AI investment cycle enters new phase as memory chip demand soars

The AI investment cycle is moving from processors to the memory and infrastructure bottlenecks that support increasingly demanding models, creating a broader set of beneficiaries than Nvidia.

• Micron’s revenue rose from around US$9 billion (NZ$15.9b) a year earlier to nearly US$42b, sending its share price up about 15% as management cited exceptional AI infrastructure demand and limited near-term supply growth.
• Nvidia has spent much of the year consolidating, while Micron and SK Hynix have both gained more than 300%; Micron, SK Hynix and Samsung now dominate advanced high-bandwidth memory, whose shortage has been dubbed “RAM-ageddon”.
• Memory-cost inflation is reaching end markets: Apple and Microsoft raised prices, while Dell, HP, Lenovo and Asus raised prices or reduced memory specifications, indicating that AI demand is affecting consumers and businesses.
• The article compares this shift with earlier technology cycles in which suppliers such as Levi Strauss, fibre-optics companies and networking businesses captured value alongside the headline winners.
• The author argues that current AI valuations differ from the late-1990s bubble: Nvidia trades at about 20 times forward earnings with earnings growth above 70%, while Micron and SK Hynix trade below nine times expected earnings despite valuations above US$1 trillion each.

Takeaway: Investors should look beyond Nvidia toward emerging AI bottlenecks, especially memory, but remain cautious because the memory industry is cyclical, competition will intensify and today’s shortage could become tomorrow’s oversupply.