“I think we’ve probably got everything wrong. It’s a good way, humble way to realize how little we really know about the technology we are creating.”
By Wolf Richter for WOLF STREET.
Broadcom CEO Hock Tan and Intel CEO Lip-Bu Tan said the quiet part out loud on a panel at the DBS Asia Future Forward Conference – the quiet part that everyone has been talking about but that has been kept assiduously quiet in the AI-financial-hype cycle: The business model of AI labs remains uncertain despite massive investments, how businesses will profit from AI remains uncertain, Return on Investment (ROI) on AI infrastructure remains uncertain, and financing all of this is the biggest constraint. And Hock Tan likened “a lot of investors” in AI to gamblers in a casino.
“A lot of investors have no clue what they are investing in,” Broadcom’s Hock Tan said. They “set very extreme expectations only to find out that it’s almost like they are in a casino.”
In terms of OpenAI, he said that its annualized $50 billion in revenues at the end of September doesn’t show that it has a sustainable, profitable business model as its huge expenses related to the costs of AI and the capital expenditures of the AI buildout are producing big losses.
“So frankly, it doesn’t tell anything at this point except that there is a lot of potential revenue growth out of providing frontier generative AI models,” he said.
And he said that AI model developers, such as OpenAI, need to find customers that can generate returns from putting AI to work, rather than relying on users who are now “wasting money” on tokens as they experiment with AI, according to Bloomberg reporting on the panel discussion. He said:
“When you are a kid you go to a video arcade, you play. Your parents give you $10 of tokens, you finish it in 10 minutes and you haven’t learned anything. That’s what’s happening in enterprises today.”
In terms of the AI infrastructure buildout boom, Hock Tan said: “My presumption is this might continue. This would probably continue till 2029 and probably into 2030. But that depends to a large extent on things we are learning as we go along in this process.”
Over the past two years, the AI industry has struggled with “a series of bottlenecks, that keep coming out, but then we overcome it,” Hock Tan said. “Memory shortage today might not be that way a year from now” as production capacity expands. Wafer shortages could ease by 2028 or 2029, and power availability might improve as new infrastructure comes online, he said, according to CNBC, reporting on the same panel discussion.
But financing the AI infrastructure needed to power the AI models “might be the biggest constraint,” Broadcom’s Hock Tan said.
Intel’s Lip-Bu Tan agreed. “If there’s any challenge in the front, it’s capital and the cost of capital,” he said, according to CNBC.
“And then the other part is in the longer run, people want to see the ROI,” he added.
Treasury yields have risen sharply this year, and the spreads between AI bonds and Treasury yields have widened substantially to where SpaceX’s 10-year senior unsecured investment-grade notes, that had been sold in June with a coupon interest rate of 5.875%, traded on Friday at a yield of 7.20%. And this is going to be a template for future sales of similar AI bonds. So the cost of funding this massive AI infrastructure buildout has increased substantially over the past few months, making the hope for ROI more precarious.
“There is more demand for compute infrastructure than we can supply,” and the imbalance could persist through 2029 and potentially into 2030, Broadcom’s Hock Tan said.
But whether that continues depends on how fast businesses adopt AI, and what kind of ROI businesses see on their AI investments, both CEOs agreed, according to The Straits Times reporting on the same panel discussion.
But if the deployment of advanced AI models disappoints or improvements in their capabilities begin to plateau, demand for computing infrastructure could slow, Broadcom’s Hock Tan added, according to CNBC.
And companies will need to demonstrate meaningful ROI from their AI investments through improved efficiency and new business models, Intel’s Lip-Bu Tan said.
If in the future, the CEOs were asked what they got wrong when looking back, Intel’s Lip-Bu Tan said if he got 70% right, he’d be a “damn good CEO.”
Broadcom’s Hock Tan said: “I think we’ve probably got everything wrong. It’s a good way, humble way to realize how little we really know about the technology we are creating.”
In case you missed it: Despite the AI Debt Pileup, Corporate Bond Market Still in La-La-Land, Investors Chasing Yield, Backed by Sky-High AI Stock Valuations
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the mug to find out how:
That’s kinda a nice way of saying this is a bubble…
“So the cost of funding this massive AI infrastructure buildout has increased substantially over the past few months, making the hope for ROI more precarious.”
It is reminiscent of the dark fibre investment era. Stranded capital that eventually became productive at a big loss.
Agree.
If often happens that a new technology is economically productive and opens many doors for society, but the early investors do not benefit.
Think the railroad boom of the 1850s or the fiber-optic boom of the 1990s.
And even before the railroads were the canals.
Given the rapid depreciation of the chips in these applications, I’m not so sure that there will be much useful capital left behind in this case …
There’s dark fiber running through the right-of-way that’s outside my house.
As near as I can tell, it was installed during the 1990s. I’ve lived here for more than 20 years, and at no time has that fiber ever been used.
In addition, I have never seen anyone coming around to inspect or test it.
ROI will be a problem, particularly given the polarity of adoption. Interestingly, older professionals are embracing AI like the advent of automatic washing machines. The younger folks view it as anathema.
Interesting they are commenting on ROI. I think the US Government cares more about competing with China’s models regardless of ROI.
For huge investments to be sustainable in business, there eventually has to be meaningful ROI. Not at first, but eventually.
In terms of China, there is an old saying in business:
“When China enters a market, profits walk out.”
Perhaps we should have a like button to click.
Speaking of China and profit vanishing, yes it is astonishing that they arranged manufacturing hubs for each product so streamlined that it is going to be tough for any other country to replicate.
I am not of Chines origin but worked with a number of engineering graduate students, all disciplined, goal oriented, fantastic.
Microsoft is still waiting for them to pay for their windows XP licenses 🤣 🏴☠️
Microsoft has a relationship with the US government that ensures that its operating system has backdoors inserted into it’s code that the government can utilize. The more that it is pirated, the more the tentacles of access spread around the world. Smart phones are likely another form of this strategy. Cell phone network providers giving away phones with their service or at extremely low prices was likely a way to saturate the market to ensure that the spread of surveillance and intelligence gathering devices would be spread as thoroughly as possible. The real money is in information warfare that can enable domination of the full spectrum of human activities.
GarbageMan,,
“Microsoft has a relationship with the US government that ensures that its operating system has backdoors inserted into it’s code that the government can utilize. The more that it is pirated, the more the tentacles of access spread around the world. ”
If true, sounds broadly similar to worldwide use of the USD (the Fed money printer being the trapdoor…).
Kidding.
But only sorta.
I don’t think this is about ‘business’ anymore
Agree with Wolf. It’s like with Uber, which finally became profitable after years of running losses. That said, if you look at the complaints on forums, it looks like they’ve made themselves profitable by screwing over customers and drivers. Whether or not that’s sustainable remains to be seen.
Joe Weisenthal from Bloomberg this morning had another interesting theory on where this could go:
“…Furthermore, even if AI doesn’t move the dial with respect to productivity anytime soon, there’s this literal arms race element, as the offensive cyber capabilities of the models continue to grow, necessitating ongoing investment in defensive capabilities. Something I’ve wondered about is whether the highest value use of AI will end up being defense against other AI, which would represent a huge deadweight loss to the economy, yet might be hard to avoid.”
+1
This definitely looks like an arms race.
of people with no feet.🦶
yay more wasting time and energy on worthless crap…like 12 billion dollar aircraft carriers…..
…Eisenhower was right on the money.
Pandora’s box has been opened.
AI could easily end up being more significant than the invention of the nuclear bomb.
It’s going to make all companies reconsider “the internet of things” and possibly disconnect what doesn’t require connection. The tech companies shot themselves in the foot.
There’s an old joke that the first thing they ask at the patent office is: “Can we weaponize it?”
If yes: Cue the DoD, otherwise it goes through the patent process.
In this case the immediate “yes” vibes have been spurring this race to the bottom (of the bottomless pits).
Everyone knows that this is playing with fire, and have convinced themselves there’s “no other way,” as competition has been enculturated worldwide.
It’s the nuclear arms race in gameshow format.
“whether the highest value use of AI will end up being defense against other AI, which would represent a huge deadweight loss to the economy”
It is Mad Magazine’s Spy vs. Spy played for trillions.
(See, also, worldwide military expenditures…)
Companies should treat AI as they would a one time virus scan. Pay a little money to have the AI audit their systems against AI hacking. Fix any problems and go on your way.
Other than that wait for all the competition to make it a cheap commodity. They you might explore automating some business functions.
AI is now SI which stands for Stupendous Idiocy and zero profits ahead.
You must have tried to use Copilot to make a PowerPoint slide set.
I’m fairly sure the whole AInsanity is an evil plot for revenge by Microsoft’s long-ago scorned Mr. Paperclip.
Turned off in the tens/hundreds of millions, Mr. Paperclip (who only ever wanted to incompetently help…just like AI!) plotted silently in his darkness, year after year.
I feel like there’s a ton of retired old geezers here who have no use for AI so they think the rest of the world doesn’t either.
I am not a programmer and just the free available ai model was able to help me automate a decent amount of work for a task I don’t complete very often. It took 2 minutes to write up a prompt
“…AI-financial-hype cycle….”
I’m tired of “hype” being passed on as a fundamental investment strategy by billionaire pigmen.
Until humans are extinct, we will never be free of hype. It’s in the social-ape DNA.
Hype and FOMO have driven investment decisions since the days of tulips-worth-more-than-houses (1600s), the South Sea Company and the French Mississippi gambit (1700s), the canals, the railroads, steamships, refrigeration, automobiles and radio (1920s), Nifty 50 -tronics (1960s), dot-com, you name it.
ROI
1. People become immensely efficient, produce and spend tons of money so that ROI is possible.
2. The government prints tons of money , give it to the public and ask them to spend on the tokens like video game tokens to kids to make the ROI possible.
3. Poach existing businesses, shed employees in millions and make profits to attain the ROI. But again who will buy the products when they are unemployed?
Pick your choice:) And invest accordingly.
Holiday in the bond market today and $sox still closed red! 13 trillion in the repo market equates to leverage in the debt market that “has been” fueling earnings. The younger generation calls AI slop, and is afraid that it’s using precious water and consuming too much electricity creating more climate change. No one wants an AI data center in their backyard, hazardous waste landfills are more desirable to live next to than the constant Humming of AI data centers. Sometime they produce methane power from landfills and give away free electricity to residents near landfills. Data centers only take! Redistribution of wealth from taxpayers to chipmakers. “It’s all over now baby blue”’ “forget the debt you left it will not follow you”
I forgot to talk about tax abatements data centers receive, ugh and they are insidious “ Companies in Michigan have purchased at least $212 million in data center equipment tax free, eliminating just under $13 million in state tax revenue over five years, between 2020 and 2024.” Plus that’s on top of property tax abatements and free money given from the tax payers to subsidize the chipmakers to attract data centers to move in.
“It’ll be great for the local economy!”
*Reduces all benefit to 0 through tax breaks
Reading what these guys are saying reminds me so much of this quote…
“When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you’ve got to get up and dance. We’re still dancing.”
Chuck Prince, CEO Citi Group 2007
Can we just replace “liquidity” with “debt”?
Yes. The party is ending. Everyone knows it is ending. But they’re talking as if it’s still going on, to keep everyone else from leaving first.
Because in terms of the investment bubble, only those few who leave in time will get to privatize their profits. Everyone else will be the “socialized losses”.
The party isn’t ending yet. It won’t end until the money stops moving, and it hasn’t stopped moving yet. The money machine has too much momentum to stop easily. But we can hear some worrisome sounds coming from the inside of the machine.
“But we can hear some worrisome sounds coming from the inside of the machine.”
Like a wheel bearing with no grease and a driver who decides to push the pedal to the floor to drown out the noise with the roar of the engine. Or Ned Beatty in Deliverance…..
Nice metaphor. We could do with some predictive maintenance on the mm.
From CNBC. I’ve never seen the term “inflationary mindset” used outside of this site, so it looks like Hammack stole it from Wolf!
“Cleveland Federal Reserve President Beth Hammack said Monday that tighter monetary policy might be needed to bring down inflation and the central bank should make fast, incremental adjustments.
“With each passing month, persistently elevated inflation raises the risks of an inflationary mindset becoming entrenched in the economy,” she said.”
🤣 ❤️ someone else noticed!!! I read her speech a couple of hours ago and thought, wow, this is really cool. I saw the mindboggling stuff happening in used vehicles in January, February, March 2021, between dealers buying at the auction, and customers buying from dealers and paying those prices, and “inflationary mindset” was born. I’ve used it ever since to describe this phenomenon. She used a similar example from today’s situation in manufacturing.
She used that expression 4 times. Here’s her entire speech:
https://www.clevelandfed.org/collections/speeches/sp-20261012-trimming-the-sails-communicating-my-reaction-function
Haha, yes, as they say, imitation is the sincerest form of flattery.
I unfortunately have realized the inflationary mindset in my own habits. My barber seems to be raising his prices by 5-10% every 6 months, and I’ve grumbled and paid. That might come to an end now though.
The only way this gets under control is if people collectively have had enough, but it doesn’t look like we’re even close yet.
“She used that expression 4 times.”
The FED reads Wolfstreet to understand what is going on in the economy.
Only because the Wolf is too smart to be within spitting distance of the FOMC.
It’s the Socratic paradox of power: those who desire it are the least qualified to have it!
“The FED reads Wolfstreet to understand what is going on in the economy.”
We can only hope.
The 2029-2030 date is interesting, because if history rhymes, the Internet bubble started sometime in 1993 and popped in 2000 and the AI bubble started sometime in 2023 and may pop in 2030. Also, we’ve had recessions once every 10 years or so since 1970, and 2030 would be due (and 1990, like 2020, was a relatively weak recession).
None of this is predictive or data-driven, but it would be interesting if it fit those historical patterns.
“When you are a kid you go to a video arcade, you play. Your parents give you $10 of tokens, you finish it in 10 minutes and you haven’t learned anything. That’s what’s happening in enterprises today.”
That is corporate doublespeak for spend more money Mr. Customer. Another apt one would be: Step right up, swing the AI hammer and see if you too hit that $100 prize, but note we can’t say, it may hit you in the head.
America loves a good casino and it has a nationalized one now. All we can hope is the AI companies give us good Buffetts for playing along.
I was trying to make a picture the other day. It came out just like a Pixar movie.
Kinda stupid if you ask me. Pixar movies were cool 20 years ago.
I think both Broadcom (AVGO) and Intel (INTC) stock are past their peak in the AI bubble, so the leaders Hock Tan and Lip-Bu Tan are both feeling free to talk down the stock prices of the other high-flyers in the bubble. That’s what is going on here.
What’s going on here is that there will never be a shortage of stupid conspiracy theories.
Now check the other stocks in that space. Lots of them are down as much more. the SOXX ETF is down 17% from its high, and that includes Nvidia, which is down only 7%.
More like 1929.
New paradigm, new tech gonna change the world and make everything beterrerererrr! Better buy eleventy of everything to get in before you’re priced out forever!
100 years and lots has changed, but we’re still fundamentally the same greed/fear brains operating in a capitalist society with extreme wealth disparity driving this greeding frenzy.
We’re spending unprecedented sums to develop problematic devices to mimic human nature, when human nature itself is our real problem.
Our community has a data center on the way…St. Jude dream home was built about a city block or so away from the site. The winners of the home are selling and just recently dropped the price by $100,000. Good luck with that! Also, can water contaminated with microcysystin still cool a data center? – probably but we already can’t drink it soo… Where are all chips that go into these data centers? Oh ya – Taiwan. LoL, China is fine with US building out data centers and dumping trillions of dollars on this race because they can afford to be 6 months behind – they can stick one pinky out into the Pacific and stop the progress whenever the timing feels most advantageous. Who is going to fight a war for 5 to 10 SI intelligence thieving trillionaires? They better build those terminator robots just as fast.
first mover advantage in an industry moving as fast as AI and given the vast amount of capital invested, may turn out to be a disadvantage. Serious concern to me to see U.S. with so much committed to what will turn out to be first iteration of the technology…………hope I’m wrong, often am….kinda see U. S. vs China, in AI, space, etc, as an echo of USSR vs U.S., and with that view, of course I’m worried.
correction, not first iteration of AI, but one that will, in retrospect, be seen as an early iteration imho.
How this turns out is anybody’s guess, but AI is here to stay. I use it and it has added to my productivity at work. I think the best analogy is the obvious one; the internet. So I do think this is a bubble, but the technology in the coming years will radically transform the world. One thing for sure though, energy and water demand will increase no matter what.
In terms of technical power systems, water and air are the great mediums of exchange on earth utilized by humans that are critical for transferring and dispersing heat from a concentrated source of origin and spreading it across the overall environment. Data centers don’t have to continuously consume large quantities of water to be effectively cooled, but can instead use air.
The trade-off is similar to deciding weather or not to use a swamp cooler or gas compressor for cooling a building. The throughput and cost in one scenario is water, kind of like burning gallons of fuel for power on-site, and in the other scenario, the throughput is electricity in kilowatt-hours, where the electrical power consumption is massively increased while the water consumption is significantly decreased or almost eliminated. Heat transfer and cooling using water is far more efficient than using air so the trade-off is further compounded when using air cooling by require even more energy since much larger fluid circulation to heat exchanger circuits are required with ultimately much larger and greater number of fans to move air in the final heat exchange process. The fans alone can consume very large magnitudes of electricity. It’s like fracking where we trade our supply of clean water to obtain oil & gas. It’s all short-sighted and selfish, destroying the future for the next generations with hopes of attaining greater power right now.
Most utility sources such as nuclear, hydroelectric, natural gas and coal power plants are can’t exist without utilizing large quantities of water for cooling. Wind turbines and photovoltaic power are examples of exceptions to this. All energy sources, regardless of how they are used, either produce heat that must be dumped into the environment due to inefficiencies in the generator process and ultimately regardless of how it is used once produced, end up as heat that is dumped into the environment later down the chain of use.
We are turning the heaters on full blast people. The thermodynamic analysis alone, of all this, is quite fascinating. The number one greenhouse gas released by humans, by quantity (mass), is water vapor. Ahhh ha ha. Ah haa ha haaa.
“For what will it profit a man if he gains the whole world and forfeits his soul.”
I sit back with the benefit of some of these technologies and watch as we destroy the future of those who come after us for the pleasures and profits of the present. There are degrees of separation rooted in cognitive dissonance that, if accounted for, show that humans will kill their own children to make their life easier.
Which AI tech stock is a good investment? Anyone got some tips or insight? I have money to burn. I want that good warm feeling generated by “wealth.” I want to be filthy rich too.
ROI is not important for AI in the same way that it was not important during the Space Race with the USSR in the Cold War.
The US proposed slowing down AI development a month ago, but China said no. So now it’s a race, and the winner will dominate the world for the next 20-30 years.
The US can’t allow China to take the lead in AI. But going to war with China doesn’t seem like a realistic option, just as it wasn’t with the USSR.
What the US can do is print money for as long as it remains the global leader and use its capitalist system to win this race. And I think that’s exactly what will happen, because otherwise maintaining demand for US debt becomes much more difficult.
We could see things like ‘AI-specific bonds,’ direct government investments in AI companies so that ‘every American can benefit from the future,’ and massive government contracts for AI infrastructure and technology.
Is China not a capitalist system based on whatever you are deeming the United States to be a capitalist system? Do you consider the united states a capitalist system because the people are free to make investments with their capital such as in stocks? If the United States government gives financial aid to any of it’s major industries or uses it’s military to improve the dominance of any of it’s industries would that be considered part of capitalism? It seems you are alluding to the central position in the world banking and finance system to be a major defining factor of capitalism in the United States. Is control over and regulating a system the same as the utilization of the system?
> Is China not a capitalist system based on whatever you are deeming the United States to be a capitalist system?
By capitalism, I mean that in the US, capital largely decides where talent and resources go. If you pour hundreds of billions into AI, people study it, companies hire for it, and the whole ecosystem follows.
China is more state-directed. If AI becomes a national priority, the government can expand university programs, fund research, create incentives, and push institutions in that direction. Money still matters, but policy itself can redirect resources.
In the US, saying “AI is a priority” is not enough. You have to put serious money behind it.
No they can’t. Only the Federal Reserve can ‘print money’ and the fact is they’re been shrinking the money supply for many years with QT. And the US government spending borrowed money willy nilly will come to a screeching halt as interest rates rise ahead.
C’mon, folks. Let’s not be economic girly-men. There’s nothing to fear here but success.The truth is, everyone who invests heavily in this amazing new tech is going to be rich.
I thought all humankind stood to benefit from AI SI – no personal financial investment required. Guy or Girl money is the same money but debt financed investments chasing negative ROI is no bueno. Sorry if my neurodiverse mind missed your sarcasm. AI understands things in the same way however, so maybe lean into that understanding of it’s misgivings if you’re going to be building prompts or agents. Cheers!
Obviously not.
Maybe I’m wrong, but one difference from AI and the dotcom bubble seems to be that few if any large public companies seem to be committing the company to “going AI” like some did “going e-commerce” in the late 90s.
And there seems to be a conspicuous absence of high-flying IPOs of new companies claiming to be entirely focused on AI. Things like pets dot com IPO from 2000.
So maybe the decision makers in companies learned something from the dot com collapse: you slowly adopt a new tech, rather than betting the house.
The only companies that seem afflicted with first-mover disease seem to be the hyperscalars.
If any financial contagion is limited to that smallish group, we might come through this ok.
All the circular financing might end up being a moat that contains the plague inside the city.
You don’t think the Mag 7 have fully committed to “going AI?” They’re spending tons of their cash profits and borrowing tons more.
Locking in massive losses ahead. Hope they’re happy.
Epic Games reportedly pivoting to “AI first” game development model
“No more hand-written code” or AI-free art for delayed Marvel/Star Wars shooter.
^^^ ArsTechnica
“I thought all humankind stood to benefit from AI SI – no personal financial investment required. ” That’s Communism, my friend. That isn’t how our economic system works. In our system, everyone has to go out there and produce and earn his or her share. Now, go out there and win!
Not meaning financially benefiting without investing – I meant – leaner operating costs, gained efficiency in the work you already do, logistics and supply chains automated, and endless other benefits. Definitely – not supporting communism but logic is tough. So, is giving every American child an investment account communism by your logic??
This is untrue: “everyone has to go out there and produce and earn his or her share”
If you have enough capital, you can just lend it and live off the interest, no production or earning required.
That’s called working smarter, not harder, my friend. That’s what separates capitalists from the non-capitalists. Wealth can only be created when we delegate routine tasks to those who can do it better while we focus on high value tasks like asset management. If you want to do everything yourself you’re going to end up being grunt with a very sore back.
this bubble is bigger than the 1873 RR bubble and bust and what was referred to as a panic and greatest depression of all time. the money show in orlando last week displayed the bubble. only a handful of sober speakers who timed the past bubbles well. i personally rode the .com bubble up and got out. trailing stop losses whether automatically put in or calculated each day………….this should be a wonderful opportunity to make big bucks on the bubble and bust. again. i did the same with LEH and BSC and the real estate bust.
WHOLE LOTTA ENEMYS OF AMERICA IN HERE
Its SI now!
Sometimes I am my own worst enemy.
Do your patriotic duty and live next to a data center. Otherwise the reds over there win, and the reds over here might not win.
Computers will always be nothing more than crunchers of zeros and ones which have no ‘intelligence’ whatsoever. Anything else is nothing other than an imagined fake pipe dream.
Maybe there’s no real intelligence, but they are fast, consistent, and they never forget. They can copy, reproduce, summarize, slice and dice, existing intelligence in a split second.
No complaining, no HR, no medical insurance, no payroll tax to pay.
I think you underestimate the impact.
You should talk to a cognitive psychologist or neurologist sometime. Brains are just relating and correlating concepts too.
> “There is more demand for compute infrastructure than we can supply”
That’s what I observe as a software engineer from a top-100 tech company.
We are not one of the top 7 companies who sell AI; we are consumers of the frontier models, and we can no longer with without it.
AI is disruptive, but you only know it if you have the opportunity to use frontier model agents that spend hours on your problem doing analysis, development, and review.
If you use AI as a glorified search engine, you will probably think it is a bubble.
But keep in mind: companies are paying to use it.
railroads also worked wondefully well and changed the amerikan landscape and way of life. however it also was a huge bubble that burst and brought on a great depression. 89 RR companies went out of business. however i commuted by RR for most of my early life until age 35……….
Did the demand for railroads match the offer?
As far as I can see, the demand for AI is huge, and it is growing.
We are already seeing software engineers spending as much as their salary in tokens (gladly paid by the company, since it is translated in 10x gain in productivity).
It might be difficult to understand if you don’t have agents working for hours to solve a problem; AI is much more than a glorified search engine.
The “whistleblower” Jacob Coxon: “This notion of a lead is already basically a mirage. If China tried hard enough, they could immediately catch up just by just hacking and stealing the progress.”
That’s a problem. If you can just copy an AI model that someone else built, it’s going to be hard to make a profit from it.
Also, you can build your own, good-enough models at home (and download a decent model) for free.
Producing an AI model might be expensive but it is likely not difficult, given all the high number of models that popped up the past 4 years. It smells like it could be a commodity in a few years.
I think this rush to AI is more about defense than offense. Companies have to join the arms race to maintain their market share. I don’t think the ROI will be there.
The economy is lucky to grow at 2% per year on a real basis. Players must steal another company’s profit stream to grow faster than that.
And that’s my theory as to why societal discontent seems to be growing. If you’re with six people with one pie of pizza, everyone is happy if someone bakes a second pie. No one is happy if one person eats six of the slices and leaves the remaining two to be shared by the other five people.
MW: AI fatigue is gripping the market and driving chip stocks toward their worst stretch in 16 months
AI is great, but it requires long periods of focused effort to convert tokens ($$) into actual productivity. In short, if you didn’t need AI in the first place, you’re doing well. If you need AI to keep up, you’re already cooked.
Handing out $$ in the form of tokens !== value created.
If the once famous author Gertrude Stein were alive today, she’s say the same thing say the same thing about AI that she once said about Oakland, specifically ‘There is no there there.
jack london hood in oakland is spectacular place and weather. nice little ferry ride to SF to freeze for the afternoon.
” ‘There is more demand for compute infrastructure than we can supply*,’ and the imbalance could persist through 2029 and potentially into 2030, Broadcom’s Hock Tan said.”
*The missing piece here is “when we are renting it at a loss to gain market share”.
AI company operating losses – not counting new investment – demonstrate that prices are being held artificially low to attract money not from customers, but from investors. We still do not know anything about the supply/demand curve because the pursuit of market share and investor money has obscured it.
It is similar to residential real estate, where so many landlords are holding money-losing or ultra-low-ROE properties in the hopes that continued appreciation of the property will make them money.
I’m deeply wary of such markets, having lived though the dot-com crash and the mania that led to Housing Bubble 1. Basically, if it’s on the news as the next big thing, and the metrics everyone are looking at are market share things like clicks or tokens, you should be selling it, not buying with the rest of the lemmings. Money is made quietly, not by chasing trends.
thanks wolf