Goldman's trading desk is going all-in on AI hardware plays right now. Their thesis? Cheaper inference costs are finally making AI practical at scale, and that means China's AI buildout isn't hype — it's actually happening.
The real money isn't in the AI software narrative anymore. It's in picks and shovels. As inference gets cheaper, usage explodes. More usage = more compute demand = more hardware orders. Classic supply chain leverage play.
Goldman's positioning suggests they see Chinese demand as the next leg up for semiconductor and data center infrastructure names. Not the buzzword AI stocks everyone's chasing, but the boring hardware that actually makes it all run.
This is a rotation trade. If you believed the AI story was real, you bought $NVDA at $200. If you believe adoption is accelerating NOW because economics finally work, you're looking at who supplies the infrastructure for that adoption wave.
Watch semis, server OEMs, memory, and power infrastructure. China building out AI capacity at scale could be the 2024-2025 trade that catches people off guard.
The real money isn't in the AI software narrative anymore. It's in picks and shovels. As inference gets cheaper, usage explodes. More usage = more compute demand = more hardware orders. Classic supply chain leverage play.
Goldman's positioning suggests they see Chinese demand as the next leg up for semiconductor and data center infrastructure names. Not the buzzword AI stocks everyone's chasing, but the boring hardware that actually makes it all run.
This is a rotation trade. If you believed the AI story was real, you bought $NVDA at $200. If you believe adoption is accelerating NOW because economics finally work, you're looking at who supplies the infrastructure for that adoption wave.
Watch semis, server OEMs, memory, and power infrastructure. China building out AI capacity at scale could be the 2024-2025 trade that catches people off guard.