Is it a bubble?

Investment Analysis Clubs Macro Economic Trends and Risks
1

OK, time is short today because the cleaning people are coming, so I have to clean up before they get here. Weird, right? You probably do it too.

Anyway, there are a couple of stories in the Times & Journal which might lead to topics, but I don’t have time, so I’m just gonna reprint one I wrote yesterday over at Shrewdm.com on the Berkshire board, about AI and the “bubble” that’s being debated everywhere. I make no claim that I am right, just that it might lead to some conversation hereabouts:

There’s a consistent history of transformational technologies generating excessive enthusiasm and investment, resulting in more infrastructure than is needed and asset prices that prove to have been too high. The excesses accelerate the adoption of the technology in a way that wouldn’t occur in their absence. The common word for these excesses is “bubbles.” AI has the potential to be one of the greatest transformational technologies of all time. As I wrote just above, AI is currently the subject of great enthusiasm. If that enthusiasm doesn’t produce a bubble conforming to the historical pattern, that will be a first.
Ok, I’m a just say it: it’s a bubble. When you can’t point to where the profits will come from and you’re still throwing endless gobs of money at it, that’s a bubble. That it might work out in the end - as the fiber laid in the dot com era or the houses which got absorbed (at distress prices) after 2008 didn’t go up in flames is irrelevant, the bubble is the definition of *the investment dollars* going up in flames.

A quick trip thru history tells me that the tulip mania lasted about 4 years. The Nifty Fifty era lasted over a decade, seemed a reasonable strategy for a while but only got crazy during the last 4-5 years and peaked just a year or two before the OPEC embargo, which crushed it. The real estate fiasco of the aughts went on for roughly 4-5 years.

The Railroad manias of the UK peaked from 1843-45 and crashed in 1847, while that in the US took a bit longer (and later) from 1866-73 ending in the (eponymous) Depression of 73. So again, half a decade, plus or minus. The Dot Com era began in 94, but the excesses became apparent in the late 90’s, say ‘97 or ‘98, and it was all over by 2001, if not sooner. So again: 4-6 years, depending on where you plant your “starting” and “ending” flags.

I’m thinking we’re about a year, maybe two into this “AI/data center era”. Suddenly everyone will wake up and say “Whew, what a hangover!”, and sanity will return. A few will do well rationalizing distressed properties, others probably not so much.

None of this means Google or Meta or whoever will go bankrupt, but there will be a reckoning of stock pricing which naturally drags everything else down, at least for a while. But the railroads continued. The internet continued. The telecommunications fiber is being used. The housing got soaked up, heck, we have a housing shortage just a few years later! (The tulip thing didn’t work out well for anyone, I guess.)

Anyway, the Goofy clock says Armageddon still could be a couple years away. Party on.

19 Likes
2

The US government and other government debt are worse global bubbles.

Ducking before someone else pushes some sort of button.

3

Experience is that two or three companies will dominate the AI business. And they likely will do very well. Others will fall by the wayside.

Which are the winners? Long term investors want to know. Companies are investing to try to be one of the winners. Bubble or not the race is on!!

2 Likes
4

And if you’re holding the S&P 500, you’ll have a significant asset allocation to those 2 or 3 companies. No need to try to pick the winners.

intercst

5 Likes
5

So a person will need 10 years, in the worse case scenario, of liquid living expensive before the stock market recovers.
I suppose there might be a few people that can time the market collapse and could exit & re-enter the market. I ain’t one of them.

3 Likes
6

Yep. I ain’t one of them either. Plus I’d have a $1 MM plus capital gains tax bill if I tried.

intercst

2 Likes
7

The problem is the S&P often carries dozens of losers in addition to its winners and also rans. You can usually beat it by buying its top performing members and avoid the rest.

2 Likes
8

Read a couple of investing articles talking about the “bubble”. Their arguments of this time it is different was based on companies are spending their own money building out the infrastructure. So far, they are not borrowing money and incurring debt. One could look at it as R&D that either pans out or doesn’t.

Now if they start borrowing money, it could turn into a classic bubble.

1 Like
9

Probably not usually. Research shows that 95% of professional money managers and a higher percentage of amateur stock pickers under perform the S&P 500 over time.

intercst

1 Like
10

That’s actually not 100% true.

I mean, it’s literally true - Meta and Alphabet aren’t themselves going out and issuing bonds to build out the infrastructure. But the data centers are being built by issuing debt - it’s just that the debt isn’t being issued by the end users of the data center, because they really love their pristine balance sheets. Here’s a short summary:

The problem is:

  1. 1.The AI buildout should be financed by debt investors.

  2. 2.Debt investors want to buy debt of the big tech companies, because they are good credits.

  3. 3.The big tech companies do not want to sell debt, because they are good credits.

This is a very easy problem to solve! This is known technology! What you do is:

  1. 1.There’s a box, and the box will build the data centers. (Schematically the box is often a “special purpose vehicle” or a “joint venture,” though in some cases the box is an actual public company but not a huge software company.)

  2. 2.The big tech companies promise to make some payments (rental payments, etc.) to the box. These payment obligations are not “debt,” in the sense that the tech companies’ accountants decide that they do not count as debt on the companies’ balance sheets, and the credit ratings agencies decide that they do not count as debt for credit ratings purposes.

  3. 3.The box goes out and sells debt to investors. The pitch to investors is “the big tech company promised to make payments to this box, and the box will pass those payments along to you, so if you think about it the debt of the box is just as good as the debt of the tech company.” And the investors believe that.

Put the Data Center in the Box - Bloomberg

So you do end up with a ton of debt being issued. That debt isn’t issued by the big tech companies, but instead by the special purpose vehicles (SPV) that are the ones that formally own the data centers. That debt is secured by the promises of the big tech companies that they will pay rent (or whatever) to the SPV that owns the data centers for the use of those data centers.

Astute observers might note that if you take the terms “Big Tech Company” and “Data Center” and “Rent” in that scenario, and change them to “Residents” and “Homes/Condos” and “Mortgage Payments,” you end up with a very similar structure to that which caused the Great Recession when it collapsed. Similar magnitude, too - if Sam Altman is right that it’s going to take about one or two trillion dollars in AI infrastructure to build the AI Thingy, and 65-75% of that cost is financed through debt that depends entirely on the promise of Big Tech Companies to keep paying to use all that infrastructure, you’ve got a similar sized mountain of debt that would go from valuable to worthless if Big Tech Companies stopped paying their mortga….I mean, Data Center Rent Payments if AI doesn’t work out to be profitable.

That’s why some folks are worried about what happens if the AI bubble burst. Because all of that debt is going to be collateralized and distributed and held by a lot of folks. It has to be - it’s too much money. OpenAI doesn’t have half a trillion dollars. Neither does X/Grok, or Microsoft, or even Google. It has to be borrowed. Someone has to own that debt. And so if AI does end up borrowing a trillion dollars to build the infrastructure and does end up not being able to make that money back, it’s going to cause no small amount of problems.

15 Likes
11

I noticed another thing that will likely affect the eventual overall ROI of AI. Seems like some of the AI companies are selling their services, not for cash, and not fro future cash, but for “investment”. And investment isn’t revenue, and doesn’t fall to the bottom line as “profit”. The latest example is Disney today with their $1B investment into an AI company in return for AI services (to use their characters in AI). Sometimes investment results in sales, heck, usually they do, but the question here is how much sales, and at what price? Is it at a price that will produce enough profit, or is it at a large discount because of the initial investment (by a potential big customer)?

8 Likes
12

OK, an amusing AI story:

Last quarter I rolled out Microsoft Copilot to 4,000 employees.

$30 per seat per month.
$1.4 million annually.

I called it “digital transformation.”
The board loved that phrase.
They approved it in eleven minutes.
No one asked what it would actually do.

Including me.

I told everyone it would “10x productivity.”
That’s not a real number.
But it sounds like one.

HR asked how we’d measure the 10x.
I said we’d “leverage analytics dashboards.”
They stopped asking.

Three months later I checked the usage reports.

47 people had opened it.
12 had used it more than once.
One of them was me.

I used it to summarize an email I could have read in 30 seconds.
It took 45 seconds.
Plus the time it took to fix the hallucinations.

But I called it a “pilot success.”
Success means the pilot didn’t visibly fail.

The CFO asked about ROI.
I showed him a graph.
The graph went up and to the right.
It measured “AI enablement.”

I made that metric up.

He nodded approvingly.
We’re “AI-enabled” now.
I don’t know what that means.
But it’s in our investor deck.

A senior developer asked why we didn’t use Claude or ChatGPT.
I said we needed “enterprise-grade security.”
He asked what that meant.
I said “compliance.”

He asked which compliance.
I said “all of them.”

He looked skeptical.
I scheduled him for a “career development conversation.”
He stopped asking questions.

Microsoft sent a case study team.
They wanted to feature us as a success story.
I told them we “saved 40,000 hours.”

I calculated that number by multiplying employees by a number I made up.
They didn’t verify it.
They never do.

Now we’re on Microsoft’s website.
“Global enterprise achieves 40,000 hours of productivity gains with Copilot.”

The CEO shared it on LinkedIn.
He got 3,000 likes.
He’s never used Copilot.

None of the executives have.
We have an exemption.
“Strategic focus requires minimal digital distraction.”

I wrote that policy.

The licenses renew next month.
I’m requesting an expansion.
5,000 more seats.
We haven’t used the first 4,000.

But this time we’ll “drive adoption.”
Adoption means mandatory training.

Training means a 45-minute webinar no one watches.
But completion will be tracked.
Completion is a metric.

Metrics go in dashboards.
Dashboards go in board presentations.
Board presentations get me promoted.

I’ll be SVP by Q3.
I still don’t know what Copilot does.

But I know what it’s for.
It’s for showing we’re “investing in AI.”

Investment means spending.
Spending means commitment.
Commitment means we’re serious about the future.

The future is whatever I say it is.

As long as the graph goes up and to the right.

11 Likes
13

Hah! My company has been pushing AI hard for about 1 year. We’ve been told “There is no training, use it and figure out how it can be used to change our business”. There’s also a lot of talk about game theory, yada yada, and we’ll either lead our industry or be swallowed up!

Seems like a ripe opportunity for a consultancy gig…for both AI developers and AI users.

7 Likes
14

That is not my experience. Beating the S&P most of the time is not difficult. Index fund investors accept meager results.

1 Like