The computer business, by definition, is a fast-changing business and always has been. Leaders are leapfrogged by new technologies. I remember using a Wang computer.
Software can be easily superseded but at least it doesn’t require massive capital investments. But companies that rely on hardware invest billions of dollars in manufacturing plants and data centers.
It Really Is Possible to Spend Too Much on AI
Tech giants are betting they won’t end up like Intel by overinvesting in computing infrastructure
By Asa Fitch, The Wall Street Journal
It has become a tech-industry truism: Spending too little on chips and other computing infrastructure for artificial intelligence is riskier than spending too much…
But investors have begun questioning this logic, fretting that the spending spree might be inflating a bubble that will inevitably pop. Indeed, spending too much can be very bad. Just ask Intel…
Overinvesting may have been the correct move for Intel in its circumstances then [2021]. But technological missteps and changed circumstances in the chip market undercut the effort within a couple of years.
The result has been nothing short of a disaster. Manufacturing projects have been put on hold or canceled. The company has been bleeding cash with free cash flow in negative territory in all but three of its last 14 quarters…
Large financial commitments to AI data-center projects might no longer make sense, becoming a burden for tech companies for years to come. If assumptions about the value of AI-related assets and contracts change, write-downs are also inevitable. A lot of startups could fail…[end quote]
This is yet another warning about the dangers of the AI bubble.
Wendy
Yes, the AI spending is big. But we well know that two or three players will become leaders and get most market share. Name recognition is very important.
The mad scramble is on to see who the leaders will be in AI. The rest will specialize or fall by the wayside.
Growth has been bolstered by data-center investment and stock-market wealth. A reversal could raise the risk of recession.
The economy’s dependence on AI comes with risks. Stock price/earnings ratios are near record highs. If lofty profit predictions prove wrong, share prices may tumble and investment could slow. The S&P 500 fell about 2% last week on concerns about a bubble, despite rallying 1% on Friday.
Falling stocks could trigger a reverse wealth effect: Americans would consume less, which would tend to depress sales, profits and, potentially, employment.
To prevent a recession; isn’t it more likely a safety-net for overextended companies will be implemented similar to 2008 bail-out, seemed to be implied today?
Yes, except the politics of doing that while ending medicare obamacare educational debt helps to the general populace is, perhaps, too obviously grossly a tell as to who controls our politics to what ends….
Over extended companies refers maybe to major purchasers of AI chips and data centers. When they find they are not getting profits as planned they are likely to sell those assets. The major risk is that technology has moved on so their asset is obsolete.
Most other companies supplying the AI boom (APH, CAT, GLW to name a few) find AI as an opportunity for growth but not the basis of their business. They are at risk only if they have done major expansions in anticipation of AI business.
At the end of Q3 2025, Berkshire Hathaway had trimmed its Apple (AAPL) stake significantly. But, it is still the conglomerate’s largest holding. Berkshire Hathaway (BRK-A, BRK-B) made a big investment in Alphabet (Google’s parent company) during Q3 2025. That investment is somewhere between $4.5B - $5B. Chrome says the purchase was in GOOGL - the voting stock variety. I guess with Alphabet, there’s search, self-driving cars, internet services, etc. , one could argue the company is more than just AI, right?
I like ETN. They will be involved in some of the data centers construction and receive payment and resulting profits from such work.
Morningstar: Founded in 1911 by Joseph Eaton, the eponymous company began by selling truck axles in New Jersey. Eaton has since become an industrial powerhouse largely through acquisitions in various end markets. Eaton’s portfolio can broadly be divided into two parts: its electrical and industrial businesses. Its electrical portfolio (representing around 70% of company revenue) sells components within data centers, utilities, and commercial and residential buildings, while its industrial business (30% of revenue) sells components within commercial and passenger vehicles and aircraft. Eaton receives favorable tax treatment as a domiciliary of Ireland, but it generates over half of its revenue within the US.
Read an interesting article today about the “AI bubble”. Had a couple of points in relation to past bubbles showing things are different this time - for now.
1 - when compared to other bubbles, AI build out is only 0.8% of GDP. Still a lot of money but less when compared to the tele com era and even going back to the rail road build out where spending was 1.4% and 3% of GDP respectively.
2 - companies, so far, have been spending using current cash flow not issuing debt/borrowing. Although a few are starting to, so that can change.
@JLC AI may be less than previous bubbles, such as telecom and railroads, but it’s more than the dot-com bubble. Also, financial crises are caused by pressure points that crash the system such as the subprime mortgage debt that crashed Shadow Banking operators like Lehman and AIG.
Some companies, like Google, have been been spending using current cash flow not issuing debt/borrowing. Others are borrowing already. The loans will really start flowing next year.
And they have solid earnings and strong balance sheets. Very unlike the dot com bust where many players had anticipated growth but no earnings.
The Magic seven will survive an AI bust. Their earnings might be trimmed but they are not headed for bankruptcy. And some may yet become AI industry leaders.
I’m sure you are right and they will be fine in the long run. However, the volatility when the bubble pops will cause many speculators a lot of trouble.
They will need the cash flow for more data collection, model training and QA testing as long as their AI’s cannot recognize the difference between 8, 10 and 7.
For large investments, it’s usually a good idea to estimate what the ROI is going to be. I’m pretty sure all companies do this when determining where to place their capital. Sometimes they are correct and sometimes they are wrong.
A few years ago, there was a company that invested heavily into “the metaverse”. Almost all their capital spending went to that project, billions, even tens of billions of capital, were devoted to the project. And they made a huge splash about it, even to the extent of renaming the very popular company “Meta” instead of their former name. I’m trying to figure out what the ROI on that investment was exactly? Any ideas how to do so? Meanwhile, the market cap of the company has grown by leaps and bounds, so apparently the market still believes.
A lesson in how sometimes the Next Big Thing might not turn out to be the next big thing after all, even if our tech overlords are spending billions on it.
Related to the thread title, looks like AI isn’t getting quite the level of uptake among business users that those tech overlords are hoping for either:
The Meta story in and of itself is a support to the proof that investing in things that don’t pan out are part of a healthy business.
If the metaverse investments are equal (or near) zero.
Meta announces it will quit burning money in the 'verse firepit.
Meta stock value rises by 4% on rerating.
~small potatoes.
Assumptions above, postulates below:
Meta investments in AI will be non-zero value in the future (some say their conference call notes indicate they are achieving ad/sales revenue value NOW.
Meta is swept up in the valuation compression that MAY happen in the future. Investors assign ZERO value to AI investments
How many multiples of 4% will Meta suffer as a loss? 3x? 5x?
That value compression and the blunted sales revenue from tech recession malaise will be:
Short term
Not substantially terminal loss of value
Similar examples would be applicable as corollaries for goog, amzn, msft, etc.
This would NOT be applicable to upstart companies like Open AI, Nebius, etc. in my opinion unless those companies become profitable in a significant way.