Is AI the biggest bubble we have ever witnessed? There are many voices claiming so, while others are of the view mankind has reached a new plateau in its evolution. As this is not a new topic per se, and my skepticism is well-known among my readers, I am not going to drill deeper on this front. However, there is a subtopic that deserves much more attention than it actually gets. Maybe because it is so boring and likely ahead of its time, the majority currently does not care. I do, though. And you probably should either.

Summary and key takeaways from today’s Weekly
– I wanted to have written this Weekly for long, now was finally the time.
– On the surface impressive, under the hood a shame: accounting gimmicks push profitability, while earnings and free cash flow started to diverge.
– This is a huge warning sign. One day the party will be over – with potentially dire consequences.

AI, or artificiel intelligence, is widely either seen as a great scam full of exuberance or as the safest way to have made money with the respective stocks over the last years. In the latter case, it is a foregone conclusion that we are just at the beginning of something really big.

Barely anyone is NOT talking or at least thinking about AI.

What is almost safe to say: expectations are sky-high. When the majority knows about a trend, you can barely call it a secret. When almost everyone is invested, as it seems, valuations unlikely will be cheap, like is the case now.

This does not automatically imply these stocks cannot rise any further. They can.

But with rising expectations, the risk of disappointments rises, too. It is a hallmark of bubbles that they last longer than any rationally thinking mind assumes.

Today, I’m taking a completely different approach to this topic. I do not want to debate about valuation multiples, but about the insanity of investments, and its potential ramifications the retail crowd is completely ignoring. They will have their excuses later, but it is likely smarter to be prepared in advance.

Brace for some surprising insights.

---

If you’re looking for attractive, overlooked stock ideas the majority has a blind eye on, look no further. I am offering compelling non-mainstream cases in concise 12-page member-exclusive reports.

Want to outsmart the market? Get active now! Become a supporting paid-member and receive my best stock ideas (plus updates).

Join me and my members on our journey to beat the markets!

^{both as per 6 August 2025 market close – since August 2022}

Natural stupidity and the potential AI disaster

Since about four or five years, we have a recurring pattern.

The major tech stocks, namely Nvidia (ISIN: US67066G1040, Ticker: NVDA), Alphabet (ISIN: US02079K3059, Ticker: GOOGL), Meta (ISIN: US30303M1027, Ticker: META), Microsoft (ISIN: US5949181045, Ticker: MSFT), and amazon (ISIN: US0231351067, Ticker: AMZN), are spinning the wheel of artificial intelligence.

They are pulling up other names with them, but due to their size, those mentioned above have reached hefty weightings in the S&P 500 as well as the NASDAQ index.

And of course, they are the ones making the big headlines.

Below is a chart showing that it’s the top ten (by market cap) that have driven earnings momentum. There are of course a few other companies in the top ten, but the big tech stocks dominate this ranking (see here).

Until here, likely everyone will agree. No surprises.

There are of course also privately-held companies like Elon Musk’s xAI or the ChatGPT-owner OpenAI, but these are neither known to be profitable, nor are they publicly listed, hence not suitable for today’s Weekly.

The cloud- and AI-providers, as I would like to call them, continue to invest heavily in their AI infrastructure, benefitting Nvidia as the primary supplier for the required GPU hardware. They all level each other up like in a perpetuum mobile, they continue to post rapidly rising sales and earnings, beat and raise on their results announcements like clockwork, and up their spending projections even more.

Only the sky seems to be the limit.

Has the wheel of fortune finally be invented?

It pretty much seems so. But one name is missing in the list above: Apple (ISIN: US0378331005, Ticker: AAPL).

Not because I have forgotten to mention them, but because Apple is seen as a stupid laggard that has missed on one of the biggest wealth-generating opportunities in our lifetime. It is seen as having practically lost this race already. Okay, such actions like the still-missing big Siri update that should have been aired last year is not supportive.

However, my take, like is often the case, is an entirely different one. I am seeing Apple as pursuing the absolutely correct strategy. You can call me crazy, but I urge you to continue reading. More on than later.

Let’s first start with the financial shenanigans that slowly cannot be denied anymore.

Does anybody remember the initial investment thesis for Big Tech stocks? I am referring to the time about pre-2022, more between 2018–2022.

Think about it for a moment, before reading on.

3…

2…

1…

Got it?

I asked Grok (my default search engine) for help, as I strangely was not able to access such sites. When I click on the links Grok gave me, the sites pretend not to exist anymore.

Grok gave me ten such sites, and summed up the pitches which all point(ed) in the same direction. Anybody who is active in the markets for at least three to four years, will instantly know and likely have an “aha” effect.

^{source: my Grok inquiry to provide me with sources}

Indeed, it was the supposedly brilliant model of little competition in their respective fields, some have seen as monopolies or oligopolies, which was not even unreasonable. Add to that low-investment requirements due to scalability (software, ads, cloud services, and pricing power), resulting in high-margin businesses.

We should not forget the clean balance sheets, full of net cash.

This powerful combo allowed for predictably high and rising free cash flows on which analysts could create their valuation models on.

Let’s now do an over-the-air update.

This still holds true – for the laggard Apple with the by far smallest bar.

Everyone else has massively hiked their investments – by factors, not just a few percent. Capex did not experience a one-time bump, but it continues to rise every year.

The above are quarterly numbers. Here are the ones over the last twelve months on a quarterly basis.

The core message remains the same: with the exception of Apple, all other mentioned tech companies have started what it looks like a competition on who can out-capex the rest of the group.

Undoubtedly, momentum is strong and the trend is up.

Even though we are just half-way through 2025, first projections for next year are being made. Not just by analysts, but by the companies themselves.

Here are for example the plans of Meta.

No, it’s neither a typo, nor just a single case.

Here’s Alphabet.

In the link with the quarterly capex overview, the respective twitter post had also pulled quotes from all the companies.

If you haven’t clicked and read it, here it is again.

Capex is on the rise. The sky is the limit. No one wants to be left behind.

Period.

So fine, so good. What’s the issue? The companies are investing for their future.

That’s only half the truth that both, the companies themselves, but also most posts on twitter, and videos on YouTube, are telling you. Whether they don’t want to or just don’t know is not even important.

The real issue is this one here.

The above was shown by an article with the headline:

“Hidden Dangers in the AI Boom: The Asset-Light Model of the ‘Big Seven’ is Being Disrupted, and the Return on Massive Investments Faces Uncertainty”.

I liked the following paragraph from the article which I did not want to rewrite, as it hits the nail on the head, as they say.

This scene may be highlighting the hidden risks behind the AI boom — no one doubts the potential of artificial intelligence to enhance growth and productivity over the long term, but the huge financing that supports this boom may put pressure on businesses and capital markets.

^{source: Futubull, see here}

This article is from a Chinese website. Everyone is free to think of it what one wants. But it is based on common sense and how companies’ financials work.

And I think that the core message and its potential impacts are broadly underestimated.

First, to double-check, I did the same exercise via tikr. Free cash flows since 2024:

While constantly higher earnings were not questioned (at least until here, but rest assured I am not skipping that), we can clearly see that from 2024 onwards free cash flows at the very minimum have not risen, to put it mildly.

If we look closer, we can see a modest downward trend.

At the same time (longer timescale), sales are going up…

… and so do operating earnings, especially after the dip during 2022:

The divergence should be clear by now: figures on the profit and loss statements go up, while free cash flow tilted in the other direction. This is obviously not sustainable, as investments need to be written down gradually due to depreciation.

In layman’s terms, this means higher capex should equal higher depreciation expenses, which pressure earnings. Please learn that sentence by heart.

The thing is only, investments (capex, or real cash outflows) occur immediately, but are being depreciated over a longer time period – the expected useful live of such an asset. That’s why cash flow is pressured stronger by capex than earnings are by depreciation in the P&L. Especially when investments are strongly ramped up, there’s a time lag until depreciation catches up.

Now looking at the companies, we see the following. Apple’s capex and depreciation have moved sideways. Check, that’s not the interesting case.

But looking at the rest of the pack, we see the big divergence, capex (the falling line) increasing strongly on the negative axis, while depreciation (the up trending line) increases either, albeit at a much slower pace due to what I just described.

Capex hits immediately, deprecation occurs over the asset’s estimated lifetime.

So far, so unspectacular.

However, what shall be questioned is the pace of the depreciation – or in other words, the estimated useful lifetime of these assets.

In 2021, the tech companies already used an accounting gimmick to boost their earnings. Yes, indeed. They prolonged the estimated useful life of their to-be deprecated assets.

Here’s the annual report from 2021 of Microsoft.

This is not my phantasy. On p. 44 in the same report they wrote:

The same from Meta in the same year:

Alphabet? You bet.

Amazon followed a year later.

What a difference can such a small change of just one year make?

In the case of Microsoft we have seen above that it bumped earnings by a billion USD (lowered depreciation by this amount). You can decide for yourself whether this is nothing. It probably was not much in relation to total earnings of several multi-billions.

But hiking capex by a third, while deprecation falls by 10% has a bitter taste to it.

While I am personally a fried of using stuff longer instead of throwing it away, we need to keep in mind that it is not about personal usage. I can work on a four-year old MacBook the same I could when I purchased it. That is not the point of reference.

In a world of ever and seemingly faster changing technological advancements and requirements, the consequence is rather LOWER, not higher, expected useful lives of state-of-the-art hardware. The examples above were in 2021 and 2022 – well before the data center investment spree started, and before Nvidia’s GPUs came into the spotlight with the same dynamic like now.

It is NOW about AI, chatbots, a race to be the fastest, the best, and most capable. Everyone wants to be the first, the best, etc.

Where do these investments go you might ask? In this context, it is weird that estimated useful lives of these assets – here predominately GPUs, RAM, and other components that in a couple or three years make huge performance jumps – seemingly can be deprecated slower, instead of faster.

Could this be the reason for the predictable, reliable, and amazingly strong earnings growth, and not the true underlying dynamic?

In the case of Meta, we have a fresh update.

They again increased their internal estimates of useful lives – to 5.5 years. With this gimmick, they lowered depreciation by 1.58 bn. USD, and in return bumped earnings by 1.34 bn. USD – for the first half alone.

But that’s not all. This is a blended, i.e. average figure for all their equipment.

Some assets like buildings obviously have a significantly higher estimated lifetime. However, most of the investments are unlikely to go into buildings, but rather in quickly becoming-obsolete hardware.

So is it fair to assume that they use Nvidia’s GPUs for almost six years, if we assume the average? Even if the result is four years – will they remain competitive?

Looking back to the annual report 2022, we can see that they back then hiked their assumption even twice in 2022, first from four to 4.5 years, and then at the end of the year to five years.

Now, Meta in a matter of two years increases their estimated useful life for their equipment by almost 50%, now 5.5 years. The next annual report would be a great opportunity to go up to six years.

No seriously, this is dangerous.

To not let up any doubt, in their latest quarterly report, they confirm that the majority (“mostly”) of their capex goes into data centers, network infrastructure and servers. From the provided table we can see that “servers and network assets” is not only the single-biggest, but also the by far fastest growing category, now representing almost 60% of their property and equipment.

And this shall justify a higher “estimated” useful life?

What’s the bottom line?

Well, cash flow tells the true story, while earnings are bended frequently to make them look stronger than they truly are. As the majority of “investors” focusses on headline numbers, earnings, and the PE ratio, this works fine. They believe this BS.

Famous short-seller James “Jim” Chanos estimates / calculates that Meta’s estimated lifecycle might even trend towards 22 years.

There’s more to it.

Likewise noteworthy, where does Meta intend to get the cash from for their ambitious plans?

They won’t be able to pay with “earnings”.

Cash and short term instruments are down. Still high, but down against their plans to raise capex even more. Their entire free cash flow was used to repurchase own shares.

So in essence, Meta is burning strongly through its cash reserves while it very likely massively overstates its profitably. Based on the last twelve months figures, and assuming the true useful lifetime is only half of what they use, deprecations expenses would be 16 bn. USD higher.

I think that would read entirely differently than the pretend perpetual growth story. Reminds me of the Metaverse insanity from a few years ago.

To not make this Weekly too long, I leave out the other tech stocks Microsoft, Alphabet, and amazon. We have seen that they in the past, too, were not shy to play accounting games either.

Currently, the market seems to be betting on the success of these big tech companies to transition from their former asset-light to now asset-heavy businesses. I don’t know where the evidence is, but the first who stops investing will likely lose the race entirely.

Even worse and far-reaching, we recently had the number of US GDP growth which came in at 3% and was highly celebrated. I am sure there were plenty of pull-forward effects due to tariff issue.

Tim Cook confirmed that for example during the recent conference call of Apple.

But even leaving this aside as if did not happen, the BEA (U.S. Bureau of Economic Analysis, see here) reported inflation-adjusted Q2 GDP growth of annualized 3.0%.

U.S. nominal GDP (current dollars, not inflation-adjusted) increased from 29.252 tr. USD in Q1 2025 to 29.837 tr. USD in Q2 2025 – a plus of 585 bn. USD. For a reason, I wrote down the difference in nominal USD, 585 bn. USD in US GDP growth.

With big techs spending like crazy, they account for a big part of that.

Concretely, just Microsoft, Meta, Alphabet, and amazon spent c. 98 bn. USD, or 17% of the entire GDP growth in the last quarter (just a quarter while the GDP figure is for an entire year).

I know that they don’t spend everything in the US, but big parts of it. I also know that this is a major over-simplification, but it does not negate the fact that the tech companies have become big, really big.

At least that’s some kind of explanation why the consumer side looks rather weak while headline numbers come along rather strong.

With expectations being sky-high, earnings not as clean as they look, and Apple being seen as the ultimate loser, I wanted to add my view to this discussion.

Apple seems to be doing maybe not everything right, but many things.

They are avoiding this stupid capex race which would cost them much money now. And the other thing that barely anyone talks about is that these are still cyclical businesses. All of them. Their current cash flows come from their core businesses. Segments where huge investments are made into are unlikely to throw off much cash, if at all. The next downswing will pressure sales, earnings, and cash flows.

What will remain? Higher depreciation! Assuming no more bending, twisting and stretching their books, there’s even a decent chance this investment-mania will bite huge holes into the future results.

Apple will be free of that.

Which doesn’t mean I am favoring the stock right now. It is too expensive for me for what it offers. But I have no doubts they will go their way.

Apple is not standing still, and they will come up with new stuff for which they will first be criticized, but then envied, and finally shamelessly copied.

Apple is no garbage-company. They have the financial means to acquire growth. Until here, Apple primarily grew organically, and only bought small companies. But who knows, there have been rumors they might be interested in Perplexity. During the call, management said that M&A is not off the table.

Part of their strategy is to grow their recurring services business. Of course will some people cancel their subscriptions during economic hardship. But this fast-growing, higher-margin business is highly underrated.

It is currently very unpopular to hear or read, but I am almost sure it will happen this way. Human behavior does not change, and certain patterns simply repeat.

This gold rush is similar to commodity companies spending and spending, until the cycle shifts, and they are forced to write off their previous over-spending. On top, these accounting gimmicks in the past have rather been used by companies that tried to hide their financial struggles, until they collapsed. Honestly, I do not even rule that out either, even though this is a very bold call at the moment.

Diverging earnings and free cash flows are a major warning sign!

In essence, I expect the same perma-bulls who first celebrated asset-light businesses with strong free cash flow growth stories, to come up with new excuses. Now, they say FCF doesn’t matter (anymore). After the party is over, and when AI assets will be needed to be written off, it will be seen as a one-time, non-cash, and tax-enhancing event. The previously burnt cash – who cares? Just ignore it!

I hope this brings in some clarity into the capex cycle of the big techs.

Conclusion

I wanted to have written this Weekly for long, now was finally the time.

On the surface impressive, under the hood a shame: accounting gimmicks push profitability, while earnings and free cash flow started to diverge.

This is a huge warning sign. One day the party will be over – with potentially dire consequences.

By becoming a Premium or Premium PLUS Member, you get instant access to all my already published research reports as well as several updates.

Likewise, you qualify for eight, respectively four more exclusive reports with my best investment ideas plus updates on the featured businesses over the next twelve months.