Nvidia Passes Apple to Become the World's Most Valuable Company - Powered by Demand for AI Chips (nbcnews.com) 52
"Nvidia dethroned Apple as the world's most valuable company on Friday..." reports Reuters, "powered by insatiable demand for its specialized artificial intelligence chips."
Nvidia's stock market value briefly touched $3.53 trillion, slightly above Apple's $3.52 trillion, LSEG data showed... In June, Nvidia briefly became the world's most valuable company before it was overtaken by Microsoft and Apple. The tech trio's market capitalizations have been neck-and-neck for several months. [Friday] Microsoft's market value stood at $3.18 trillion, with its stock up 0.8%...
Nvidia's shares hit a record high on Tuesday, building on a rally from last week when TSMC, the world's largest contract chipmaker, posted a forecast-beating 54% jump in quarterly profit driven by soaring demand for chips used in AI.
The article points out that by the end of the day Friday, Nvidia's valuation had dropped to $3.47 trillion, while Apple had risen to $3.52 trillion...
Nvidia's shares hit a record high on Tuesday, building on a rally from last week when TSMC, the world's largest contract chipmaker, posted a forecast-beating 54% jump in quarterly profit driven by soaring demand for chips used in AI.
The article points out that by the end of the day Friday, Nvidia's valuation had dropped to $3.47 trillion, while Apple had risen to $3.52 trillion...
Let's be clear here (Score:5, Insightful)
It's similar to when PCs started replacing Junior accountants only on a much much larger scale and without new forms of work to replace and absorb the job losses.
Hell sometime in the next 5 to 10 years we are going to have to figure out what to do with 7 million taxi and Uber drivers. Those waymos are all over Phoenix and San Francisco and they work. The only thing holding them back right now is getting enough miles and enough legal precedence that they can sort out The liability.
Basically it's an industrial revolution going on. Folks who yell Luddite every time this is brought up don't know the actual history of the two industrial revolutions. There was massive technological unemployment for decades following them until new tech, and a whole lot of dead people in two world wars, brought us back to some semblance of full employment. It also helped that we basically blew all of Europe to Kingdom come and it had to be rebuilt generating a ton of jobs...
The point is we are not equipped as a species to deal with what's coming. Too many centuries of, if you don't work you don't eat and we are suddenly going to have tens of millions if not hundreds of millions of people who we just don't need to work.
What do you do when you've got 30 or 40 million people that you just don't have a use for? Traditionally some loudmouth comes along and gives them firearms and points them in the direction of whoever still has a job and money
Re: (Score:2)
AI requires an avatar (robot, delivery drone, or car) to start "replacing" humans. Right now the world is about 100 to 150 years from a robot that can meaningfully replace humans for most tasks. When we start seeing (non-gimmick) human-less farms, deliveries, restaurants, or factories that's when we can start worrying from a policy standpoint. Right now, robots are really dumb and will remain so for, like I said, 100+ years. How many lights-out factories exist?
Re: (Score:3)
AI requires an avatar (robot, delivery drone, or car) to start "replacing" humans.
Most 1st world jobs do not require any physical manipulation of reality.
I get paid to move electrons, not atoms.
Re: (Score:2)
Most 1st world jobs do not require any physical manipulation of reality.
Bizarre claim. Haven't you ever heard of blue collar jobs? This is one of those statement that its hard to tell if someone's serious or not.
Re: (Score:1)
There are just absolute tons of jobs where people do zero traditional "work". They are only manipulating data on computer systems. My job is almost that, except there is a bunch of paperwork too. However, since I started working remotely, most of that is done for me by the clerical department. And frankly, most of it could be done by machines, which can stuff envelopes. The humans are more reliable than those machines though, and they perform some other knowledge-related functions as well.
However, software
Re: (Score:2)
Haven't you ever heard of blue collar jobs?
Google says that 18% of jobs in America are blue-collar.
Blue-collar jobs include manufacturing, construction, transportation, mining, etc.
Re: (Score:1)
Most 1st world jobs do not require any physical manipulation of reality.
Bizarre claim. Haven't you ever heard of blue collar jobs? This is one of those statement that its hard to tell if someone's serious or not.
The end of the 20th century marked the writing of the Balfour agreement which depending on who you ask clearly defined which national economic agenda would allow a country to be 1st world while deciding which nations will specialize in the "blue collar" work.
They even defined "dark" nations which to me meant it matters to them more how you look doing something than what you were actually doing.
Re:Let's be clear here (Score:5, Insightful)
Predicting the future is largely ridiculous, like your 100 - 150 years. When we look back in time and claim someone predicted this or that would happen, we ignore all the other someones who claimed something different. It is just cherry picking.
Clear as Mud. (Score:1)
AI requires an avatar (robot, delivery drone, or car) to start "replacing" humans. Right now the world is about 100 to 150 years from a robot that can meaningfully replace humans for most tasks. When we start seeing (non-gimmick) human-less farms, deliveries, restaurants, or factories that's when we can start worrying from a policy standpoint. Right now, robots are really dumb and will remain so for, like I said, 100+ years. How many lights-out factories exist?
The first human flight, was in 1903. Less than 70 years later, we put a human on another planet. 30 years ago you were dialing up to the internet on a landline, and couldn’t even imagine gigabit fiber offerings in your home.
If there’s one thing humans have proven they consistently suck at, it’s predicting the future. But it’s quite an extreme stance to assume we’re going to basically get jack shit done in the next 100-150 years with regards to autonomous machines and AI. As
Re: (Score:3)
Well, Honda ASIMO came out in 2000 and today our best robots - Optimus, Unitree, etc. are barely marginally better. Hardly the same as going from Wright Flyer to Lindbergh flying across the Atlantic within 25 years. Extrapolate 25 years on that timeline, robots will still walk with very little confidence. We're decades away from a robot that can confidently or even non-confidently in a controlled setting accomplish all of the following tasks:
1. From a table on which there are about 20 screws of varying size
Re:Clear as Mud. (Score:4, Insightful)
Well, Honda ASIMO came out in 2000 and today our best robots - Optimus, Unitree, etc. are barely marginally better. Hardly the same as going from Wright Flyer to Lindbergh flying across the Atlantic within 25 years. Extrapolate 25 years on that timeline, robots will still walk with very little confidence. We're decades away from a robot that can confidently or even non-confidently in a controlled setting accomplish all of the following tasks:
1. From a table on which there are about 20 screws of varying sizes randomly strewn about, pick up only the eyeglass frame screw.
Computer vision should be able to do this task rather easily, assuming high enough quality cameras.
2. Insert and screw it into an eyeglass frame using a screwdriver picked up from the table.
Why would you waste effort doing that when you can build a screwdriver that pops out of the manipulator arm and is motor-driven? Yeah, it takes a lot of work to screw in a screw in the way that a human would do it, because humans don't have any motors to spin a screwdriver directly, so we have to do weird, complex motions with our arms and hands and fingers and wrists if we want to twist things. But that doesn't mean that the task is hard to automate. It just means that nobody in their right minds would automate it in a way that mimics how humans do it.
3. Insert DRAM into a motherboard.
Solder on the DRAM. Problem solved.
Besides, this actually *has* been done. There's even a YouTube video of it [youtube.com].
4. From a small bag of various trinkets, feel around and pick out only the rubber band from it.
Probably correct. I doubt there are robot arms with adequate touch sensing to make this possible. It's not that it couldn't be done with capacitative touch sensor skin or something, but it wouldn't be a very practical robot, so nobody builds one, as far as I know.
5. Get one Tylenol out of a pill bottle.
You start with a special hand design that has a small opening that is just bigger than a pill, you use vision to see if a pill is properly aligned, and you shake the container as needed to realign the pills until one drops out. This shouldn't even be all that hard. Pointless, but not all that hard.
Besides, nobody would ever do this. Instead, they would put the pills in blister packs, and the robot would slice off a single pill packet and open it.
6. Sculpt and shape a recognizable face on a piece of clay
I'm not even sure this is hard, either. Nobody has done it, but conceptually, it could probably be done with a modified CNC process that iteratively converges on the correct shape by compressing the surface. I don't even think this inherently requires AI — just a complete 3D model of the face scaled to the exact same total volume as the clay, a bunch of iterative algorithms, and a very precise LIDAR setup or similar to be able to measure how close or far the clay object is from ideal at any given point on its surface.
7. Assemble a lego set
Pick-and-place machines used in electronics manufacturing do things like this with far greater precision than would actually be needed to achieve this goal. The fact that nobody has done this proves only that nobody cares enough to do it, not that it is difficult.
8. Insert a thread into a needle.
Done, with video evidence [youtube.com].
9. Sew a pocket into fabric.
This appears to be a solved problem [clnusa.com], so presumably only reason this isn't being done automatically is some combination of cost and momentum.
10. Fold clo
Re: (Score:2)
it takes a lot of work to screw in a screw in the way that a human would do it, because humans don't have any motors to spin a screwdriver directly, so we have to do weird, complex motions with our arms and hands and fingers and wrists if we want to twist things. But that doesn't mean that the task is hard to automate. It just means that nobody in their right minds would automate it in a way that mimics how humans do it.
Humans are inefficient, but some humans are more efficient than others [smbc-comics.com].
Re: (Score:2)
it takes a lot of work to screw in a screw in the way that a human would do it, because humans don't have any motors to spin a screwdriver directly, so we have to do weird, complex motions with our arms and hands and fingers and wrists if we want to twist things. But that doesn't mean that the task is hard to automate. It just means that nobody in their right minds would automate it in a way that mimics how humans do it.
Humans are inefficient, but some humans are more efficient than others [smbc-comics.com].
ROFL. Thanks for the link.
Re: Clear as Mud. (Score:1)
No, no, no.. A single robot has to do ALL of that !
If you are making specialized machines to do each task it's already been done, automated manufacturing machines/robots can so almost any one task better n faster than humans.
It's like AGI when the same machine /software can do all general purpose stuff.
Re: (Score:2)
Well, Honda ASIMO came out in 2000 and today our best robots - Optimus, Unitree, etc. are barely marginally better. Hardly the same as going from Wright Flyer to Lindbergh flying across the Atlantic within 25 years. Extrapolate 25 years on that timeline, robots will still walk with very little confidence. We're decades away from a robot that can confidently or even non-confidently in a controlled setting accomplish all of the following tasks:
1. From a table on which there are about 20 screws of varying sizes randomly strewn about, pick up only the eyeglass frame screw. 2. Insert and screw it into an eyeglass frame using a screwdriver picked up from the table. 3. Insert DRAM into a motherboard. 4. From a small bag of various trinkets, feel around and pick out only the rubber band from it. 5. Get one Tylenol out of a pill bottle. 6. Sculpt and shape a recognizable face on a piece of clay 7. Assemble a lego set 8. Insert a thread into a needle. 9. Sew a pocket into fabric. 10. Fold clothes from a laundry hamper or pile.
These strike me as the same arguments made by the welders at the old industrial scale plant my dad used to work at back in the 1980s. Except, no robots needed to do every task the workers did. Instead, specialized robots were brought in that only had to take stock, take welding rod, and perform precisely defined actions, repeatedly, all day every day. And only one welder was kept on staff for touch-ups if something went wrong.
We aren't replacing generalized workers. But most jobs aren't generalized. I know
Re: (Score:3)
AI requires an avatar (robot, delivery drone, or car) to start "replacing" humans. Right now the world is about 100 to 150 years from a robot that can meaningfully replace humans for most tasks. When we start seeing (non-gimmick) human-less farms, deliveries, restaurants, or factories that's when we can start worrying from a policy standpoint. Right now, robots are really dumb and will remain so for, like I said, 100+ years. How many lights-out factories exist?
The first human flight, was in 1903. Less than 70 years later, we put a human on another planet.
The Moon isn't technically a planet.
30 years ago you were dialing up to the internet on a landline, and couldn’t even imagine gigabit fiber offerings in your home.
Thirty years ago, I switched between dialing up from home and having much faster service on Ethernet from the university. I could imagine much faster offerings in my home. I just didn't have any clue how it was going to happen, how quickly it would happen, or who was going to pay for it.
Thirty years ago, I predicted that streaming services would basically bring the broadcast TV industry to its knees. Twenty-five years ago, I predicted that poor salaries in the TV indus
Did you even bother reading the post? (Score:2)
And call centers don't need an avatar. You're mixing up robotic automation which has been guzzling down jobs for 40 years with the coming AI automation boom.
Seriously let's talk about those robots. Because we've been rolling them out in factories and farms for over 40 years. Googled The phrase "70% of jobs lost since the 1980s to automation" and you'll find an article from bus
Re: (Score:2)
All those people can be put to work building a wall between the US and Mexico.
When they're done with that, they can start building sea walls along the coasts.
That will keep them busy for quite some time.
Build the wall or starve.
A sea wall? Why? Are we suddenly living in Pacific Rim's universe? Trying to block the Kaiju?
Re: (Score:1)
They became Power-Point jockies, talking about synergy and growth acceleration strategies.
Re: Let's be clear here (Score:2)
I agree.. the current hypervaluation of Nvidia is akin to the tool-sellers in the guld rushes of the past; they were the ones that made it big, not the people working away in the hills and riverbeds of the US West.
Re: (Score:2)
Re: (Score:2)
Re: Let's be clear here (Score:2)
Well, NVIDIA is not really building gaming chips. They are building AI chips. Most of what they do technologically is to there to help compute AI workloads, not to render triangles.
Game rendering needs really fast single precision floating point calculations. Most of modern GPUs transistor footprint is about tensor multiplication at various precisions.
Not saying you can't do better with a truely specialized design. But NVIDIA gardware is quite generic in its design which makes it easy to adapt to new genera
Re: (Score:2)
Nvidia got famous for building GPUs which are chips that do multiply-add really fast. The GPU's key feature is that it gives software extensive control over what gets multiply-added. It turns out that operation is good for more than manipulating triangles and simulating light reflection. But its original purpose is to perform the calculation
y = Ax + b
which is an affine transform. It turns out you can also do:
y = \Beta x + b,
which is linear regression, and
y = Wx + b
too, because they're exactly the same thing
Re: Let's be clear here (Score:2)
Oh yes, I can totally remember hardware accelerated 8-bit floating point number calculations before AI workload got popular. Oh wait, no I don't!
Re: (Score:2)
Oh, you're right! Nvidia doesn't build GPUs anymore because they now support an additional datatype in addition to the dozen or so they supported before!
How Greed, becomes Poor. (Score:2)
The reason for the insane value is because everyone is expecting those chips to start replacing workers, particularly those in the 12 to $24 an hour range.
The reason for this insane valuation, is because Greed N. Corruption, CEO can’t wait to get their hands on the technology that will KILL their revenue strea, wait, it’s gonna do WHAT?!?
Funny part about Greeds delusions and replacing a metric fuckton of paycheck-earning meat sacks, with machines. That metric fuckton of meat sacks, used to maintain your fucking revenue stream, moron.
Enjoy those plummeting sales numbers and stock crashes after you fire the reason you got rich in the first place.
Re: (Score:2)
Those waymos are all over Phoenix and San Francisco and they work. The only thing holding them back right now is getting enough miles and enough legal precedence that they can sort out The liability.
No, they DONT work.. And those are ideal conditions, everywhere else is harder, not just a little much exponentially harder. Self-driving cars have been promised next year for the last 10 years, and they are still 1 year away by optimist measures, and more than 10 years away for everybody else.
Re: (Score:2)
The reason for the insane value is because everyone is expecting those chips to start replacing workers, particularly those in the 12 to $24 an hour range.
AI-related corporate profits are of two varieties. One is the human worker replacement one. The other is the new industry creation one. It's the latter that is driving the insane valuations. The former is just another version of private equity slashing jobs, and many/most investors realize that such strategies don't produce true value. Missing out on the latter is terminal for a company, and it's that exhilaration and fear that is driving the mad dash for AI.
As just one example, imagine an AI replaceme
Re: (Score:2)
The reason for the insane value is because everyone is expecting those chips to start replacing workers, particularly those in the 12 to $24 an hour range. It's similar to when PCs started replacing Junior accountants only on a much much larger scale and without new forms of work to replace and absorb the job losses. Hell sometime in the next 5 to 10 years we are going to have to figure out what to do with 7 million taxi and Uber drivers. Those waymos are all over Phoenix and San Francisco and they work. The only thing holding them back right now is getting enough miles and enough legal precedence that they can sort out The liability. Basically it's an industrial revolution going on. Folks who yell Luddite every time this is brought up don't know the actual history of the two industrial revolutions. There was massive technological unemployment for decades following them until new tech, and a whole lot of dead people in two world wars, brought us back to some semblance of full employment. It also helped that we basically blew all of Europe to Kingdom come and it had to be rebuilt generating a ton of jobs... The point is we are not equipped as a species to deal with what's coming. Too many centuries of, if you don't work you don't eat and we are suddenly going to have tens of millions if not hundreds of millions of people who we just don't need to work. What do you do when you've got 30 or 40 million people that you just don't have a use for? Traditionally some loudmouth comes along and gives them firearms and points them in the direction of whoever still has a job and money
Luckily (not really) for us, we're right in the midst of setting up a new worldwide war. We won't call it a world war, because we don't want to prove ourselves wrong on WWIII being the last technological war, and nobody seems dumb enough on the surface to start hurling nukes at one another, despite the bluster from the unhinged world leaders to the contrary. Granted, if someone does, we'll solve the unemployment problem really, REALLY quickly. Otherwise, we'll solve it slowly, by killing each other off on b
Nvidia is the greatest company- (Score:4, Funny)
Re: (Score:2)
...great time to snap up Apple stock.
Apple is currently $6 off its all-time high [marketwatch.com]. Not that investors shouldn't buy it, but it's not at a bargain price unless you're looking a decade down the road.
Re: (Score:2)
Okay, you are right. Keep me out of the investment room.
How can this be maintained? (Score:4, Insightful)
I'm not sure where we are on this cycle. Sometimes I think something is going to collapse, but then it just stays around or even grows for some unknown reason. AI is still a thing, but we have many people offering the same tools for the job, and supply is starting to outpace demand. When funding in the AI sector starts drying up, I'm expecting to see a lot of patent wars and people trying to claim a lot of things AI-based, similar to how the smartphone industry had all the patent litigation for years, then after that, the market shakes out into a few providers. Likely OpenAI will survive, maybe Microsoft, perhaps 1-2 more, but once there is a point where growing LLMs gives diminishing returns, all the market push for GPU and matrix multiplication hardware is going to contract, likely sharply. Especially when newer LLMs have fewer sources of info as people close their gates to AI, both in access to their sites, as well as via access to IP. Maybe something can be set up so LLMs can feed to each other without causing the model to collapse [techtarget.com].
Apple also is going to have an important week coming this week. Not as big as a new iPhone, but new Mac announcements, and even though it isn't radical as Apple Silicon, new Macs with features that people look for are going to be a positive income bump. The M1/M2 Macs are starting to lose their warranties, and businesses generally don't want to be on hardware that doesn't have a service plan.
As for Nvidia, there will be always a market for GPUs. It may not be as big when the AI bubble or the crypto bubbles were going full steam, but they definitely have a niche.
Re:How can this be maintained? (Score:5, Interesting)
Have a theory this is a manifestation of wealth inequality (which is also a manifestation of inflation beyond money printer go brrrr, but that's a different idea).
So it works like this- every general thing of general value has already been snatched up, so if you are sitting on a mountain of money you need something to invest in. It doesn't even matter what it is, as long as you've covered every bet on the roulette wheel (and crowded out anyone else from making a bet), something will pay off.
So AI, crypto; it doesn't matter. Something will pay (and you can always threaten to crash the economy should the losses get too great).
Re: (Score:2)
Also it doesn't help bubbles that those with all the money will be the dumbest investors, and the more boosted the bubble becomes the more this feature is exacerbated
Re: (Score:2)
Have a theory this is a manifestation of wealth inequality (which is also a manifestation of inflation beyond money printer go brrrr, but that's a different idea).
So it works like this- every general thing of general value has already been snatched up, so if you are sitting on a mountain of money you need something to invest in. It doesn't even matter what it is, as long as you've covered every bet on the roulette wheel (and crowded out anyone else from making a bet), something will pay off.
So AI, crypto; it doesn't matter. Something will pay (and you can always threaten to crash the economy should the losses get too great).
And the government will always cover the losses if you're big enough. I know we're at a point where it's gonna suck when the government doesn't bail out these shitbirds, but at some point they're going to have to be allowed to fail in spectacular fashion. Maybe the government could help out the people that didn't create the mess when that happens, but I wouldn't hold my breath on that one. Little person hurts = fuck you assholes, shoulda planned better. Big money intentionally tanks entire economy = oh, poo
Re: (Score:1)
How can this be maintained?
How long is always the question and it's very difficult to determine from 50,000 feet, which is where the economists we listen to are looking from because they are paid and they hang out with other people with jobs.
Down on the ground it looks like it can't last that much longer because I'm hearing from more and more people who are employed but can't pay their bills.
As for Nvidia, there will be always a market for GPUs.
Yes, but the consumer market can shrink an awful lot if people don't have disposable income.
Re: How can this be maintained? (Score:2)
yeah, I wonder. I am thinking of selling tech stock at this point and diversify. This feels way too much like bullshit.
Are the companies selling a service or itself? (Score:1)
Remember the beginning of the AI hype? Companies like Nvidia were counted out and were supposed to have folded or been absorbed by a more "major" player by now. Why? Because they were "too service oriented" which had to be the dumbest thing I ever heard. So this meant that the companies that told the world "you want this because you look smart using it....", the ones that provide no unique product past that were naturally supposed to be the industry leaders by now.
bubble (Score:2)
It's a bubble.