US Tech Stocks Hit By Concerns Over Future of AI Boom 44
US tech stocks sold off as warnings that the hype surrounding AI could be overdone hit some of the year's best-performing shares. From a report: Nvidia, the chips group that has surged to become the world's first $4tn company on the back of AI, fell 3.5 per cent on Tuesday, while software group Palantir dropped 9.4 per cent and chip designer Arm shed 5 per cent.
The tech-heavy Nasdaq Composite closed down 1.4 per cent, the biggest one-day drop for the index since August 1. The blue-chip S&P 500 fell 0.7 per cent. European and Asian markets largely followed Wall Street lower on Wednesday. [...] Japan's Nikkei 225 index fell 1.5 per cent and South Korea's Kospi slipped 0.6 per cent. Futures price indicated moderate declines when Wall Street opens.
Traders pinned some of the declines in the US on a critical report on Monday authored by a branch of the Massachusetts Institute of Technology. Researchers said "95 per cent of organisations are getting zero return" from their investments in generative AI, the technology that has sent US stocks soaring to record highs in recent months.
The tech-heavy Nasdaq Composite closed down 1.4 per cent, the biggest one-day drop for the index since August 1. The blue-chip S&P 500 fell 0.7 per cent. European and Asian markets largely followed Wall Street lower on Wednesday. [...] Japan's Nikkei 225 index fell 1.5 per cent and South Korea's Kospi slipped 0.6 per cent. Futures price indicated moderate declines when Wall Street opens.
Traders pinned some of the declines in the US on a critical report on Monday authored by a branch of the Massachusetts Institute of Technology. Researchers said "95 per cent of organisations are getting zero return" from their investments in generative AI, the technology that has sent US stocks soaring to record highs in recent months.
Next quarters numbers (Score:3, Insightful)
Woops... (Score:5, Insightful)
They already did (Score:4, Informative)
Algorithmic trading has been a thing for 3 decades now. Only a few markets still use open outcry real traders - metals IIRC - and the rest AFAIK just babysit the computers so whether they should still be called traders is open to question.
Re:They already did (Score:5, Insightful)
Re:Woops... (Score:5, Funny)
warnings that the hype surrounding AI could be overdone
And in other shocking and unexpected news, scientists have reported that the Pacific Ocean may be damp.
Maybe investors read "The Economist" (Score:5, Interesting)
Headline: "Palantir might be the most overvalued firm of all time"
https://www.economist.com/fina... [economist.com]
That could explain the 9.4% drop.
Re: (Score:2)
One can only hope Palantir dies. But someone will buy their surveillance tech and continue.
Pay attention to Warren Buffet (Score:5, Insightful)
There's a reason why Berkshire Hathaway is sitting on record amounts of cash at the moment. Buffet doesn't believe in paying over the odds for the things he invests in, and literally everything is way overpriced right now. It'll take more than a 1.4% correction to fix that too.
Re: (Score:2)
GPU (Score:4, Interesting)
I'm really surprised the market stayed so bullish on Nvidia for so long. GPUs are just a temporary stop-gap measure to run AI models, as it is the best generic, off-the-shelf hardware for that purpose when the boom began.
Purpose-built chips for running AI inference (like Groq) blow GPUs away, both in performance, and more importantly, power efficiency. Unless Nvidia is taking this surge in money and using it purely for R&D on entirely new non-GPU AI hardware, this bubble is going to pop for them in a massive way when GPU sales go back to being driven by graphic rendering.
Re:GPU (Score:5, Informative)
'GPU' is whatever nVidia wants it to be and however many of whatever type of compute element they want can be in a 'GPU'. Unlike the crypto boom, where nVidia didn't really embrace it and largely got left behind, nVidia has fully embraced their AI role and so long as the market will reward it, they are all in on whatever makes a 'GPU' competitive for the solution. In some fields the high end Blackwell GPUs actually represented a retreat in performance, largely owing to the architecture catering to AI sensibilities.
Re:GPU (Score:4, Insightful)
This nVidia is absolutely all in AI and while I think it is a 'bubble' they will also be the last to be exposed to the pop.
Look at the VC investments and stock prices of the companies delivering AI applications, they are all around that magic 200X earnings number that usually signifies the point beyond which there is not realistic positive return for new investors, unless they happen to get really really lucky and pick the big winner in the specific vertical its kinda like Highlander at this stage, there can be only one.
Think about it like pouring money in IBM because you believed in an OS/2 future in 1988 at that point vs buy MSFT shares; but both were trading at 200X earnings. With MSFT you'd still have made out all right probably, (could not find their actual PE in 88 quickly). My point is there is probably one coding assist, one photo enhancer application, etc to rule them all and everything else will be also-rans in a given vertical. The rest will fold, get acquired cheaply if they have any unique IP, or muddle along without much growth. Unless you picked the big winner you'll never get out what you put in.
On the other hand after each gold rush in a given space peters out, the AI exuberance and capital flows will be pushed into some other vertical in succession. The need for generic-ish hardware good for running ML models will persist. nVidia can expect to keep shoveling chips out the door until either - 1) the market matures and the AI boom full ends, or 2) they hit scaling problems that prevent them from really producing 'better' SI from a performance vs power vs foot print perspective and can't compete with their own previous generation products availible because of the previous vertical app market decline.
The "cloud" is going to be a drag on them too. In the 2000 internet bubble, there was a stigma attached to the used anything, so hardware from failed start ups ended up discounted and sold into secondary markets cheaply. On the hand buying so many units of compute from AWS, well that is a pure commodity and AWS can just leave their compute plant installed and rent it to the next client. Amazon does not need to replace or upgrade it, unless the cost of operation is improved enough to justify the cap-ex. To that end we can expect some flattening of Nvidia sales before they big drop. So you if you watch the quarterly reports, you'll know exactly when to get out. (This is my theory anyway)
Re:GPU (Score:4, Interesting)
Re:GPU (Score:4, Interesting)
Given Tesla's market cap, nothing surprises me anymore. Even now they remain the most valuable car company and have been for a decade now, that despite showing no sign of even remotely ever catching the traditional car companies in sales volume or revenue, and worse, now losing ground to competitors from China.
Nothing about NVIDIA's stock surprises me.
But that said I think your analysis is a bit off. NVIDIA don't just sell GPUs. They sell a whole world of HPC tech, as complete end to end solutions. It's like comparing Intel to a manufacturer of super computers.
Re: (Score:3)
Don't forget, Tesla approved a pay package for Elon Musk that is worth ALL THE PROFITS they've ever made.
Re: (Score:2)
Re:GPU (Score:4, Insightful)
Purpose-built chips for running AI inference (like Groq) blow GPUs away
For a single technique, sure. The field is moving quickly enough that you run the risk of winding up owning a bunch of suddenly obsolete hardware. The advantage of using GPU-derived chips is that they can do multiple different types of task.
Re:GPU (Score:4)
I'm really surprised the market stayed so bullish on Nvidia for so long. GPUs are just a temporary stop-gap measure to run AI models, as it is the best generic, off-the-shelf hardware for that purpose when the boom began.
Purpose-built chips for running AI inference (like Groq) blow GPUs away, both in performance, and more importantly, power efficiency. Unless Nvidia is taking this surge in money and using it purely for R&D on entirely new non-GPU AI hardware, this bubble is going to pop for them in a massive way when GPU sales go back to being driven by graphic rendering.
I agree that GPUs are temporary solutions that have excelled during this time of AI exploration and research. Models are changing constantly at a furious pace. It's not a surprise that ASICs are not well suited for software that is architecturally changing. Google et al. do use their internal chips for those applications that are more set. Again not a surprise. However, for most training and a lot of inference, all the hyperscalars still buy a lot of Nvidia GPUs.
This wild west of AI model progress has been ongoing for many years now. The big question is how long the wild west will last? A few more months, a few years, or even longer?
Nvidia detectors claim/propose that AI model stability is imminent, which will lead to the downfall of Nvidia as ASICs dominate. Perhaps that's true or even inevitable, but the important question is when. It's certainly not now.
Re: (Score:2)
But as far as Groq replacing Nvidia for the whole process, it currently cannot do that. Maybe one day it "could," but then NVidia "could" make dedicated inference hardware too.
A few things... (Score:4, Interesting)
Yes you have that report, you have another about how terribly overvalued Palantir is, and you have the general reaction to GPT-5, which was amazingly underwhelming.
Lots of signs that the 'march to AGI' actually isn't going to get more fundamentally capable than we've managed so far without some sort of unpredictable breakthrough that can't just be guaranteed through more funding. Still lots of area for LLMs to be more useful as they are already, but that's not what the investors were betting on.
Re:A few things... (Score:5, Insightful)
A lot of us are old enough to have worked through the dot-com bubble and crash. The AI boom over the last year or two feels a lot like the late 90s leading up.
That said, if the market hadn't gone so all-in on the Internet and world wide web, investing stupid amounts of money into so many idiotic crackpot ideas that were doomed to fail, we arguably wouldn't have the Internet that we enjoy today.
In other words, you have to throw a lot of shit at the wall to see what will stick. 90% of just about everything we humans create is unremarkable. It's the 10% that we end up with that matters.
And we will get the 10% out of AI. Which is another way of framing what you are saying.
I've been a pretty big skeptic when it comes to the hype vs actual productivity gains, but that doesn't mean I haven't found a few areas where it has been useful in my workflow. LLMs are a cool magic trick. They can create some breathtaking illusions and are exceptionally useful at a few narrow applications such as searching and pattern matching.
While I don't want to see the ugly side of the crash (layoffs and unemployment), I am looking forward to when the hype train is over so everyone shuts up about their overstated fantasy and we can get on with doing actual work. The blockchain hype was annoying too, but since it wasn't obvious how blockchain could be applied to all types of work we didn't hear every single CEO at every single company talk about how AI is going to fundamentally change the way we work.
Re:A few things... (Score:4, Insightful)
What do you mean enjoy? Virtually everything I loved about the Internet in the 90's is now dead.
I remember when the worst thing I had to worry about was broken layout in Netscape because IE did CSS a bit differently. Now, everything is an app, home pages are dead, web search is useless, and a huge chunk of web sites are either functionally useless or give me a blank page if I don't have a Chromium-based browser.
I like that the Internet bandwidth is more plentiful than what it used to be, but I'd love to have the old WWW back. AI isn't going to make anything better.
.com promised profitability, AI promises usefulnes (Score:3)
A lot of us are old enough to have worked through the dot-com bubble and crash. The AI boom over the last year or two feels a lot like the late 90s leading up.
That said, if the market hadn't gone so all-in on the Internet and world wide web, investing stupid amounts of money into so many idiotic crackpot ideas that were doomed to fail, we arguably wouldn't have the Internet that we enjoy today.
In other words, you have to throw a lot of shit at the wall to see what will stick. 90% of just about everything we humans create is unremarkable. It's the 10% that we end up with that matters.
And we will get the 10% out of AI. Which is another way of framing what you are saying.
I've been a pretty big skeptic when it comes to the hype vs actual productivity gains, but that doesn't mean I haven't found a few areas where it has been useful in my workflow. LLMs are a cool magic trick. They can create some breathtaking illusions and are exceptionally useful at a few narrow applications such as searching and pattern matching.
While I don't want to see the ugly side of the crash (layoffs and unemployment), I am looking forward to when the hype train is over so everyone shuts up about their overstated fantasy and we can get on with doing actual work. The blockchain hype was annoying too, but since it wasn't obvious how blockchain could be applied to all types of work we didn't hear every single CEO at every single company talk about how AI is going to fundamentally change the way we work.
The dot-com bubble produced useful services and promised profitability. The AI bubble only produced promises of useful services. I don't know of a single product that is using LLMs for a useful service. I am sure they're out there, but we're over 3 years into this bubble. In 1998, I knew of many useful commercial websites, most of which are still thriving today. I know many companies selling me toolkits and promises of things I can do with AI...but why aren't startups thriving on AI-based services? In
Re: (Score:2)
lol (Score:3)
I love how stupid the stock market is. Throw money at the trend! Oh shit we threw money at a trend!
Re: lol (Score:2)
If you look at historical charts of the stock market, do crashes start looking like noise on an ever- upward-trending curve?
That whooshing sound you hear... (Score:2)
... is the bubble bursting.
Re:That whooshing sound you hear... (Score:5, Interesting)
Just about 40 years after the first big "AI" bubble back in the 1980s. At DEC there was cautious optimism, and even that turned out to have been considerably overblown. The guys in the AI group, who wryly called themselves "the artificial intelligentsia", were the first to admit that they were mainly working on expert systems. There was a lot of progress, and some useful software was written. But nothing like the immense wave of marketing hype had suggested.
Back about then, some enthusiastic junior guy brought our manager the latest Gartner Group report explaining what a great future DEC had and all the fabulous technology. Barely glancing through it, he grunted "Still drinking our own bathwater, I see" and returned to his spreadsheet.
The really funny thing about that is that the whole of today's "AI" hype, with the LLMs and all, is just that: drinking our own bathwater.
The software absorbs as much of the Web's written content as it can, and essentially learns how to spit out more of the same. No wonder it reflects our prejudices! No wonder people discover it can't actually "think"!
Re: That whooshing sound you hear... (Score:2)
I thought it was just me flushing the toilet while reading this article on the toilet this morning.
Re: (Score:2)
The present paradigm (Score:3)
The idea of AI needing as much power as the rest of the nation is an obvious non-starter. At least to me.
Recommissioning all of the old nuc plants to provide power for AI, and we're going to need all the water in your city to cool the servers.
And remember, useable fissiles are not in infinite supply. Let's at least see about reducing the power requirements.
Re: (Score:2)
Re: (Score:2)
Yes efficiency has been lacking in the designs, every new consumer gaming GPU I see get released I begin to wonder when you'll need a dedicated circuit to run a desktop.
Efficiency, and also the Chips now have a lot of internal leakage due to the tiny size of their sub components.
Palantir sahre price is 524.23 (Score:2)
Re: (Score:2)
Re: Palantir sahre price is 524.23 (Score:1)
It's the new new new economy!
Um... ALL US tech companies are overvalued... (Score:2, Informative)
With the fact that companies have been laying people off so they can do stock buybacks to keep the stock value up... this means that it is only a matter of time before they don't have the revenue to keep the clown show going, and once Wall Street realizes that these companies can't do exponential growth forever, things are going to crash hard.
Wall Street has been playing shenanigans for years, and is completely separate from Main Street, that there is going to be an order of magnitude correction. We alread
God I cannot wait for the bubble to burst (Score:3)
I am not at all religious, but I pray daily to speed the time that AI and ChatGPT go the way of NFTs and the only people talking about it are idiot tech bros that I can ignore.
Real progress is being made, but... (Score:3)
Real progress is slow and uneven
Progress sometimes leads to profit, or not
Progress sometimes makes things better, but not necessarily more profitable
Salesweasels sell immature tech and promise extraordinary profit
Clueless managers and executives believe the hype
Investors panic when the profit doesn't magically appear quickly
Meanwhile, people learn to use the new tools. Sometimes they find the tools useful, sometimes not
The tools keep improving
biggest one-day drop for the index since August 1 (Score:2)