WSJ: Audacious $5-$7 trillion investment would aim to expand global AI chip supply.
See full article...
See full article...
Report: Sam Altman seeking trillions for AI chip fabrication from UAE, others
- Thread starter JournalBot
- Start date
I think it's inconceivable that the billionaire that's had like 3 wives, like a a dozen kids, including impregnating one employee at one of his companies, and tried to bribe another employee to sexually please him on a jet would... try to get board members to do drugs with him or take part in marital infidelity
Upvote
31
(31
/
0)
So we have another Sam asking everybody for money for "the future". Stay da f*CK away from my pension.
Upvote
15
(15
/
0)
You completely forgot that "the line goes up" and no limit is ever even implied? Everybody will obviously be rich!
/S for the cognitively disinclined.
Upvote
13
(13
/
0)
You only need to wait long enough for them to get really close and only need a few trillion more to finish.
Upvote
12
(12
/
0)
He should really be raising money to manufacture whatever drug he's on and give it out. It's gotta be cheaper than $7 trillion...
EDIT: I accidentally wrote billion instead of trillion because that's a number I'm not even used to typing.
Upvote
6
(7
/
-1)
I mean, if I were a sociopath with no empathy whatsoever, I would consider Adam Neumann's story a huge success....
Upvote
14
(14
/
0)
Agreed on the whole We Work 2.0 comment, but the market cap part of this is quite wrong. Nvidia is ~1.8T USD in market cap. The Shenzhen Stock exchange market cap is ~3.14T USD, while the Shanghai Stock exchange market cap is ~6.3T USD -- ~9.4T USD in total.
That being said, 1.8T market cap is pretty impressive given the narrow scope of products Nvidia produces.
Upvote
6
(6
/
0)
Thank you ! This is what I was hoping to find in the comments. This post should be promoted because it provides much needed context to the article
Upvote
10
(10
/
0)
I guess someone wanted to sound like a bigger asshole than Elon Musk, if only for a day. Well done, sir.
Upvote
8
(9
/
-1)
Yup it is all a counter to the idea that China has control of far too much.
Upvote
3
(3
/
0)
They're probably just jealous TBH. I have no idea whether or not that is true (don't give a damn anyway) But you should've seen people's comments here when Sam got kicked out from OpenAI... If you think LLM's "hallucinate" you should've seen the amount of bullshit being produced here, a good percentage of people were sure he must've raped someone, or something along those lines.
Upvote
-18
(1
/
-19)
robert.walter
Ars Praefectus
Even if a small part of this isn’t bullshit or a fraud and is successful this scares the hell out of me.
He could hire all the top people away from every company and cripple development.
He could hire all the top people away from every company and cripple development.
Upvote
0
(3
/
-3)
Upvote
3
(3
/
0)
He is not going to take over any companies or industries. He wants the world to produce a hundred times more chips when he rolls out superintelligent AI.
Upvote
4
(4
/
0)
Inflation adjusted, it took less than $300 billion to get us to the moon. WTF is $7 trillion for again?
Upvote
7
(7
/
0)
The only quarter-sensible explanation is "we have superintelligent AGI today, but it runs slowly and only the US and Chinese govt. can run it at all; if the world wants it to work for all of humanity, we need $7T today or the world will imminently be dominated politically and economically by these two countries to whom $7T will soon be a drop in the bucket."
Having an incrementally better LLM doesn't justify even $1T; the whole CHIPS Act was only $11 Billion.
Upvote
9
(9
/
0)
Altman's prompt to Gpt4: "How much money to become the Master of the universe?"
Upvote
6
(6
/
0)
This number is very much a fantasy, in fact a joke. There is no need to take it seriously, and there has to be some error in the reporting chain. Sam Altman is a smart man. Unless he genuinely thinks so, in which case he ought to be removed from his current leadership position and additionally be blacklisted from all corporate affairs of any company in the future. This has to be the dumbest thing I've read this entire decade.
Upvote
-5
(1
/
-6)
This is a mistake by the source that WSJ quoted. That's the only way I can think of it. If Sam Altman actually said that, I think he should be immediately told to give it a few months rest at home and never again be allowed entry into corporate America, that's how inexplicably deranged this idea is.
Upvote
-1
(2
/
-3)
This is surpassing bitcoin scammyness by a large margin. I sincerely hope some cooler heads prevail when thinking about allocating this kind of money.
Upvote
9
(9
/
0)
The Infrastructure of the future for a world of abundance. (assuming it's true, but Sam has embraced it either way
)
There's a research paper about it as well now!
Last edited:
Upvote
-9
(2
/
-11)
Tech goes through bubbles all the time, where excited investors spend too much on whatever the latest shiny toy is. Last year it was AI, the year before it was crypto, the year before that it was autonomous driving, and the year before that it was hydrogen-powered trucks.
A key part of the bubble strategy is to cash out before it bursts. This may be what Altman is doing. He's making a claim that the AI bubble will inflate enormously, and hoping that people with ready cash, like the oil sheiks, will fund it. That keeps the stock price of AI companies higher for longer, letting the Valley techbros skim off still more.
The risk here is that the wrong tech gets backed. Even if large language models turn out to really useful, it's not at all clear that Nvidia-style GPUs are the right way to do them. A lot of different hardware and software approaches are being tried right now, and it's way too early to put all the chips on one of them. If LLMs don't turn out to be all that valuable, like, say, Alexa-style voice recognition boxes, then that money is gone.
A key part of the bubble strategy is to cash out before it bursts. This may be what Altman is doing. He's making a claim that the AI bubble will inflate enormously, and hoping that people with ready cash, like the oil sheiks, will fund it. That keeps the stock price of AI companies higher for longer, letting the Valley techbros skim off still more.
The risk here is that the wrong tech gets backed. Even if large language models turn out to really useful, it's not at all clear that Nvidia-style GPUs are the right way to do them. A lot of different hardware and software approaches are being tried right now, and it's way too early to put all the chips on one of them. If LLMs don't turn out to be all that valuable, like, say, Alexa-style voice recognition boxes, then that money is gone.
Upvote
-1
(3
/
-4)
Everyone says whatever they're hawking is going to save the world. I got a scrum master certification to add some padding to the resume. That dude said agile project management could literally save the world. SBF was going to save the world, Facebook was going to create an interconnectedness that makes the world a better place, Elon Musk is going to take us interplanetary. It is the tech bro pitch that the world sucks and the thing that gives them piles and piles of money isn't a business, it's what's going to save then world. I think it would be refreshing at this point if someone had a new thing that was just a cool new thing, that wasn't going to disrupt society and "save the world".
The type of people giving trillions of petro dollars aren't looking to save the world, hell a lot of them are cool with modern slavery. They're wanting a return on investment.
The type of people giving trillions of petro dollars aren't looking to save the world, hell a lot of them are cool with modern slavery. They're wanting a return on investment.
Upvote
21
(22
/
-1)
Bubble or not: what Altman has actually done in the past, is released to the world a computer system that makes you feel like it understands you.
Upvote
-13
(1
/
-14)
Probably, given there's no guarantee GPUs are even the best inference platforms (for transformers, let alone if a new architecture replaces them). If AMD or Intel sort out software support they'll immediately start hemorrhaging market share, and it's clear the big players already are working with custom solutions for AI processing.
Already CPUs are sometimes better for inference in practical applications.
Upvote
4
(4
/
0)
I have been saying it for many months now: That tech bro is the next Zuck / Musk.
Upvote
1
(2
/
-1)
Is this a compliment, an insult, a statement of profound suspicion, or some/all of the above?
Upvote
11
(11
/
0)
Good point. This, and continued research into new algorithms. The chip design & algorithm design research sometimes need to progress in step with each other.
Upvote
2
(2
/
0)
Looking forward to when we look back on LLMs as a giant waste of investor money, like crypto and the metaverse.
Upvote
10
(11
/
-1)
It would be interesting to be a fly on the wall during the conversations involving people whose opinions actually matter.
It's not like semiconductor companies are little mom-and-pop shops run by frightened idiots who lack access to capital markets; or like the suppliers of the assorted equipment and specialty materials used to build fabs, testing, and packaging facilities hate money and seek to live by stability and tradition.
Altman may have the understanding of hardware difficulty one might expect from a software guy who has been told that he's a genius master of the universe; and it's possible that that reputation will help him considerably in rounding up relatively cheap capital; but he's going to be running up against some mixture of people who simply can't scale up any faster than ASML can make tools and universities can make specialist hires and people who don't just want to hear about how you have eleventy-billion dollars today; but how you will make it worth their while to spend several years building a fab that will need to be kept busy for at least a decade or two.
That's going to require some very hard selling of some combination of fantastically expensive minimum order quantities going out 20 years or a less expensive but harder speech check "it's not just me whining about GPUs being expensive; demand for semiconductors really will be several times higher than it is today in the future, indefinitely".
It's not like semiconductor companies are little mom-and-pop shops run by frightened idiots who lack access to capital markets; or like the suppliers of the assorted equipment and specialty materials used to build fabs, testing, and packaging facilities hate money and seek to live by stability and tradition.
Altman may have the understanding of hardware difficulty one might expect from a software guy who has been told that he's a genius master of the universe; and it's possible that that reputation will help him considerably in rounding up relatively cheap capital; but he's going to be running up against some mixture of people who simply can't scale up any faster than ASML can make tools and universities can make specialist hires and people who don't just want to hear about how you have eleventy-billion dollars today; but how you will make it worth their while to spend several years building a fab that will need to be kept busy for at least a decade or two.
That's going to require some very hard selling of some combination of fantastically expensive minimum order quantities going out 20 years or a less expensive but harder speech check "it's not just me whining about GPUs being expensive; demand for semiconductors really will be several times higher than it is today in the future, indefinitely".
Upvote
12
(12
/
0)
Let me get this straight:
1) Needs more money than Apple and MS valuation combined for Ai chips???
2) Appears to be serious, and thinks investors will come forward to make it happen
3) Sam must be assuming with enough money a chip fab can just be spontaneously willed into existence, and this is not a 20-year project requiring millions or billions of man-hours for design, construction, and operation
This feels like the old "Step 1: Draw a Circle. Step 2: Draw the rest of the rabbit" bit but with semiconductor manufacturing, except step 1 might as well be "simulate the entire universe"
1) Needs more money than Apple and MS valuation combined for Ai chips???
2) Appears to be serious, and thinks investors will come forward to make it happen
3) Sam must be assuming with enough money a chip fab can just be spontaneously willed into existence, and this is not a 20-year project requiring millions or billions of man-hours for design, construction, and operation
This feels like the old "Step 1: Draw a Circle. Step 2: Draw the rest of the rabbit" bit but with semiconductor manufacturing, except step 1 might as well be "simulate the entire universe"
Upvote
11
(11
/
0)
ACCORDING TO JOURNALISTS WHOM ACCORDING TO PEOPLE FAMILIAR WITH THE MATTER WHOM ACCORDING TO SAM ALTMAN...
Third Party reporting. People will really believe anything nowadays.
This world is doomed.
Third Party reporting. People will really believe anything nowadays.
This world is doomed.
Upvote
-10
(3
/
-13)
There's only a few countries that have banned/restricted the use of ai in weapons.
Bet there's plenty of testing going on in Ukraine at the moment.
older article https://www3.nhk.or.jp/nhkworld/en/news/backstories/441/
Bet there's plenty of testing going on in Ukraine at the moment.
older article https://www3.nhk.or.jp/nhkworld/en/news/backstories/441/
Australia's got plenty of space for fabs (and lots of sand) and is geo-physically(and politically) more stable than where most of the current fabs are.
Last edited:
Upvote
2
(2
/
0)
Upvote
0
(0
/
0)
The funny thing about the phrase “$5-7 trillion” is that there are TWO TRILLION DOLLARS between the ends of that range.
Upvote
11
(11
/
0)
More than the budget of every country. Even the US is "only"six trillion.
Upvote
5
(5
/
0)
Uncritical copy pasting of another article with no research into the truth of these obviously nonsensical claims. I guess I should pay more attention to the article authors on Ars going forward.
Upvote
-11
(1
/
-12)
Upvote
4
(5
/
-1)