We are well beyond just getting started. The “experts” are in reality salesmen. Don’t be fooled.
1 Like
This means there is a product or hardware that is not AI. There is software on that product that is not AI. There is an additional more complex layer of software for some of the features that we call AI. Okay for some features there is AI.
AI is a cost in that case. AI adds some features. It is code.
It is not magic dust. You do not get to be the AI whisper.
AI is useful. It costs money. It is a small productivity gain. It halucinates nonstop. I need to find the right question four or more questions in at times a few days later. If I asked a teacher, professor, or studied the book my video game education would be better and more structured. AI gueries are the wrong answers until the question is something you 3/4s know the answer to in advance.
That is completely untrue.
Your AI bear case…not bad for routine soft words
AI Overview
[image]
The main bear case for the Nasdaq in late 2025 centers on escalating global trade tensions, a weakening labor market, high valuations, and deteriorating economic data, which collectively suggest a potential for a prolonged global growth slowdown. The tech-heavy index entered a bear market in April 2025 after President Trump announced new tariffs, sparking investor anxiety over the negative impact on growth and corporate confidence. Strategists are also concerned about high valuations and rising bond yields, which make equities less attractive compared to less risky alternatives.
Key Arguments for a Bearish Outlook
The April 2025 announcement of new tariffs on over 180 countries triggered significant drops in the Nasdaq, as many tech companies rely on imports from countries affected by the tariffs.
Concerns persist that cracks in economic data, combined with the potential negative impact of tariffs, could lead to a significant global growth slowdown.
While some data remains stable, there are signs of a weakening labor market, which could indicate a more significant economic downturn.
Some analysts argue that stock valuations are stretched, meaning they are “priced for perfection”. Any future disappointment or bad news could send stocks tumbling.
Rising government deficits and long-term bond yields are seen as potential risks to the market’s stability.
Sentiment surveys show a sharp rise in pessimism among individual investors, a reading not seen in over a year.
Economic Indicators Pointing to a Bear Case
Weakening consumer sentiment and a weakening composite PMI signal a potential slowdown in economic activity.
Both consumer and corporate confidence are deteriorating, which often precedes a recession.
Data on airplane traffic and other indicators show signs of a potential slowdown in key economic sectors.
1 Like
Just for the record, and not that it’s important to the argument, very few people use Alexa to order from Amazon, which even Amazon admits. People do use it to check the weather, to play music, to tell the time, and less frequently to control smart home devices, but “ordering from Amazon” cases are vanishingly small.
That doesn’t mean they shouldn’t have it, just that the monetary use of it to the parent is like any other products, it isn’t incrementally adding after the sale.
Oh I don’t doubt that is true. If not for the Wrights, we fly later. If not for Vanderbilt, the railroads don’t happen for another decade. On the other hand, Jay Cooke went bankrupt after flying high in that same industry, pardon the mixed illustration. The Panic of 1873 - and another in 1893 - took down over 100 of the biggest railroad corporations in the industry at the time . The fact that they built a great technology and benefited the nation was of little solace to their investors, who ended up in the same breadlines as everyone else.
Names like the Atchison, Topeka & Santa Fe, Northern Pacific, Erie, Union Pacific and dozens of other well recognized names: all bankrupt, picked up for pennies by other investors later on. So it was with much of the fiber laid in the late 1990’s (ironically, some along railroad lines).
Nobody is saying AI isn’t a good thing, just that there’s also timing involved with any investment or new technology. This is likely to be one of those.
11 Likes
Indeed. Warren Buffett famously observed that “[i]f a farsighted capitalist had been present at Kitty Hawk, he would have done his successors a huge favor by shooting Orville down".
Airplanes literally changed the world. The air travel industry has been the bane of investors.
The importance of the tech doesn’t necessarily guarantee success to investors in the industry.
4 Likes
I think what people are trying to suggest is that the AI cycle has gone nuts and the excesses are catching up. The evidence are historically high valuations and the lack of a convincing narrative that AI revenues in the next few years will come anywhere close to justifying the current AI expenditures. Sounds kinda like a bubble.
And if you haven’t and have only seen the market go up, you might think it’s different this time when it really isn’t.
AI will be like most every other game-changing technology. The great majority of startups will go bankrupt. The great majority of early spending will be wasted and after the carnage there will be a few winners who win really big.
But here is the other AI issue that you aren’t considering. For a technology to boost the economy it has to create wealth for the consumer. It has to create a larger and wealthier consumer class.The industrial revolution replaced low paying farm jobs with higher paying manufacturing jobs. Computers replaced data filers with higher paying data analysts. AI is a technology that AI-optimists argue will replace humans in almost every endeavor. So where is the wealth produced by AI going to go when so many humans are going to be unemployable? Who will be the new consumers pumping up the economy?
9 Likes
Because of the AI excesses in good part, the markets are at extreme highs. The bigger they are, the harder they fall. No one here gets out alive.
The coming depression is wanted as an excuse to do things by this admin. The admin is working towards that day.
The “recession is coming” crowd is always fear mongering. Every year, they see a bubble, a crash and harken back to CSCO and Pets.com.
Their guru Jim has been pounding the table for a crash for last 25 years.
You may be right, but you will know cyclical top after the fact, and not during or before. There were people predicting next recession from 2010…
During dot com era, Cisco had 200 PE, and their customers are non-revenue making startup’s. Today $NVDA PE is barely 40, and forward PE is 30. Facts are not aligning with many of your narratives .
I am not sure why you keep saying AI revenue doesn’t justify the investments. $NVDA is having steller revenue growth, OpenAI, in just 2 years achieved $13B revenue run rate and has 800 million Monthly Active users. If those numbers don’t impress you, I can understand. But they are some of the fastest growth the world has ever seen.
DO you have any numbers behind your assumptions?? I would love to see them.
I’ve seen these market cycles and I’m also very familiar with the “doom-and-gloom” crowd. Their arguments often sound serious, mature, even wise. But if you track their thesis and actual results over 1, 3, 5, or 10 years, the record is clear, they rarely make money.
Take John Hussman, for example: down more than 50% over 15 years, yet still managing over $500M, and erstwhile Berkshire crowd celebrated him as a guru . Those CAPE-ratio warriors always fighting “excess valuation”. Back in 2014, many of them were pounding the table that $AMZN was worth maybe $10 a share — and they were getting hundreds of recommendations — while others were pointing out AWS could eventually generate $100B in revenue.
The point isn’t to smear anyone. At the time, those bearish arguments did sound intelligent, even compelling. And yes, they may be right in the short-term. But people consistently underestimate the long-term.
2 Likes
This doesn’t feel like the 2000-2002 market to me.
I just had my best day with drug stocks (PFE, LLY, MRK) since Pfizer came out with Viagra in the late 1990’s.
intercst
3 Likes
I made a fortune ignoring Hussman.
intercst
5 Likes
Primarily from sales of hardware (GPUs), not from AI directly.
Yes, but they are likely to burn through $8.5 billion in cash to achieve that.
Notably, OpenAI is not profitable. They are still running at losses and supposedly are not expecting to be cash flow positive until 2029.
There were a TON of companies in 2020 and 2021 that were growing revenue at crazy multiples (see the Saul board) and that were largely not profitable. Most of them are still well below their ATHs.
2 Likes
$NVDA GPU’s are the foundation for the AI training, AI inference, AI reasoning, etc… it is not just their GPU’s but a whole lot of other tech goes into it… $NVDA is the flag bearer and cornerstone of this AI revolution
So? They are growing revenue, taking market share, getting into new markets, use cases… that’s all that matters for now. IF they are running the company to be profitable then they will sacrifice the growth. We have gone through no profits, not enough profits, not enough cash flow… argument with $AMZN.
If you are worried about the market then you have too much allocation. Reduce or rotate to stocks that you are comfortable with.
5 Likes
I know of no one of the Berkshire board (here or at Shrewdm) who thinks he is anything but an idiot.
Cisco’s biggest customers were the telecoms, foundational behemoths with billions in revenue, and startups building out infrastructure like Global Crossing, etc. Again, (and I would wish for the last time) nobody is saying there are no uses for AI. It will be used for everything from content creation to cross docking, the question is can it ever hope to recoup the kinds of investments that are being made in this mad rush to have more, bigger, best?
6 Likes
In complete agreement. Full stop.
But to borrow from the analogy above, Boeing selling a bunch of airplanes doesn’t make airlines a great investment, just like NVDA selling a TON of GPUs doesn’t mean that the companies buying such will necessarily be a great investment.
3 Likes
Well even after 15 years they didn’t change their opinion… but I have been in tons’ of conversation with those folks when they were here… I know how deeply they worshipped him, and when questioned how much personal insults they indulge… the polite answer was.. he is a Phd, what is your credential…
I know you love those folks… Let us just say we disagree on this one.
Only time can tell.
I didn’t pay attention…
in 2016 in his first annual letter Sundar Pichai "mobile-first to an AI-first world"
I know $GOOGL is working on AI, acquired companies, developed many foundation technologies, but in 2018 when he said, I thought it is ridiculous to say more profound than electricity or fire??? What is he smoking??
“the most important things humanity is working on” and “more profound than, I don’t know, electricity or fire”.
But, in July 2024, Pichai said,
“The risk of under-investing is dramatically greater than the risk of over-investing for us here”
Still I was busy trading $GOOGL, because I thought AI will decimate their search franchise, it may very well… but I wish I had held on to the shares I bought…
If you are worried, just size your positions appropriately. You cannot invest with fear.
Bears work to figure that out. Bears are on that page. Just as Bulls are on their page.
Sometimes, bears know a lot more about things than the bulls.
With so many excesses in the market, it will pay to be a bear.
Bulls can become pigs. We are there now.
Slight disagreement… Actually my thinking is other way around. I don’t own any $NVDA, so there is some bias there, 
. $NVDA has the risk of buyers cutting down on chip purchase, missing numbers on a quarter and stock price correcting violently. However, the software/ application developers using $AI are relatively in a better position and can weather any eventual cooling.
For a guy arguing so strongly about AI, I also own no $NVDA, that is what makes market. We need to do what works for our individual situation.
Could you elaborate?
I mean, I think the risk that we’re in an AI bubble is the possibility that the AI we’re trying to build isn’t worth as much as it costs to build. That’s not a claim that AI generally isn’t worth what it costs to build, because almost any kind of modern software product can be (and nowadays usually is) considered to be AI in some way. Rather, it’s the push to develop incredibly powerful general AI systems that require enormous amount of chips and compute and energy, and therefore require a ton of capital to push towards.
Right now, the capital is flooding in. OpenAI (as a proxy for all the developers) can get as much money as it possibly wants from investors to go out and buy NVDA chips, because the investors are convinced that OpenAI’s eventual product will make a ton of money.
The bubble pops if the investors decide that the story isn’t likely to play out that way. If there isn’t enough economic demand for the type of general AI that OpenAI can build to make the investment worthwhile, then they’ll stop giving OpenAI the money to buy more chips. Which means that not just NVDA, but OpenAI (again a shorthand for all the software firms) get burned.
That’s how all these bubbles pop, if bubbles they are. “We’re not going to need that much fiber.” “People don’t get enough benefit from connecting their devices to the Internet of Things as much as we had hoped.” “There’s no way to deliver a useful AR/VR Metaverse to people at a decent cost.” “Blockchain is too inefficient and expensive to use for much other than cryptocurrency and financial transactions.” If the tech is going to generate as much economic value for investors as people hope, that doesn’t happen - but if that’s not the case, it’s hard to see how the software people don’t get hit, too, once the investors stop giving them the capital to buy the chips and pay for the power they need to generate and operate their models.
9 Likes
That’s your thinking. The capacity build is not really a bubble. The capacity overbuild occurs in every industry. I think it is important to set some basic framework for this conversation.
First of all, between 1995 to 2000 :
- The nasdaq went from 1000 to 5000, i.e., 5x rise
- Many startups raised significant capital, without viable business model
- Large # of tech companies IPO’ed,
- 1996: 276
- 1997: 174
- 1998: 113
- 1999: 370
Compared to the above, the tech IPO’s is far less
- 2022: 181; ChatGPT was launched only in November and no AI related company IPO’ed.
- 2023: Total 181; only 15% of these are tech IPO’s
- 2024: Total 225: Only 15% of these are tech IPO’s
- 2025: Total 262
Now, many tech IPO’s are existing profitable companies like Arm holding, Reddit, instacart, etc. Unlike previous cycle big AI companies like OpenAI, Perplexity, etc… are raising only private capital and not public capital.
- The heavy investment in those startup’s and subsequent failure of those companies resulted in significant losses for the investors.
- The sky-high valuation are limited to few names like $PLTR, and Application/ Software names that could play a role in AI like $MDB are still very reasonably valued like 10x of sales. The revenue growth of $MDB, $SNOW will not be impacted, and they may continue their growth. Even $PLTR will continue to grow, their stock market valuation and revenue growth will converge. This will happen during the bubble burst, right now, it is not happening.
The bursting of $PLTR, or few other stocks doesn’t even require AI bubble bursting.
Where AI bubble bursting may have significant impact in the valuation of big tech companies like Apple, NVDA, MSFT, AMZN, META, AVGO, TSLA… with an exception of $TSLA, all these companies have significant revenue, earnings, and distinct technology IP. The AI stock valuation bursting is not going to impact their business. Their existing valuation is not so stretched, there is a good chance they will catch up in few years… unlike 15 years it took for $MSFT.
So the valuation has to run even further for the stocks to crash 50%, 60%, 70% and take 5, 10 years to catch up.
3 Likes