: Beneficial to Asian memory manufacturers

UBS specifically points out that the nature of memory spending—whether it is price-driven or quantity-driven—has a fundamentally different impact on the macroeconomy.

If the increased spending primarily reflects rising memory prices, its contribution to US real GDP will be quite limited. In this scenario, capital flows will be more reflected in the transfer of profits from AI infrastructure investments to memory manufacturers. Since major global memory manufacturers are concentrated in Asia, the relevant economies will thus receive a more direct positive contribution to GDP.

This assessment has significant implications for investors' asset allocation: as the AI investment cycle deepens, the Asian memory industry chain may benefit more than previously expected by the market.

Risk warning and disclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.