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

411–420 of 1001 posts

Re: Nvidia’s $589B DeepSeek rout

#411

Post by jdietrich »

jaggs wrote: post #384

The real hidden message is not that bigger compute produces better results, but that the average user probably doesn't need the top results. In the same way that medium range laptops are now 'good enough' for most people's needs, medium range (e.g. DeepSeek R1x) AI will probably be good enough for most business and user needs. Up till now everyone assumed that only giga-sized server farms could produce anything decen…

>medium range (e.g. DeepSeek R1x) AI will probably be good enough for most business and user needs

Except R1 isn't "medium range" - it's fully competitive with SOTA models at a fraction of the cost. Unless you need multimodal capability or you're desperate to wring out the last percentage point of performance, there's no good reason to use a more expensive model.

The real hidden message is that we're still barely getting started. DeepSeek have completely exploded the idea that LLM architecture has peaked and we're just in an arms race for more compute. 100 engineers found an order of magnitude's worth of low-hanging fruit. What will other companies will be able to do with a similar architectural approach? What other straightforward optimisations are just waiting to be implemented? What will R2 look like if they decide to spend $60m or $600m on a training run?

Re: Nvidia’s $589B DeepSeek rout

#412

Post by tokioyoyo »

brokencode wrote: post #376
DiscourseFan wrote: post #343

Earlier quoted context omitted.

No, because what this implies is that the Chinese have better labor power in the tech-sector than the US, considering how much more efficient this technology is. Which means that even if US companies adopt these practices, the best workers will still be in China, communicating largely in Chinese, building relationships with other Chinese-speaking people purchasing chinese speaking labor. These relationships are alrea…

What a stretch. One Chinese model makes a breakthrough in efficiency and suddenly China has all the best people in the world? What about all the people who invented LLMs and all the necessary hardware here in the US? What about all the models that leapfrog each other in the US every few months? One breakthrough implies that they had a great idea and implemented it well. It doesn’t imply anything more than that.

Chinese tech companies are also investing into AI. DeepSeek team isn't the only one (and probably the least funded one?) within mainland. This is mostly a challenge to the "American AI is yeas ahead" illusion, and a show that maybe investing only in American companies isn't the smartest method, as others might beat them in their own game.

Re: Nvidia’s $589B DeepSeek rout

#413

Post by httpz »

paxys wrote: post #341

IMO this is less about DeepSeek and more that Nvidia is essentially a bubble/meme stock that is divorced from the reality of finance and business. People/institutions who bought on nothing but hype are now panic selling. DeepSeek provided the spark, but that's all that was needed, just like how a vague rumor is enough to cause bank runs.

Nvida has a P/E of 47. While it may be a bit high for a semiconductor company, it's definitely not a meme stock figure.

Re: Nvidia’s $589B DeepSeek rout

#414

Post by jmward01 »

So, what are investors thinking to warrant this? If it is 'DeepSeek means you don't need the compute' that is definitely wrong. Making a more efficient x almost always leads to more of x being sold/used, not less. In the long term does anyone believe we won't keep needing more compute and not less?

I think the market believes that high end compute is not needed anymore so the stuff in datacenters suddenly just became 10x over-provisioned and it will take a while to fill up that capacity. Additionally, things like the mac and AMD unified memory architectures and consumer GPUs are all now suddenly able to run SOTA models locally. So a triple whammy. The competition just caught up, demand is about to drop in the short term for any datacenter compute and the market for exotic, high margin, GPUs might have just evaporated. At least that is what I think the market is thinking. I personally believe this is a short term correction since the long term demand is still there and we will keep wanting more big compute for a long time.

Re: Nvidia’s $589B DeepSeek rout

#415

Post by amazingamazing »

jmward01 wrote: post #414

So, what are investors thinking to warrant this? If it is 'DeepSeek means you don't need the compute' that is definitely wrong. Making a more efficient x almost always leads to more of x being sold/used, not less. In the long term does anyone believe we won't keep needing more compute and not less? I think the market believes that high end compute is not needed anymore so the stuff in datacenters suddenly just became…

selling more does not necessarily mean you make more money. more efficiency could lead to less margins even if volume is higher.

moreover, even things are incredibly efficient, the bar to sufficiently good AI in practice (e.g. applications), might be met with commodity compute, pretty much locking nvidia out, who generally sells high margin high performance chips to whales.

Re: Nvidia’s $589B DeepSeek rout

#416

Post by fspeech »

The key question is: has demand elasticity increased for Nvidia cards? An increase in elasticity means people are more willing to wait for hardware price to drop because they can do more with existing hardware. Elasticity could increase even if demand is still growing. Not all growths are equally profitable. Current high prices are extremely profitable for Nvidia. If elasticity is increasing future growth may not be as profitable as the projection from when Deepseek was relatively unknown.

Re: Nvidia’s $589B DeepSeek rout

#417

Post by nl »

paxys wrote: post #404
belevme wrote: post #368

Earlier quoted context omitted.

I don't think it's fair to say NVDA is meme stock, having reported 35B revenue last quarter.

Nvidia's annual revenue in 2024 was $60B. In comparison, Apple made $391B. Microsoft made $245B. Amazon made $575B. Google made $278B. And Nvidia is worth more than all of them. You'd have to go very far down the list to find a company with a comparable ratio of revenue or income to market cap as Nvidia.

Nvidia's revenue growth rate was 94% and income growth rate was 109% for the Oct 2024 quarter. This compares to Apple's 6% and -35%.

Nvidia is growing profits faster than income.

Nvidia's net profit margin is 55% (vs Apple 15%) and they have an operating income of $21B vs Apple's $29.5

These are some pretty impressive financial results - those growth rates are the reason people are bullish on it.

Re: Nvidia’s $589B DeepSeek rout

#419

Post by siwakotisaurav »

KiwiJohnno wrote: post #399
paxys wrote: post #341

IMO this is less about DeepSeek and more that Nvidia is essentially a bubble/meme stock that is divorced from the reality of finance and business. People/institutions who bought on nothing but hype are now panic selling. DeepSeek provided the spark, but that's all that was needed, just like how a vague rumor is enough to cause bank runs.

Not quite, I believe this sell off was caused by DeepSeek showing with their new model that the hardware demands of AI are not necessarily as high as everyone has assumed (as required by competing models). I've tried their 7b model, running locally on a 6gb laptop GPU. Its not fast, but the results I've had have rivaled GPT4. Its impressive.

None of the models other than the 600b one are R1. They’re just prev gen models like llama or qwen trained on r1 output making them slightly better

Re: Nvidia’s $589B DeepSeek rout

#420

Post by buffington »

danpalmer wrote: post #390
belevme wrote: post #368

Earlier quoted context omitted.

I don't think it's fair to say NVDA is meme stock, having reported 35B revenue last quarter.

I'd say it's a meme stock and based on meme revenue. Much of the 35B comes from the fact that companies believe Nvidia make the best chips, and that they have to have the best chips or they'll be out of the game. DeepSeek supposedly nullifies that last part.

Didn't DeepSeek train on Nvidia hardware though?

I can't see how DeepSeek hurts Nvidia, if Nvidia is what enables DeepSeek.

Post reply on HN