Hacker News Reader

Comment by bilbo0s | Hacker News Reader

I made a comment yesterday that the whole NVDA thing seemed eerily similar to the AMZN thing in the first dot com crash. [0]. If the market can get sufficiently irrational, and we start to see that same like 80-90% markdown that we got with AMZN in the first crash, I mean, who knows? That's a golden opportunity. Not many generations get opportunities like that twice.

I guess we'll see what happens with the stock price, but long term I'm pretty confident in this Jevon's Paradox effect on AI use. I'm praying for more negativity hype out among those who maybe don't know as much about tech as the HN crowd. Because those people collectively, can move stock prices.

[0] https://news.ycombinator.com/item?id=42853937

View on HN

Having read the paper from deepseek I seriously don't get how anyone can think "Oh now is the time to sell nvidia".

Both at the tactical level and strategic level that paper means we will see _more_ nvidia gpus fly off the shelves. If anything it's showing why the h800 is a decent card and everyone in China should buy one (dozen thousands).

I had analyzed the paper as well, and gave some quick thoughts on a few of the low level things that stood out and would affect NVDA here:

https://news.ycombinator.com/item?id=42856652

I have to agree. I'm not sure the architectural and engineering changes here don't put NVDA further ahead. Some stuff all gpu makers can do. Like quantize on transfer. But NVDA is so far ahead in other areas and DeepSeek leans heavily on that industry leading tech. At least at the low level.

But I'm betting the average investor doesn't have the level of facility with ML to properly analyze the information that's come out. Heck, I'm betting most of the people working in AI are really just TF or PyTorch monkeys. The real AI experts have probably been hovered up by MS or FB or AMZN or OpenAI or what have you. So it feels like we could get lucky here and the AI related market could drop like a rock taking the good with the bad. Which I think would be awesome for most people in a position to wait for it to happen.

I can't wait for the 2000 dot com crash for AI.

What gets me about the deepseek work is that it's neither novel or theoretically difficult.

I've been using a similar synthetic data pipeline to generate examples for sota bert embedding models.

That low-hanging fruits like this are left unexplored by meta or alphabet - and if we're to be believed lead to crisis meetings for a week straight - makes one wonder wtf the engineers there are doing?

This is just a baseless conspiracy theory that I have, but I do wonder if they intentionally avoided certain avenues of research because it could erode the “moat” major tech firms have by dramatically reducing the capital costs required for training and inference. If you focus your research around work that mandates high-end hardware at scale you can lock out tons of potential incumbents

There is nothing in the deepseek paper that suggests you can't use the order of magnitude in hardware costs you saved to just train models that are ten times as large.

But there are thresholds of commercial viability in all of this. DeepSeek's technology gets you over that line with less sophisticated hardware.

There's already some pretty impressive work being done with folks using just a pair of M2 Ultras with r1 in a "home lab" context that goes way beyond what you could previously do with llama.

Yes, but the r1 model is shiny and new. I've been using it locally for a day and I'm already finding odd spots. One thing that I've seen it do a number of times is get the correct answer in the thinking tokens, then ignore it, generate thousands more tokens until it reaches it context length then give a wrong answer.

There are things, like doc strings for functions, that llama models give better results for.

…and even if you can’t, you could train models that are ten times more specific, or update them ten times more frequently.

majority of Nvidia's profits are from huge companies particularly Microsoft and META not individuals buying graphics cards. Look at their current video gaming graphics sales and even if these ai consumer purchases were 10x what there previous graphics sales were then the current valuation wouldn't make sense. This is all ignoring the fact that if the h800 was good enough amd's top line cards should be too.

>This is all ignoring the fact that if the h800 was good enough amd's top line cards should be too.

You can't pay me to use AMD cards. I have work to do. I don't have time to write drivers for cards that no one supports.

The point of the deepseek r1 paper is that the more synthetic data you generate by over trained models the better the resulting model. That means more GPUs. There is no reason why you can't train a dozen models like their nameless logic model in other domains and then use them for synthetic data augmentation.

The better the foundation model, the better the distilled thinking model.

So yes, an increase need for ever better thinking models.

Perception, not facts, primarily drive markets

A supposed future event, invoked by a unconstrained interpretation of Jevons Paradox, is not a fact.

But, I thought the CUDA moat is primarily needed for training.

On the inference side, why does one need Nvidia GPUs over Intel and AMD GPUs?

Probably because companies buy Nvidia GPUs training and then use them for inference. Dual use.

Not to mention there’s little real world evidence that AMD is better at inference speed not cost/inference over Nvidia.

Tensor Cores accelerate inference...

It's not comparable. NVDA is heading toward ~$95 billion in operating income for the next four quarters approximately. AMZN in the first dotcom bubble crash had only (hefty) sales growth and was burning enormous amounts of cash to build itself out. There was no profit - and no near-term hope of a profit - to support Amazon's valuation back then. They were highly susceptible to an exaggerated crash accordingly.

At $95b in op income, you're looking at ~33 times op income for NVDA for 2025, or lower. I don't consider that a bargain but it's also not crazy expensive given the stock market valuation context + their growth + their market dominance in a critical field.

You can't apply Jevons paradox to everything, that would be predicting the future. However, you can try to analyze whether it holds in your specific context.

Comparing AMZN and NVDA is like comparing apples and oranges. Amazon has significantly diversified from its origins as an online bookseller and started when the internet, and later mobile devices, were not yet mainstream. NVDA, on the other hand, is an entirely different kind of company.

It doesn't matter whether it's AMZN or NVDA, you should try to distinguish between the company's fundamental value and the speculative component, as both are reflected in the price.

Reply on news.ycombinator.com