$GOOGL - Google's $80B equity raise with Berkshire's $10B investment signals a strategic shift to lower-margin but higher-volume infrastructure business. The TPU advantage and superior cost structure could allow Google to dominate if compute becomes commoditized, with absolute returns potentially exceeding their high-margin search business.
$POWER - Power infrastructure is being built out massively as a direct result of AI spending, representing a long-term asset that will outlast any bubble similar to dark fiber from the dot-com era.
$AIMODELS - Foundation models themselves represent a lasting asset from the AI bubble, having cost tens of billions to build and will be usable by productized AI companies for decades, similar to fiber optic infrastructure after the dot-com crash.
$SPACEX - SpaceX benefits from AI bubble dynamics through their data centers in space thesis, which enables higher valuations and more rocket development regardless of medium-term viability.
$INTC - Intel's foundry business will benefit from TSMC's underinvestment in 2023-2024, as capital flows to alternative foundries to meet AI infrastructure demand.
Bearish:
$GOOGL - Google's equity raise could signal management believes the stock is overvalued and they're using it as cheap currency, or that they're uncertain about AI ROI and want to share downside risk with new shareholders.
$MEMORY - Memory companies face structural headwinds as AI bubble forces algorithmic improvements that reduce memory usage, while also opening markets to Chinese competition. The industry has never faced pressure to optimize memory efficiency before.
$OLDTECH - Existing companies trying to bolt AI onto old workflows will fail to see returns and slowly die, replaced by new AI-native companies that rebuild processes from the ground up. The bubble will burst when it becomes a widely accepted meme that AI doesn't work for legacy companies.