GPU makers really lucked out in the last few years. First from the crypto craze and now the AI craze.
Not to mention the whole "video game" thing that people really like
Nvidia’s Blowout Forecast Sparks $260B AI Rally
31–34 of 34 posts
Re: Nvidia’s Blowout Forecast Sparks $260B AI Rally
#32I just can't get my head around this. Sure, Nvidia is in a good position with AI and Data Centre. I absolutely think they're in a great position and they're going to sell a lot of chips. But they're trading at 200 times earnings? I don't see how anyone could reasonably justify that. And not only that, but while yes, the earnings are good, surely the AI hype cycle has pulled forward a decent amount of demand, it's a r…
I think there is a heavy quantitative induction mechanism in the markets, that is, a positive feedback mechanism from index funds trying to minimize divergence, or hedge funds trying to font-run those algorithms. So it doesn’t take much to set off irrationality, and not much exists to pull it back, because income investing basically doesn’t exist, and therefore value investing has little support.
Partial side note: obviously the FOMC has a lot to do with investment flows, but I’ve found that the tax system is just as relevant or more so in market distortions, but it is poorly understood, rarely observed, and never considered for alternative possibility, like the water that our economy swims in. For instance, the combination of 401k and double taxation killed income investing. Also, capital gains tax and long-term preference deters sale in a pump, and actually works the other way to prefer short-term selling of losses, so the net effect is to add to the positive feedback. Every tax situation is a comparison of speculative gain and loss on an efficient market hypothesis prior, versus a guaranteed tax loss.
Re: Nvidia’s Blowout Forecast Sparks $260B AI Rally
#33I just can't get my head around this. Sure, Nvidia is in a good position with AI and Data Centre. I absolutely think they're in a great position and they're going to sell a lot of chips. But they're trading at 200 times earnings? I don't see how anyone could reasonably justify that. And not only that, but while yes, the earnings are good, surely the AI hype cycle has pulled forward a decent amount of demand, it's a r…
Consumer ( or Crypto ) Business dropped by around 20%. While Datacenter is 15% increase. Even if both Consumer and DC were to double their Revenue, this will still be 100x P/E. And you can safely bet consumer wont double, and leaves those headroom for DC, i.e DC segment to grow by 2.5x. It will still be 100 PE.
Could they do that? Possibly. But I still think it is too much hype.
Re: Nvidia’s Blowout Forecast Sparks $260B AI Rally
#34Earlier quoted context omitted.
You can argue that this PE, at this market cap implies, it will eventually have the world's largest market cap (larger than apple...maybe 2X). Would think OpenAI and others would tire of buying chips from NVIDIA at some point well before that and make their own (like Google does).
Yes, this is exactly the point - some people are making arguments about Nvidia's amazing software stack. Sorry, but Amazon and Google have some pretty fucking good software engineers, Google already tapes out TPUs. Oh, and those are Nvidia's biggest potential customers. Honestly this stock market move makes it look like Google should just announce "hey we're putting the TPU on sale for general availability" they'd ju…
TPUs are available on GCP, but working with them has been frustrating.
My experience has been that it takes not insignificant effort to convert training scripts and then some weird unexpected bug takes a while to figure out, I’ve heard similar things from peers in academia.
Additionally, at least in early 2023, PyTorch had a substantial throughput reduction on TPUs so you’d probably need to use Jax (or god forbid TF) for efficiency sake.
Granted, I’ve heard some of the PyTorch XLA issues have since been improved.
Regardless, the H100 currently significantly outperforms TPUv4 in throughout on transformer loads, we’ll see what TPUv5 looks like to be fair but it’s not a given that Google/Amazon can outpace Nvidia in manufacturing and chip design when this is their core product and they also have amazing engineers + a large open source community building around CUDA.