- GPUs are still needed for training and inference, and I guess people will figure that out in the next few days
- So far, all we have is claims from one source. Seems credible enough, and I'm sure there's massive optimization potential in training LLMs, but someone else should replicate the process first before it's taken as a fact
Everything about DS success screams the future of AI is bright. The advance puts more use cases into feasible space. And then number one supplier for it drops. DS is literally using NVIDIA chips for god‘s sake
Markets aren’t rational. Or people don’t understand AI. Or both. Idk
Pet-peeve: I hate titles like this, “biggest loss in history” is meaningless unless it’s inflation-adjusted, and it almost never is. Is this a bigger loss in real terms? Or is this just a big number? I’m guessing it’s the second.
Edit: I read the article, it’s definitely the second case. Is this _also_ the biggest loss in real terms? I don’t know.
> For Nvidia, the loss was more than double the $279 billion drop the company saw in September, which was the biggest one-day market value loss in history at the time, unseating Meta’s $232 billion loss in 2022. Before that, the steepest drop was $182 billion by Apple in 2020.
Not compared to these recent losses, but compared to all history it’s very possible, the value of the dollar drops by orders of magnitude as you go back in time.
Counterpoint - generally larger companies should be less susceptible to the type of volatility that leads to the title, so I still think it's newsworthy. Even if you were to change the title to largest percentage single-day loss amount among 3 trillion+ companies it would still be true (might even be true amount 2+ trillion as well)
Inflation adjusted is still limited to the largest companies. It’s simply using 2025 dollars for companies rather than comparing 1950 or whatever dollars vs 2025 dollars.
but inflation adjusted doesn't make any sense. things are less volatile now, there's high speed trading, better knowledge, easier trades, more automation, regulation, etc. it wouldn't make any sense.
even now compared to 2020 there has been a huge change in amount of retail investors.
Things aren’t less volatile in terms of extreme events. Look in depth at say the 2010 flash crash one of the all time great examples of very short term volatility.
Similarly, longer term we still see huge shifts such as the COVID dip before all that money flooded the markets.
I don’t have an intraday chart to look at, but DJIA which is just 30 large companies lost ~1.5 trillion in 2025 dollars in that flash crash.
So on average each company lost ~50 billion over a few minutes only for them to recover almost as quickly. Though it wasn’t an even split due to differences in market cap etc and it’s not even a listing of the biggest companies.
From what I recall multiple companies briefly took a 100+B hit, but again I don’t have the numbers on hand.
Well, this is what the media does, it is a low hanging fruit that will attract readers. The reality is that NVDA price was way too high and would fall anyway. This was just a catalyst.
It's not meaningless to people who accept that the vast majority of these reportings always use non-adjusted numbers and usually call out when that isn't the case.
reply