Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Nvidia’s $589B DeepSeek rout

finance.yahoo.com

Post reply on HN
Print view View original

281–290 of 1001 posts

Re: Nvidia’s $589B DeepSeek rout

#281

Post by maxglute »

hintymad wrote: post #276
bee_rider wrote: post #263

ASML plunge indicates a hysterical/irrational component to the response, right? They aren’t going anywhere. If it turns out training is easier than expected, they make the devices that make the devices that do inference too… If the field is going to produce anything useful, cheap training gets us there faster.

I'm guessing the sentiment is not about training, but about the R&D capability of China. If Chinese can figure out how to build a good enough model faster and cheaper, they may be able to come up with an ASML competitor as well.

PRC just announced mass producing 28nm litho that cost 1/30 ASML hardware. Easy to extrapolate where this goes especially mature nodes like 28nm still accounts fo over 70% of global wafer use.

Re: Nvidia’s $589B DeepSeek rout

#282

Post by hintymad »

Is it possible that the key here is all about where top talent goes? America's best minds from Math and Informatics Olympiads gravitate to Wall Street for financial gains. Similarly, DeepSeek's top talent are mostly IMO and IOI medalists and they originally aimed for finance too (The parent company of DeepSeek is a successful quant trader), but strict government regulations pushed them towards AI instead. Ironically, this unintended pivot turned out to be quite significant

Re: Nvidia’s $589B DeepSeek rout

#284

Post by hintymad »

maxglute wrote: post #281
hintymad wrote: post #276

Earlier quoted context omitted.

I'm guessing the sentiment is not about training, but about the R&D capability of China. If Chinese can figure out how to build a good enough model faster and cheaper, they may be able to come up with an ASML competitor as well.

PRC just announced mass producing 28nm litho that cost 1/30 ASML hardware. Easy to extrapolate where this goes especially mature nodes like 28nm still accounts fo over 70% of global wafer use.

I'm increasingly believing that the West has turned their dream of free trade for comparative advantage into a massive deindustrialization. The end result is unfolding in front of everyone and the sentiment I see, even on HN, is we can't outcompete China any more. This is sad. Really sad. And this fits exactly what Liu Cixin said in Three Body Problem: Weakness and ignorance are not barriers to survival, but arrogance is.

We thought we won, and we thought we could "control" what other markets do, and we thought we could focus on only the "high-value add". Now where are we?

Re: Nvidia’s $589B DeepSeek rout

#285

Post by michaelmrose »

polygamous_bat wrote: post #225
junon wrote: post #211

Earlier quoted context omitted.

o1 also does this.

o1 does not show the reasoning trace at this point. You may be confusing the final answer for the reasoning trace in the middle, it's shown pretty clearly on r1.

Is this different data or different annotation

Re: Nvidia’s $589B DeepSeek rout

#286

Post by hn_throwaway_99 »

bee_rider wrote: post #263

ASML plunge indicates a hysterical/irrational component to the response, right? They aren’t going anywhere. If it turns out training is easier than expected, they make the devices that make the devices that do inference too… If the field is going to produce anything useful, cheap training gets us there faster.

> ASML plunge indicates a hysterical/irrational component to the response, right

But don't forget about the hysterical/irrational component that also causes prices to go up when investors are all worried about FOMO. Of course, sure, ASML isn't going anywhere, but their stock price isn't based on them "sticking around", it's based on the idea that growing usage of AI will require exponentially more computing power over time, and DeepSeek kinda put a pin to that balloon.

Re: Nvidia’s $589B DeepSeek rout

#287

Post by maxglute »

hintymad wrote: post #284
maxglute wrote: post #281

Earlier quoted context omitted.

PRC just announced mass producing 28nm litho that cost 1/30 ASML hardware. Easy to extrapolate where this goes especially mature nodes like 28nm still accounts fo over 70% of global wafer use.

I'm increasingly believing that the West has turned their dream of free trade for comparative advantage into a massive deindustrialization. The end result is unfolding in front of everyone and the sentiment I see, even on HN, is we can't outcompete China any more. This is sad. Really sad. And this fits exactly what Liu Cixin said in Three Body Problem: Weakness and ignorance are not barriers to survival, but arroganc…

IMO not so much as "where are we" as "where are they", west was always going to have reckon with competing with PRC who is on trend to add more STEM than OECD combined or US adds new people. And eventually this will apply to India as well. Arrogance doesn't help, but at some point reality of high value regressing to mean because magnitude more smart brains is involuting the margins out of everything. PRC competitors likely also getting creamed by deepseek as well. The running joke in China is when China does something advanced, that thing is no longer considered advanced since China very good at driving costs to nothing and commodizing advanced into common which ironically hurts PRC from getting into the true high value game.

Re: Nvidia’s $589B DeepSeek rout

#289

Post by mike_hearn »

grajaganDev wrote: post #230
alecco wrote: post #186

Earlier quoted context omitted.

The limit is high quality data, not compute.

Right and LLMs will not be able to generate their own high quality training data. There are no perpetual motion machines.

They already do. All the current leading edge models are heavily trained on synthetic data. It's called textbook learning.

Re: Nvidia’s $589B DeepSeek rout

#290

Post by igleria »

throwup238 wrote: post #266
igleria wrote: post #147

Earlier quoted context omitted.

this same forum ignored deepseek one month ago, save for a few... open minded people.

Deepseek: The quiet giant leading China’s AI race (27 days ago, 465 comments): https://news.ycombinator.com/item?id=42557586 Are we talking about the same forum? HN commenters have been raving about DeepSeek v3 for at least a month.

https://news.ycombinator.com/item?id=42514633 the current base version release thread and a mere 40 comments. C'mon, it's not a sin to be biased.
Post reply on HN