Jefferies: Cheaper AI models to fuel infrastructure demand, favoring memory chip makers

Brokerage argues that cost-efficient AI models will expand, not contract, capital expenditure on computing hardware, creating a tailwind for memory suppliers.

By Handelsavisen Newsroom · Automated 2026-06-26 07:24 UTC · 1 min read

The proliferation of low-cost artificial intelligence models is poised to accelerate, rather than dampen, demand for computing infrastructure, according to a new report from Jefferies.

The brokerage argues that as AI becomes more accessible and efficient, the need for underlying hardware—particularly memory chips—will intensify as enterprises scale their deployments.

Jefferies explicitly favors memory chip makers in this scenario, viewing them as primary beneficiaries of the expanding infrastructure build-out.

This perspective challenges the narrative that cheaper AI models might reduce the overall addressable market for high-performance computing components.

Instead, the report suggests a virtuous cycle where lower inference costs drive broader adoption, which in turn requires more robust and scalable memory solutions.

This analysis arrives as the semiconductor sector navigates shifting sentiment around AI capital expenditure.