LOL. The SEC, Nasdaq, and S&P board are all openly colluding to maximize both the bubble and forced liquidation at the top onto retail investors and retirement accounts. S&P changing its rules to list these companies at the precise moment that insiders are allowed to start dumping their stock is the most egregious aspect of this very, very obviously rigged game.
The Treasury Dept and Federal Reserve Bank in the 1990s did the same thing for derivatives trading as Wall Street argued that it could regulate itself and the market would make the corrections (Alan Greenspan). Proved very wrong. The financial market regulations are run to protect the rich with backing from Congress.
This isn't even "protecting the rich." It's reengineering the rules to force index funds to buy these stocks at inflated valuations. It's just nakedly transferring wealth from the public markets to VC and institutional insiders.
Given the security issues of agentic AI, there are just two things obvious:
1. The number of use cases for autonomous agents — beyond the typical ones of AI coding and first-level customer support — will be relatively small.
2. The scope of these use cases will be significantly smaller than what most of the frontier labs are selling you today.
Both issues make the hyperscalers' current datacenter funding approach look increasingly not grounded in financial "best practices." (Just look at Google's stock losses over the last 24 hours.)
You're right that there's no moat in the models. But no moat has never meant no money. It just means the value moves downstream. The railroads went bankrupt and the towns they connected boomed. Cheap bandwidth wiped out the fibre companies and made Google. So the money being torched at the model layer might not be a misallocation. It might be subsidising a cheap input for whoever builds on top. The bagholder and the winner are both real, they're just not the same person.
But there’s no current way to profit off the AI infrastructure buildout even if the data centers could be picked up for free. Someone has to pay for the energy consumption, and inference is getting more expensive, not less.
Chips actually do not physically wear out. They just become technically obsolete when new generation is put on the market providing more compute power in the same space and at better energy efficiency. GPUs / AI accelerators are in general less reliable than CPUs, so they have higher rate of failures, but this is not because those chips would be wearing out.
Whether this happens before their technical obsolescence depends on whether / when they actually get deployed in a data center. Given the delays and cancellations for DC builds this year, that isn't a hypothetical question.
One potential moat is users' chat histories with a particular model. The model develops a depth of information about the user that can make the interaction more useful. Eventually the user doesn't want to leave the history behind and they're to some degree locked in.
It's arguably a deeper moat than Google Search ever had.
A new tech appears that will change the world. Investors know it, and greed takes over, everybody wants to get rich. Valuations grow much faster than the technology, until everybody realizes pric s are crazy and valuations crash. Meanwhile, the technology continues its steady ascent and eventually (in a few years) does change the world.
toolate: I guess--and it's only a guess--it's time to reinitiate public hangings--of those convicted of contributing to the nefarious elements we all seem to be aware of.
How to reign in a Putin/Trump/or China-like "surveillance state" is anyone's guess.
"Trump Signs Executive Order Seeking Oversight of A.I. Models/The order, which signaled a shift from the hands-off approach the White House had previously taken toward A.I., followed debates over how to gain control of A.I. models without disrupting innovation."
President Trump at the White House last week. Credit...Doug Mills/The New York Times/Sheera Frenkel Tripp Mickle/By Sheera Frenkel and Tripp Mickle
Reporting from San Francisco/June 2, 2026k/Updated 12:50 p.m. ET
"President Trump signed an executive order on Tuesday that asked technology companies to give the government oversight of new artificial intelligence models before releasing them to the public, a shift for an administration that had promoted a hands-off approach to the powerful technology . . . . 'Advanced A.I. capabilities make our nation stronger, but also introduce new national security considerations that require coordinated action across executive departments and agencies,' the order said.
This is a developing news story. Check back for updates."
They're forming a commission to decided what the threshold is for a powerful model to be considered "covered." They're going to use "scanning" - as if that works.
Then if they decided the model is powerful enough, there is a thirty-day waiting period for the company providing the model to wait until the government analyzes the model. No word on how that's to be done, but if it's not adversarial threat modeling, it's a waste of time.
Bottom line: This is a scam to enable the Deep State to get its hands on models to use for espionage against other countries and surveillance of the US population.
richardstevenhack: It's no surprise to me or probably to others here (I hope) that our present government is already one of the, if not THE worst bad actor in the world right now.
However, with our still partly democratic power situation up in the air, and as we undergo changes going forward, I'm also sure that Bernie Sanders' similar (but differently founded) proposals are NOT THAT. If so, then as you also probably understand, it will depend on who has the reigns if and when such a "commission" occurs that has any hope of actually doing the job.
Though some kind of well-thought-out guardrail situation is sorely needed, e.g., as a transparent commission with interactive oversight by good and well-informed actors, for the time being, gaslighting rather than good actors is all we can expect.
In other words, though a commission of sorts has a great sound to it, that's what it's intended to sound like, by our present power group of bad actors. They always need a shining horse to ride in on so it's easier to screw everyone concerned once they get control.
richardstevenhack: Sanders is not a nonconscious light pole--he talks and talks and talks--along with all the others and has a grip on the ideas that enliven lots of people regardless of party or the size of their bank accounts. And it's because of communications, things besides sxxt happen.
I don't share your pessimism, though I understand it. And I know nothing good happens without help from all of us. I also know this--lots goes forward that does not get out to the public; and neither you nor I nor others here are the only one's who understand something about 'bad actors' and the influence of money and power. But unlike the song, one thing we all can know: that's not all there is.
But that's the key. What you/we don't see today will be what happens tomorrow. How or whether we like that outcome depends on what we do, and how open we are to engaging with others who care, and then we get on with finding those good new solutions.
When Microsoft managed to improve Bing using A.I., alarms went off at Google. And rightly so. Google has a near monopoly on search, and hence rakes in billions of advertising money. Why risk that?
And why did you get stuck assuming the investors in A.I. will only train models or only rent them to others for inference? Do you have any idea how much work can be automated using A.I.? That's the actual gold mine.
In 1995 people were laughing at Amazon, an online bookstore. Look at what it is now, then consider how near-sighted those laughing people were, and then look in the mirror.
And why quote Brad Delong to predict the economic impact of transformative technology? Does he have a good track record? As good as Paul Krugman, who in 1998 predicted that the Internet's impact on the economy would be no greater than the fax machine's, and to this day claims that because the economic impact of the internet is not visible in our GDP it must not exist?
In 1995, people laughed at Amazon because it was an online bookstore. Arguably, they were correct to do so, because in 2026, Amazon is a very successful company precisely because it is no longer an online bookstore, but rather rents out physical computers (like IBM), serves as a major delivery provider (like FedEx), owns grocery stores (like Whole Foods Market), has a film division (like Disney) and much more. In point of fact, Amazon is really a bunch of different companies under a single corporate umbrella. If you had asked someone in 1995 whether a company that was IBM, FedEx, Walmart and Disney all rolled into one could be successful, they probably wouldn't have laughed, but rather would have said "absolutely." In other words, Amazon's success seems to have very little to do with the merits of online bookstores over physical bookstores. (Actual online bookstores have much more modest success).
I'm less certain about the effect of the Internet on the economy, but looking at whether it has actually increased GDP does not seem like an unreasonable way of measuring it. Of course, production can be substituted, even if it has not actually increased, but it's not unreasonable to consider substitution not to be an increase in economic productivity. If text messages replace some percentage of phone calls, or streaming video replaces some percentage of movie theater sales, the economy hasn't precisely been revolutionized. I do think there are some cases where the existence of the Internet must have led to actual economic productivity—surely some jobs benefit from being able to look things up faster, after all—but it's a question that does require more careful consideration than just naively equating societal changes with economic growth.
So I would say that people might have been more correct than you assume. Of course, they might also not have been entirely correct, so perhaps there is something to be said for not simply listening to the skeptics, but instead of advocating for balance and cautious thinking, you seem to be advocating for uncritical acceptance. Nearly everything you post is about how no one should worry about safety, there will be so much money, transformative, gold mine, etc. Of course...who knows whether "you" are really "you" anymore, these days....
My main point is that people try predict the future while locked into today's perspective. Most people simply cannot imagine new opportunities.
Textbook example: When factories first adopted electricity in the late 19th and early 20th centuries, they merely swapped their massive steam engines for large electric motors without changing their floor plans. The true productivity boom only occurred *decades later* when factories were completely redesigned around the new technology.
When you leave out AWS, Amazon still easily outcompetes brick-and-mortar stores because of efficiencies and scale. It's the dominant sales outlet. It defines what products we consumers see, and hence influences our purchases. None of this was realized by the people laughing back in 1995.
Economists tend to be so locked into their models that they deny reality before they adjust their model. An example - payments. Next time you buy something online to get it delivered at your doorstep think about the banking system, how different you can spend your income compared to your grandfathers (who got it in cash in envelopes), and then deny the productivity of electronic payment systems. If I were to estimate increase in productivity I'd try estimate the cost 50 years ago of all the goods and services we enjoy today for say $1000.
When microsoft started to force co pilot, I forced out microsoft in favor of Linux. I also deleted all of my Office Products on my Mac, after disabling that AI, for open source. I automatically disable AI (artificial intelligence) on any browser. The only interaction with AI that will be left for me will be the same as bitcoin, higher utility bills.
LLM tech has become the market darling of what is likely to prove to be the most extreme financial mania in our lifetimes, hence the associated sector lunacy and delusion. The inevitable financial collapse which follows (all financial manias end in financial collapse) can be expected to terminate LLM tech financial recklessness. After the collapse we'll find out which applications of the technology prove durable and which entities survive.
Ultimate timing of inevitable financial collapse is only knowable in hindsight however there are certainly indicators to follow. Among the best, IMHO, is the yield curve. Correlation with inversion is often mistakenly viewed as the key but take a look at history, uninversion correlates most closely with financial mania tops, though not perfectly. We are currently in the process of uninverting rather aggressively following a period of extreme monetary accommodation hence amid extreme systemic monetary/fiscal distortions.
Human nature doesn't change. People make choices based on their subjective valuations and prices. When they come to understand they've been misled by artificially depressed rates of interest, the most important set of prices in an economy, they change their choices and highly leveraged financial structures characteristic of financial mania peaks collapse...always. Financial cycles really aren't mysterious. They play out in fundamentally the same fashion time after time for a reason.
It always seemed clear that there was no moat, even back in 2023! I’ll also note, Gary, that this comment section gets way more traffic, another sign of more awareness of your work.
Are the obstacles to building data centers in the U.S. keeping the AI bubble smaller than it otherwise would be? Hooray for red tape?! Or is the planned buildout already included in the bubble?
Google doesn't need a purely technical moat to win with LLMs, even though they probably possess the deepest technical machine learning and AI expertise in industry, aided by more data and compute than anyone else, and a hugely profitable core business model that allows them to buy market share and competitors. They are actually perfectly happy commoditizing LLMs in a way that helps sell cloud services. They do this by making LLMs easy to adopt and use and creating a loyal engineering community around their tooling and infrastructure. This is standard procedure for Google; sell premium products where you can, commoditize the rest. They gave away Android to extend their search dominance to mobile, they gave away Chrome and sign-on services to help build an integrated ecosystem of services across desktop and mobile, giving them constant and detailed access to people's information wherever they are. They give away their office suite to get access to even more data and lock users into their ecosystem. They constantly provide free tools and training for engineers so that they become dependent on Google Cloud to solve problems and remain marketable. Finally, big tech, being VC-forged, has always been winner take all, at least aspirationally; in practice, oligopoly is safer than monopoly and almost as profitable.
Our AI frenzy is based almost 100% on love of money and almost not at all on love of people. That is insanity in its purest form.
Wishful thinking is insane. Mass wishful thinking is destroying a civilization.
Half of America being forced to own some SpaceX! Not good for real world bank accounts that pay rent.
History is made by events. LLM AI is not driving events.
Executives make decisions and events follow. They probably mostly snicker at LLM AI.
LLM AI is almost 100% paper shuffling. It's not changing the real world economy in any significant way. It never will or the real world will decompose.
For the record, AGI never can either.
If we had one now, where would it store its immense world view that was greater than the mash up of 8 billion world views that allow human intelligence to produce so much?
How would it gain the trust of decision makers?
Would it hallucinate on critical matters?
Let's block that path before we succumb to the next Pied Piper who is going to form a public purpose corporation to protect us from AGI.
The irony is that the quote from Bain's report reads EXACTLY like it was written by Claude lol
A tiny example of why this stuff isn’t going away.
It was. Run it through GPTZero or Pangram
People are incredibly lazy. It shows how few people actually like what they do.
Or care what they do
Zinnngggg. These posts keep getting spicier. Will it be fines, jail time or a “whoops, our bad” for these psychopath CEO’s? 🤪
LOL. The SEC, Nasdaq, and S&P board are all openly colluding to maximize both the bubble and forced liquidation at the top onto retail investors and retirement accounts. S&P changing its rules to list these companies at the precise moment that insiders are allowed to start dumping their stock is the most egregious aspect of this very, very obviously rigged game.
The Treasury Dept and Federal Reserve Bank in the 1990s did the same thing for derivatives trading as Wall Street argued that it could regulate itself and the market would make the corrections (Alan Greenspan). Proved very wrong. The financial market regulations are run to protect the rich with backing from Congress.
This isn't even "protecting the rich." It's reengineering the rules to force index funds to buy these stocks at inflated valuations. It's just nakedly transferring wealth from the public markets to VC and institutional insiders.
What is the SEC doing?
Did, not doing. It approved Nasdaq's listing requirement rule change that enabled this shitshow.
I see, thanks for the answer
None of the above. It will be yachts and Islands
Yachts and islands for CEOs, more misery for the precariat.
Sadly, you’re mostly like correct 🫤
Argentina, anyone?
Given the security issues of agentic AI, there are just two things obvious:
1. The number of use cases for autonomous agents — beyond the typical ones of AI coding and first-level customer support — will be relatively small.
2. The scope of these use cases will be significantly smaller than what most of the frontier labs are selling you today.
Both issues make the hyperscalers' current datacenter funding approach look increasingly not grounded in financial "best practices." (Just look at Google's stock losses over the last 24 hours.)
You're right that there's no moat in the models. But no moat has never meant no money. It just means the value moves downstream. The railroads went bankrupt and the towns they connected boomed. Cheap bandwidth wiped out the fibre companies and made Google. So the money being torched at the model layer might not be a misallocation. It might be subsidising a cheap input for whoever builds on top. The bagholder and the winner are both real, they're just not the same person.
The real winner is the surveillance state
But there’s no current way to profit off the AI infrastructure buildout even if the data centers could be picked up for free. Someone has to pay for the energy consumption, and inference is getting more expensive, not less.
I’m all ears.
Yes, but the need for training never goes away and the chips don’t have the kind of infrastructure advantages as they wear out.
Chips actually do not physically wear out. They just become technically obsolete when new generation is put on the market providing more compute power in the same space and at better energy efficiency. GPUs / AI accelerators are in general less reliable than CPUs, so they have higher rate of failures, but this is not because those chips would be wearing out.
My (admittedly limited) understanding is that they do in fact wear out, physically. Under heavy data center use that can take between 1 and 3 years:
https://www.tomshardware.com/pc-components/gpus/datacenter-gpu-service-life-can-be-surprisingly-short-only-one-to-three-years-is-expected-according-to-unnamed-google-architect
Whether this happens before their technical obsolescence depends on whether / when they actually get deployed in a data center. Given the delays and cancellations for DC builds this year, that isn't a hypothetical question.
https://deadstack.net/cluster/report-half-of-planned-u-s-data-center-builds
One potential moat is users' chat histories with a particular model. The model develops a depth of information about the user that can make the interaction more useful. Eventually the user doesn't want to leave the history behind and they're to some degree locked in.
It's arguably a deeper moat than Google Search ever had.
This post's title rang back a classic novel by Chinua Achebe (Google it or ask your AI ;)
i actually used the title in homage explicitly for a book chapter once! here it was subconscious
But, if you take a step back, the title of Achebe's book was itself a reference to the famous poem by WB Yeats. "The Second Coming"
Things fall apart; the centre cannot hold;
Mere anarchy is loosed upon the world,
Thanks for reminding us.
I thought that rang a bell.
A new tech appears that will change the world. Investors know it, and greed takes over, everybody wants to get rich. Valuations grow much faster than the technology, until everybody realizes pric s are crazy and valuations crash. Meanwhile, the technology continues its steady ascent and eventually (in a few years) does change the world.
Nothing new here...
I am starting to read all AI and money articles in Bugs Bunny voice: What a rube. What a maroon!
But the path towards a surveillance state and minting a boatload of billionaires seems pretty clear to me
toolate: I guess--and it's only a guess--it's time to reinitiate public hangings--of those convicted of contributing to the nefarious elements we all seem to be aware of.
How to reign in a Putin/Trump/or China-like "surveillance state" is anyone's guess.
edited that for you How to reign in a Putin/Trump/China "surveillance state" is anyone's guess.
This just in from the New York Times:
"Trump Signs Executive Order Seeking Oversight of A.I. Models/The order, which signaled a shift from the hands-off approach the White House had previously taken toward A.I., followed debates over how to gain control of A.I. models without disrupting innovation."
President Trump at the White House last week. Credit...Doug Mills/The New York Times/Sheera Frenkel Tripp Mickle/By Sheera Frenkel and Tripp Mickle
Reporting from San Francisco/June 2, 2026k/Updated 12:50 p.m. ET
"President Trump signed an executive order on Tuesday that asked technology companies to give the government oversight of new artificial intelligence models before releasing them to the public, a shift for an administration that had promoted a hands-off approach to the powerful technology . . . . 'Advanced A.I. capabilities make our nation stronger, but also introduce new national security considerations that require coordinated action across executive departments and agencies,' the order said.
This is a developing news story. Check back for updates."
I read the White House readout.
It's a joke.
They're forming a commission to decided what the threshold is for a powerful model to be considered "covered." They're going to use "scanning" - as if that works.
Then if they decided the model is powerful enough, there is a thirty-day waiting period for the company providing the model to wait until the government analyzes the model. No word on how that's to be done, but if it's not adversarial threat modeling, it's a waste of time.
Bottom line: This is a scam to enable the Deep State to get its hands on models to use for espionage against other countries and surveillance of the US population.
Plus corporate bribes to Trump and his family.
richardstevenhack: It's no surprise to me or probably to others here (I hope) that our present government is already one of the, if not THE worst bad actor in the world right now.
However, with our still partly democratic power situation up in the air, and as we undergo changes going forward, I'm also sure that Bernie Sanders' similar (but differently founded) proposals are NOT THAT. If so, then as you also probably understand, it will depend on who has the reigns if and when such a "commission" occurs that has any hope of actually doing the job.
Though some kind of well-thought-out guardrail situation is sorely needed, e.g., as a transparent commission with interactive oversight by good and well-informed actors, for the time being, gaslighting rather than good actors is all we can expect.
In other words, though a commission of sorts has a great sound to it, that's what it's intended to sound like, by our present power group of bad actors. They always need a shining horse to ride in on so it's easier to screw everyone concerned once they get control.
I saw Sander's proposal.
My only comment on that is: good luck with that, Bernie!
It ain't gonna happen.
Not unless Sanders gets elected President - and THAT ain't gonna happen, either
The moneyed interests run the US and they have two rules:
1) Keep the money up.
2) Keep the people down.
Short of taking out those moneyed interests - and I don't see anyone in the US with either the brains or the balls to do that - nothing will change.
That is, until they get the US into WWIII.
Then Russia and China will make the change.
As in, no more US.
I don't see any other way this ends.
richardstevenhack: Sanders is not a nonconscious light pole--he talks and talks and talks--along with all the others and has a grip on the ideas that enliven lots of people regardless of party or the size of their bank accounts. And it's because of communications, things besides sxxt happen.
I don't share your pessimism, though I understand it. And I know nothing good happens without help from all of us. I also know this--lots goes forward that does not get out to the public; and neither you nor I nor others here are the only one's who understand something about 'bad actors' and the influence of money and power. But unlike the song, one thing we all can know: that's not all there is.
But that's the key. What you/we don't see today will be what happens tomorrow. How or whether we like that outcome depends on what we do, and how open we are to engaging with others who care, and then we get on with finding those good new solutions.
Have you forgotten already?
When Microsoft managed to improve Bing using A.I., alarms went off at Google. And rightly so. Google has a near monopoly on search, and hence rakes in billions of advertising money. Why risk that?
And why did you get stuck assuming the investors in A.I. will only train models or only rent them to others for inference? Do you have any idea how much work can be automated using A.I.? That's the actual gold mine.
In 1995 people were laughing at Amazon, an online bookstore. Look at what it is now, then consider how near-sighted those laughing people were, and then look in the mirror.
And why quote Brad Delong to predict the economic impact of transformative technology? Does he have a good track record? As good as Paul Krugman, who in 1998 predicted that the Internet's impact on the economy would be no greater than the fax machine's, and to this day claims that because the economic impact of the internet is not visible in our GDP it must not exist?
In 1995, people laughed at Amazon because it was an online bookstore. Arguably, they were correct to do so, because in 2026, Amazon is a very successful company precisely because it is no longer an online bookstore, but rather rents out physical computers (like IBM), serves as a major delivery provider (like FedEx), owns grocery stores (like Whole Foods Market), has a film division (like Disney) and much more. In point of fact, Amazon is really a bunch of different companies under a single corporate umbrella. If you had asked someone in 1995 whether a company that was IBM, FedEx, Walmart and Disney all rolled into one could be successful, they probably wouldn't have laughed, but rather would have said "absolutely." In other words, Amazon's success seems to have very little to do with the merits of online bookstores over physical bookstores. (Actual online bookstores have much more modest success).
I'm less certain about the effect of the Internet on the economy, but looking at whether it has actually increased GDP does not seem like an unreasonable way of measuring it. Of course, production can be substituted, even if it has not actually increased, but it's not unreasonable to consider substitution not to be an increase in economic productivity. If text messages replace some percentage of phone calls, or streaming video replaces some percentage of movie theater sales, the economy hasn't precisely been revolutionized. I do think there are some cases where the existence of the Internet must have led to actual economic productivity—surely some jobs benefit from being able to look things up faster, after all—but it's a question that does require more careful consideration than just naively equating societal changes with economic growth.
So I would say that people might have been more correct than you assume. Of course, they might also not have been entirely correct, so perhaps there is something to be said for not simply listening to the skeptics, but instead of advocating for balance and cautious thinking, you seem to be advocating for uncritical acceptance. Nearly everything you post is about how no one should worry about safety, there will be so much money, transformative, gold mine, etc. Of course...who knows whether "you" are really "you" anymore, these days....
My main point is that people try predict the future while locked into today's perspective. Most people simply cannot imagine new opportunities.
Textbook example: When factories first adopted electricity in the late 19th and early 20th centuries, they merely swapped their massive steam engines for large electric motors without changing their floor plans. The true productivity boom only occurred *decades later* when factories were completely redesigned around the new technology.
When you leave out AWS, Amazon still easily outcompetes brick-and-mortar stores because of efficiencies and scale. It's the dominant sales outlet. It defines what products we consumers see, and hence influences our purchases. None of this was realized by the people laughing back in 1995.
Economists tend to be so locked into their models that they deny reality before they adjust their model. An example - payments. Next time you buy something online to get it delivered at your doorstep think about the banking system, how different you can spend your income compared to your grandfathers (who got it in cash in envelopes), and then deny the productivity of electronic payment systems. If I were to estimate increase in productivity I'd try estimate the cost 50 years ago of all the goods and services we enjoy today for say $1000.
But if you leave out AWS, AMZN will be valued a lot lower.
Hey Marc, wanna buy some NFTs? How about a stake in the Metaverse?
When microsoft started to force co pilot, I forced out microsoft in favor of Linux. I also deleted all of my Office Products on my Mac, after disabling that AI, for open source. I automatically disable AI (artificial intelligence) on any browser. The only interaction with AI that will be left for me will be the same as bitcoin, higher utility bills.
LLM tech has become the market darling of what is likely to prove to be the most extreme financial mania in our lifetimes, hence the associated sector lunacy and delusion. The inevitable financial collapse which follows (all financial manias end in financial collapse) can be expected to terminate LLM tech financial recklessness. After the collapse we'll find out which applications of the technology prove durable and which entities survive.
Ultimate timing of inevitable financial collapse is only knowable in hindsight however there are certainly indicators to follow. Among the best, IMHO, is the yield curve. Correlation with inversion is often mistakenly viewed as the key but take a look at history, uninversion correlates most closely with financial mania tops, though not perfectly. We are currently in the process of uninverting rather aggressively following a period of extreme monetary accommodation hence amid extreme systemic monetary/fiscal distortions.
Human nature doesn't change. People make choices based on their subjective valuations and prices. When they come to understand they've been misled by artificially depressed rates of interest, the most important set of prices in an economy, they change their choices and highly leveraged financial structures characteristic of financial mania peaks collapse...always. Financial cycles really aren't mysterious. They play out in fundamentally the same fashion time after time for a reason.
It always seemed clear that there was no moat, even back in 2023! I’ll also note, Gary, that this comment section gets way more traffic, another sign of more awareness of your work.
And the AIpologist arguments are getting way more desperate.
AI profitability will come right around the time McCarthy's "few simple words" refuting Hubert Dreyfus are published, i.e., never.
Take away the market hype, anecdotal evidence, and Survivor Bias and you people have squat.
Maybe even (big) short and squat
Are the obstacles to building data centers in the U.S. keeping the AI bubble smaller than it otherwise would be? Hooray for red tape?! Or is the planned buildout already included in the bubble?
Even if shovels aren’t in the ground the capital is committed; the contracts are signed; the debt is issued.
Google doesn't need a purely technical moat to win with LLMs, even though they probably possess the deepest technical machine learning and AI expertise in industry, aided by more data and compute than anyone else, and a hugely profitable core business model that allows them to buy market share and competitors. They are actually perfectly happy commoditizing LLMs in a way that helps sell cloud services. They do this by making LLMs easy to adopt and use and creating a loyal engineering community around their tooling and infrastructure. This is standard procedure for Google; sell premium products where you can, commoditize the rest. They gave away Android to extend their search dominance to mobile, they gave away Chrome and sign-on services to help build an integrated ecosystem of services across desktop and mobile, giving them constant and detailed access to people's information wherever they are. They give away their office suite to get access to even more data and lock users into their ecosystem. They constantly provide free tools and training for engineers so that they become dependent on Google Cloud to solve problems and remain marketable. Finally, big tech, being VC-forged, has always been winner take all, at least aspirationally; in practice, oligopoly is safer than monopoly and almost as profitable.
“The technology is wowerful but unreliable”
Fixed.
Thanks for the wrap up Gary.
Our AI frenzy is based almost 100% on love of money and almost not at all on love of people. That is insanity in its purest form.
Wishful thinking is insane. Mass wishful thinking is destroying a civilization.
Half of America being forced to own some SpaceX! Not good for real world bank accounts that pay rent.
History is made by events. LLM AI is not driving events.
Executives make decisions and events follow. They probably mostly snicker at LLM AI.
LLM AI is almost 100% paper shuffling. It's not changing the real world economy in any significant way. It never will or the real world will decompose.
For the record, AGI never can either.
If we had one now, where would it store its immense world view that was greater than the mash up of 8 billion world views that allow human intelligence to produce so much?
How would it gain the trust of decision makers?
Would it hallucinate on critical matters?
Let's block that path before we succumb to the next Pied Piper who is going to form a public purpose corporation to protect us from AGI.