Theme

Memory & packaging

Why memory bandwidth, capacity and advanced packaging now matter alongside compute.

Updated October 2026

This theme looks at memory and packaging as parts of the AI system rather than as separate components. It collects our explainer on memory in AI systems together with reported results and guidance from memory makers. The articles show how bandwidth, capacity and packaging connect to the processors they serve.

Companies in this theme

Explore companies

Micron profile TSMC profile NVIDIA profile

Latest on this theme

All research

Markets

The global chip supply chain: how it works, where it is strained and why

A plain-language guide to the steps, the places and the bottlenecks, with the evidence and the caveats.

13 Oct 2026●22 min read●Explainer

Markets

Memory leads the forecasts: what WSTS and McKinsey expect

Two forecasters point to memory as a main driver of growth, and both tie it to new capacity.

13 Oct 2026●4 min read●Analysis

Memory & packaging

The AI system is only as useful as its memory

Why bandwidth, capacity and packaging belong alongside compute in the investment discussion.

12 Oct 2026●4 min read●Explainer

Companies

Samsung guides for ₩107.4 trillion (~US$80.1 billion) operating profit

The preliminary figure for the third quarter is a company estimate, not a final result. Sales are estimated at about ₩195 trillion (~US$145 billion).

8 Oct 2026●3 min●Reported

Companies

Micron's revenue rises to $133.2B in fiscal 2026

The memory maker's annual revenue more than tripled, and the fourth quarter alone brought in $54.23 billion. Memory has moved from being a cyclical commodity to being a bottleneck for AI.

30 Sep 2026●4 min●Reported

Related themes

All themes

AI infrastructure demand Chip market & cycle Policy, trade & geography