Google Boss Says Trillion-Dollar AI Investment boom Has 'Elements of Irrationality' (bbc.com) 56
Every company would be affected if the AI bubble were to burst, the head of Google's parent firm Alphabet has told the BBC. From the report: Speaking exclusively to BBC News, Sundar Pichai said while the growth of artificial intelligence investment had been an "extraordinary moment", there was some "irrationality" in the current AI boom. It comes amid fears in Silicon Valley and beyond of a bubble as the value of AI tech companies has soared in recent months and companies spend big on the burgeoning industry.
Asked whether Google would be immune to the impact of the AI bubble bursting, Mr Pichai said the tech giant could weather that potential storm, but also issued a warning. "I think no company is going to be immune, including us," he said. In a wide-ranging exclusive interview at Google's California headquarters, he also addressed energy needs, slowing down climate targets, UK investment, the accuracy of his AI models, and the effect of the AI revolution on jobs.
Asked whether Google would be immune to the impact of the AI bubble bursting, Mr Pichai said the tech giant could weather that potential storm, but also issued a warning. "I think no company is going to be immune, including us," he said. In a wide-ranging exclusive interview at Google's California headquarters, he also addressed energy needs, slowing down climate targets, UK investment, the accuracy of his AI models, and the effect of the AI revolution on jobs.
Re: (Score:3)
This is a decent point, though one supposes the rush to build datacentres would slow further, so it won't all be gravy for the hardware companies either.
There comes a time where there has to be some actual utility for the software running on the hardware that is there however, because a significant amount of what it is being used for now quite often has zero, or negative utility itself. But it may mean some people are going to get access to compute power cheaper than they may have done previously once the
Re: Hardware will be fine (Score:1)
"a significant amount of what it is being used for now quite often has zero, or negative utility itself"
Is utility in your eyes alone, or the eyes of all beholders?
If people pay attention to AI but you don't like it, does that mean it has no utility in general, or are you saying more about yourself than utility?
Re: (Score:2)
I think he measn utility in the sense of economics, as in value for the companies. So not in the eyes of all beholders, but in the eyes of the shareholders. The major current problem for AI implementations is that while there are cases where it is useful (sometimes even highly so), it's not profitable because the cost to develop, train and run the models vastly exceeds the amount of money the providers are getting from it. Image/video generation is
Re: (Score:2)
That's the core problem with the current trend. Not that AI can't be useful, but that the current business models the major players are using are fundamentally broken and no-one seems to have realistic path to profitability when factoring in how fast their costs are growing. OpenAI has currently has around 1,4 trillion [techcrunch.com] (not a typo, trillions, not billions) of datacenter commitments set up for the upcoming next 8 years, and their yearly revenue is less than 2 % of that yearly (~20 billion), if we believe Altman's own figures which are probably overly optimistic, and they're nowhere close to getting enough external funding to cover those commitments.
Am I correct in saying that the potential losses here might sink entire countries, never mind corporations? Is the AI craze likely to result in multi-country depressions? Not to mention the longer-term costs - borne by almost everyone on the planet - associated with the unconscionable increase in AGW that the AI revolution represents?
Re: Hardware will be fine (Score:1)
If it wasn't AI, would it be something else, because investors love bubbles? And why would the Fed stand by and let another Great Depression happen when they can just act as a value buyer and end panics?
Re: Hardware will be fine (Score:1)
Why not include stock market gains in revenue? When Democrats call for a wealth tax, aren't they doing so?
Re:Hardware will be fine (Score:4, Insightful)
This is a decent point, though one supposes the rush to build datacentres would slow further, so it won't all be gravy for the hardware companies either.
There comes a time where there has to be some actual utility for the software running on the hardware that is there however, because a significant amount of what it is being used for now quite often has zero, or negative utility itself. But it may mean some people are going to get access to compute power cheaper than they may have done previously once the realignment starts.
It's like the railroads. Enormous fortunes were made and then lost as the railroad boom played out and then the bubble burst. When people were driving hard to push rails across the continental US, the business case for doing so wasn't there. Yes, linking the east and west coasts had some value, but not much, since there really wasn't that much on the west coast. And there was a whole lot of nothing in between. But it was obvious to everyone that when the railroads connected the coasts and opened access to the interior, there would be enormous value. What exactly, no one knew, in the sense that no one knew where all of the railroad-enabled interior cities would be constructed or what kinds of things they would do. But it was clear that there was value in access to all of that land and that someone would do something with it.
On the other hand, realizing that value didn't happen right away. It took decades for all of the land granted to the railroads to become really valuable, because it wasn't valuable until people came and built farms, dug mines, established ranches and generally built lives and industry. The return on that massive investment was there... but it came far too late for most of the people that invested it. Lots of bankruptcies resulted, and others swooped in and snapped up the resources for bargain-basement prices, and they're the ones who became incredibly wealthy (well, they and the ones who supplied the steel, e.g. Carnegie).
It's been the same with pretty much every technology-driven bubble. Remember telecom/dot-com bubble in the 90s, with all of the "dark fiber" that was laid everywhere? Bankruptcies and consolidations resulted, and all of that fiber got lit up and used. That bubble built the Internet, and huge fortunes were made as a result -- the top half-dozen most valuable companies on the planet are all a direct result.
OpenAI and Anthropic are betting that this time will be different, that the payoff will come fast enough to pay back the investment. Google is betting this somewhat, too, but Google has scale, diversity and resources to weather the bust -- and might be well-positioned to snap up the depreciated investments made by others. If history is any guide, OpenAI and Anthropic are wrong. But, then again, AI is fundamentally different from every other technology we've created.
Re: Hardware will be fine (Score:1)
What if the railroad crash was simply an irrational panic of bank lenders spreading by contagion to fundamentally good assets (by any mainstream reckoning)?
What if you are trying to shoehorn irrational panics into a narrative where eonomic rationality is the norm, but another narrative can easily be spun about irrational market responses that wildly exaggerate little blips and lead bankers to call in notes in fear that are perfectly good, economically speaking?
What if Greenspan caused dotcom by framing irra
Re: (Score:2)
I think this makes sense. OpenAI pays Google for compute, Google uses that to build more DC capacity. If OpenAI goes bankrupt, Google keeps the compute (and whatever they've already be
Re: Hardware will be fine (Score:2)
Re: (Score:2)
Where nVidia would be exposed would be is companies have bought significant quantities of their chips, but go bust when the bubble pops without settling their invoices from nVidia first. Best case that ties up nVidia's cash while the bankruptcy process happens and they get a significant slice of the outstanding f
Re: (Score:2)
LOL as if cryptocurrencies are anything other than manufactured scarcities of literally no value. This is what you get when you employ AI to manufacture prose in support of fraud.
Re: Hardware will be fine (Score:1)
"A system built on perpetual extraction, accelerating debt, and manufactured scarcity cannot sustain genuine transparency or monetary autonomy"
How long has Japan been transparently printing money to sustain itself, busting shorts who thought like you?
Re: (Score:3)
...There is huge money to be made ultimately, once drug companies, like GSK, BASF, Dupont, 3M, etc use it to advance chemistry and materials.
This is a point people keep missing. AI is not just ChatGPT and its clones. Those are just large language models. What really matters is the use of AI in doing actual work, and that has little to do with language models.
Re: Hardware will be fine (Score:1)
What if you can interface with the software doing actual work without using arcane proprietary command syntax that was the first thing somd developer thought of?
Re: (Score:1)
"once people realize the LLMs are not going to replace actual thinking persons, anywhere where the outcomes matter"
Come on, with that UID you are old enough to know better. Remember when people started calling distributed systems 'clusters', then calling internet connected clusters 'cloud', then because everyone was desperately trying to attach some sort of meaning to this buzzword is slowly started to mean internet connected clustered systems WITH AN API.
Remember how everyone who went through this evolutio
Re: (Score:1)
Well I can't argue with that. I was there too and saw a good deal of it.
I was looking at this at a macro level. At that level I think the net number of people will be around the same but you are right we will rotate out competent people who understand this stuff for a smaller handful of expert fixers, and ultimately a number if incompetents who will inherit systems they don't understand and can't maintain, and more inflexible policy band-aides.
Re:Hardware will be fine - But not HW companies (Score:2)
NVidea is the "Sun Microsystems" of this bubble. And Sun CRASHED HARD in the dot-com crash. As anyone who wanted Sun Hardware could pick it up really, really cheap at all the Bankruptcy auctions.
Sun never fully recovered.
Re: Hardware will be fine - But not HW companies (Score:1)
What if you invested in NASDAQ?
AI says: "Volatility: Sun's stock was significantly more volatile than the general NASDAQ index, with more extreme gains during the bubble and a much steeper, more permanent crash afterward.
Long-Term Trend: While the NASDAQ Composite eventually recovered and reached new all-time highs many years later, Sun Microsystems' stock never fully recovered its peak value and the company was ultimately sold at a fraction of its peak market capitalization. "
way more than some irrationality (Score:4, Interesting)
The issue is not seeing the bubble, the issue is that most of us (having zero investment and not much savings) have zero idea how to avoid the big crash. I see no way to avoid the crash. Best we can hope is cross finder whether we will lose our job or not.
Re: (Score:1)
Here is the thing, you are posting on Slashdot. Don't tell me you are not sharp enough to find a broker, and buy some long dated at the money PUTS either on the AI and AI adjacent firms or just the market over all with funds like SPY / QQQ.
You If you really had conviction about truly big enough crash for Main Street to feel it to commit 18 or 20K; you'd make enough to keep the mortgage current and food on the table for a year right there after there return of the principle.
The thing is you don't really b
Re: (Score:2)
Doing that would be stupid because we know the government will try to bail them out.
You can't make rational economic decisions when the government will interfere with the market any time it feels like it. That just turns the whole thing into a casino.
Re: (Score:1)
There is a way to play that too. Government; especially one as divided as our current one; can't do anything fast.
There will almost certainly be in the event a major market crash the idea ballots floated.
Someone like Massie or Rand Paul will threaten to be votes to derail it.
- Sell this news (in this case exercise those PUTs) because we all know after Washington does its things for a couple weeks (maybe longer) something will get done.
There is a pretty simple script here. People say you can't time the mark
Re: (Score:2)
I think we are going to see a 'correction' we go down 10% or or less from recent highs and trade sideways for a while.
Most of the big guys in AI are already down more than 10% in the past week. Today isn't looking any better. Once Nvidia reports tomorrow after the bell will we see stabilization, assuming they have good news to report.
Re: way more than some irrationality (Score:1)
Why not front-run the Fed? Buy, hold, and let the Fed keep the irrationality going forever?
Re: (Score:2)
Governments don't interfere with markets when they "feel like it"--they'll usually do it to prevent catastrophe, promote a particular public good or strategy, or (most common in the USA) because the government has been bought and paid for by the people being bailed out.
Re: (Score:3)
"you are posting on Slashdot..."
so does SuperKendall.
"..buy some long ..."
with what money?
"you'd make enough to keep the mortgage current"
what mortgage?
"... after there return of the principle" /. posters by definition has multiyear financial plans.
Sure,
Let them eat cake.
Re: (Score:3)
Re: (Score:2)
Here is the thing, you are posting on Slashdot. Don't tell me you are not sharp enough to find a broker, and buy some long dated at the money PUTS either on the AI and AI adjacent firms or just the market over all with funds like SPY / QQQ.
The market can remain irrational longer than you can remain solvent.
The better strategy, IMO, is to keep your money safe and wait for the bubble to burst, then pile in for the recovery. Where to keep money safe is a good question, though. Just holding cash might be risky if inflation comes back, and the current administration seems anxious to pump up inflation.
Re: (Score:2)
It is quite clear to everybody it is a bubble and a lot of the AI stuff is sand-castle based or vapor based... At least those of us understanding what the current crop of AI does
There's a pair of seriously bad assumptions underlying your analysis:
(1) What AI does right now is all it's going to do. Given the way capabilites have grown recently, this is a ludicrous assumption. Keep in mind that ChatGPT was launched November 30, 2022... it's less than three years old! And the reasoning models are barely a year old. There is no reason whatsoever to assume that this technology has peaked.
(2) We already know how to take full advantage of AI. Every time a new technology comes alon
Re: (Score:2)
The AI thing absolutely is a bubble, but it's not "sand-castle based or vapor based". It's very real. The problem is that the massive wave of investment is going to have to start generating returns within the next 3-4 years or else the financial deals that underpin it all will collapse. That doesn't mean the technology will disappear, it just means that the current investors will lose their shirts, other people will scoop up their assets at firesale prices, and those people will figure out how to deploy it effectively, and create trillions in economic value.
The problem is that the investors - and lenders - potentially losing their shirts include major international banks and pension funds, not just private shareholders. Recently, a J.P. Morgan analysis [tomshardware.com] estimated that at least $650 billion in annual revenue will be required to deliver mere 10% return on the projected AI spend. And already banks like Deutsche Bank [ft.com] are looking to hedge their lending exposure to AI related projects.
If the AI bubble crashes hard, it could be a repeat of the 2007 global financial cr
Re: (Score:2)
The AI thing absolutely is a bubble, but it's not "sand-castle based or vapor based". It's very real. The problem is that the massive wave of investment is going to have to start generating returns within the next 3-4 years or else the financial deals that underpin it all will collapse. That doesn't mean the technology will disappear, it just means that the current investors will lose their shirts, other people will scoop up their assets at firesale prices, and those people will figure out how to deploy it effectively, and create trillions in economic value.
The problem is that the investors - and lenders - potentially losing their shirts include major international banks and pension funds, not just private shareholders. Recently, a J.P. Morgan analysis [tomshardware.com] estimated that at least $650 billion in annual revenue will be required to deliver mere 10% return on the projected AI spend. And already banks like Deutsche Bank [ft.com] are looking to hedge their lending exposure to AI related projects.
If the AI bubble crashes hard, it could be a repeat of the 2007 global financial crisis.
Yep. That's all true even if AI is the most transformative technology ever invented, even if it generates trillions per year in economic output -- it might not do it soon enough to prevent another crash. You don't have to believe that AI is "sand-castle based or vapor based" (which it's really not) to see a big problem coming.
Re: way more than some irrationality (Score:1)
When in the history of stock markets has there not been "a big problem coming"?
Re: (Score:2)
When in the history of stock markets has there not been "a big problem coming"?
When in the history of humanity has there not been a big problem coming? Or occurring, for that matter.
Re: (Score:2)
A big issue in this bubble is how incestuous the deals are.
Example: Company A pledges to invest 1 billion dollars into Company B. Company B pledges to buy 1 billion dollars worth of compute time from Company A. Company A has increased their revenue by 1 Billion, and Company B has increased their market value by several Billion (1 Billion invested x whatever percentage of shares they gave). On paper, everybody is raking in $$, but in reality... no value is created. Add in another layer of Company A using
Re: (Score:2)
Just up the page is an article headlining this type of incestuous investment I am referring to : https://slashdot.org/story/25/... [slashdot.org]
Irrationality was used to designate the dot-coms (Score:3)
Irrationality. I remember it well. Quoting Wikipedia: "Irrational exuberance" is the phrase used by the then-Federal Reserve Board chairman, Alan Greenspan, in a December 1996 speech given at the American Enterprise Institute during the dot-com bubble of the 1990s.
Irony? Google attacks its competitors funding smh (Score:2)
A) everyone but Google.
Lack of intelligence (Score:2)
All these investors are expecting results on the intelligence front. And since that's a total fake sell, the expectations will inevitably crash.
AI isn't for you (Score:2)
It always strikes me odd that people ask the question if there are no consumers who will buy their products?
You think somebody with a billion dollars hasn't asked that question?
What if they come up with a different answer than the old one we're told Henry Ford did. (Fun fact Ford paid better not because he wan
Re: AI isn't for you (Score:1)
Will they keep suicide illegal because they want the pleasure of killing you themselves?
Re: (Score:2)
The goal is to be the last man alive, sitting on a pile of 8 billion plus skulls.
There is no other definition of 'a winner'.
Re: AI isn't for you (Score:2)
What if you are just spouting nonsense? Have you considered that?
For all the profits to materialize the important question to be answered is where the money are going to come from. If you think automating non trivial part of a work force will not have any effect... You have some learning to do.
As expected (Score:3)
Exciting new tech emerges and advances rapidly
Investors don't understand it, but are desperate for "the next big thing"
Hypemongers, futurists and pundits tell amazing stories of a fictional future
Lying salesweasels convince clueless executives to spend way too much on immature tech
Zillions of startups emerge with impressive looking but kinda fake demos
The general public misuses the tech to create slop and scams
Meanwhile, the tech continues steady progress
Oversold profitability vs functionality? (Score:2)
I'd add that we had smaller hype cycles in recent memory:
Re: Oversold profitability vs functionality? (Score:1)
What if your accuracy fetish is not shared by the majority?
Then it's a good time to be a lawyer (Score:2)
What if your accuracy fetish is not shared by the majority?
Well, we've seen this before...sloppy startups launch with sloppy code, but a good idea. They get users. They get marketshare. They get revenue...with real revenue comes real responsibility...there's a data breach and now they realize they cannot be sloppy. They cannot write their core code in Python and port it to a real language. Now they need process and release engineering and backups and rollback....all the things that slowed down their competitors.
With Vibe coding, we'll see it all over again.
An example... (Score:2)
In addition to the rest of us having raised electric rates at home... there was an article on /. just last week, about two datacenters built in Nvidia's home town... that are sitting unused, because they're waiting for electrical power to b built out.
There's also the issue of the chipmakers claiming 5 years, when the real lifetime is 3... or, as Cory Doctorow noted, if you *really* push them, 56 *days*.
Summary at 47 Comments Mark (Score:2)
Condensed into 10 bullets from the provided Slashdot thread (source: perspectives.pdf).