4 ms·
The other way is certainly also true. Your short piece is rational, but lacks insight into the inference and training dynamics of ML adoption unconstrained. Th
by snake_doc 2y ago
The other way is certainly also true. Your short piece is rational, but lacks insight into the inference and training dynamics of ML adoption unconstrained.
The rate of ML progress is spectacularly compute constrained today. Every step in today’s scaling program is setup to de-risked the next scale up, because the opportunity cost of compute is so high. If the opportunity cost of compute is not so high, you can skip the 1B to 8B scale ups and grid search data mixes and hyperparameters.
The market/concentration risk premium drove most of the volatility today. If it was truly value driven, then this should have happened 6 months ago when DeepSeek released V2 that had the vast majority of cost optimizations.
Cloud data center CapEx is backstopped by their growth outlook driven by the technology, not by GPU manufacturers. Dollars will shift just as quickly (like how Meta literally teared down a half built data center in 2023 to restart it to meet new designs).
- optimiz3 2y agoEveryone can say things that sound smart. When it comes to markets the only thing that matters is if your portfolio was green or red.
- Jlagreen 2y agoMy entry into Nvidia is 2016, my portfolio has never been red since then.
- TeMPOraL 2y agoSince when gambling on a RNG output makes you smart?
- timeon 2y agoIt seems to me like both of you are saying same thing.
- snake_doc 2y agoNo, the only thing that matters is if the portfolio delivers returns in excess of your cost of capital. If your portfolio is green, you can still be a poor performer.