AI’s Next Constraint May Be Cost Inflation, Not Compute

MFS argues that investors are increasingly focused on who benefits from artificial intelligence, while overlooking a second-order effect: AI is creating supply bottlenecks that may raise costs across large parts of the economy.

• Following extensive meetings across the AI ecosystem, MFS concluded that demand for chips, memory, power and data-centre infrastructure is likely to outstrip supply well into 2027, potentially making current AI spending forecasts too conservative.
• The firm believes shortages in high-bandwidth memory, storage and computing infrastructure will increasingly appear as higher costs of goods sold (COGS), squeezing margins for companies that lack pricing power.
• MFS argues that the key investment distinction is no longer simply between AI winners and losers, but between businesses that can pass through rising costs and those whose returns may be eroded by AI-driven inflation and competition.

Read the full report for a deeper look at AI supply-chain bottlenecks, memory shortages and the investment implications of AI-induced cost inflation.