On September 20, UBS forecasted that artificial intelligence capital expenditures will reach nearly $1 trillion in 2026 and increase to approximately $1.4 trillion by 2027, with a significant rise in memory costs being the primary driver of this growth. UBS estimates that memory-related expenditures will surge from $71 billion in 2025 to $367 billion in 2026, and reach $923 billion in 2027. Other AI-related costs are expected to be $631 billion in 2026, decreasing to $525 billion in 2027. This implies that the rise in memory costs will account for approximately 60% of the incremental growth in AI capital expenditures this year, and by 2027, the increase in memory costs will even surpass the total net increase in AI capital expenditures.
