Why are $MU (Micron) and $SNDK (SanDisk) struggling despite strong earnings and guidance?
Let's look back at the AI investment cycle.
First, GPUs were in short supply, and $NVDA soared. Then HBM and memory capacity became the bottleneck, fueling a massive rally in memory stocks.
But now, the bottleneck may be shifting from technology and production capacity to CAPITAL.
Google Cloud revenue surged 103% YoY, yet Alphabet expects to spend $195–205 billion on CapEx in 2026, while Q3 free cash flow turned negative at -$9.1 billion.
The market used to ask: "Do we have enough chips?"
Now it's asking: "Do we have enough money?"
I believe the next phase of the AI investment cycle comes down to two critical questions:
1. Where will the money come from?
As hyperscalers' free cash flow becomes insufficient to cover massive capital expenditures, can bond markets, banks, and institutional investors continue financing AI infrastructure? And at what cost?
2. How will the money ultimately be earned back?
Can AI companies like Anthropic and OpenAI generate sustainable profits through token usage, APIs, and enterprise services? Will that revenue eventually justify the billions spent on GPUs, memory, and data centers?
If AI financing and monetization can form a sustainable economic cycle, today's depressed AI hardware valuations may finally have a solid foundation for recovery.