We Begged You to Buy Real Estate. Hotels Are Up 42%.

Depending on which news source you trust this week, we are living through either the best of times or the worst of times. Pick your poison. Bloomberg, one of the few mainstream outlets that still does actual reporting instead of vibes, has been chewing on something we have been warning readers about for months: the deeply circular nature of the artificial intelligence economy.

Let us be clear about something before Wall Street tries to paint me as an AI skeptics.

Nobody is a bigger believer in the long-term future of artificial intelligence than me. I walked into this technology a year ago a skeptic and walked out an enthusiastic convert.

That said, conviction does not mean blindness, and what is happening right now is circular in a way that should make people nervous.

Look at the money. Nvidia is financing hundreds of billions of dollars in commitments to OpenAI, one of the largest purchasers of Nvidia chips on the planet. The same pattern shows up in the new arrangement with the Korean conglomerate SK Corp., another major customer. These companies are financing their own customers.

A meaningful chunk of OpenAI's capital is coming from the very vendors selling it equipment. That is circular by definition, and calling it anything else is just marketing.

Blackstone updated its outlook for AI capital expenditures to roughly $750 billion, up from $600 billion back in January. That is a staggering increase in spend, and to date there has been no meaningful return on any of it.

Go back and look at the fiber optic buildout during the Internet boom. Most of the early adopters who raced to lay cable across the country went bankrupt, and the larger telecom companies eventually bought that dark fiber for pennies on the dollar and booked it as a win.

History does not always repeat, but it rhymes loudly enough to notice.

Will artificial intelligence follow the same script?

Probably not exactly, because the major players here are actually profitable and throwing off enormous cash flow, unlike the Internet era casualties. Still, if the return on all this capital expenditure never materializes, somebody eventually has to answer for 45 to 50 times earnings and 10 to 15 times sales. Time will answer that question.

We do not lose sleep over it here, because roughly three quarters of what we do is grounded in deep value principles, with the rest built on quantitative momentum. Headlines are not our business.

We are not overly exposed to AI spending, and we are not overly exposed to the general market madness surrounding the Iran conflict either, which remains its own tangled mess of competing narratives.

The war is over and we won, the war is over and they won, or, more likely, the war is nowhere close to over and nobody wins.

Market volatility tied to that story will keep rattling portfolios, and we will keep treating it as a buying opportunity, because that is what disciplined investors do with fear.

Earlier this year I spent considerable ink getting down on my knees and begging readers to invest in real estate. As commercial real estate headlines turned uglier by the week, the portfolio kept getting cheaper, eventually trading around 70% of net asset value across most holdings, with a 6% dividend yield backed by companies capable of growing payouts well ahead of GDP and inflation over the next decade.

It was a generational buying opportunity, and for the most part, readers ignored me. I heard some choice names. The sharpest criticism landed when I recommended bank stocks, and both banks and REITs have since outperformed the broader market from the depths of that criticism.

We sidestepped tech for the most part. Real estate and banks, in my view, are two of the true long-term beneficiaries of artificial intelligence, and the return will show up with the users rather than the builders.

Infrastructure is becoming commodity-like territory. Owning the data center will likely prove far more profitable than being the contractor who built it and paid for all the equipment sitting inside.

Over time, I expect the better trade to be AI's landlords rather than AI's construction crew. We will find out whether that call holds up, but real estate has already begun quietly taking a commanding lead over many of the larger, more heavily traded technology names.

The numbers back this up. The S&P 500 is up 8.6% year to date in 2026, and the tech heavy Nasdaq 100 is up 10.5%. REITs broadly are up about 20%, and our portfolio is matching that figure closely.

I told readers we were seeing a bottom in commercial real estate, that the worst case scenario was that values would bottom and bounce along the floor while collecting cash until the inevitable rebound.

We have exceeded the worst case.

The NAREIT All Equity REIT Index is up roughly 15%, several other REIT ETFs are up around 20%, and our average holding in the total return portfolio is right there with them, well ahead of the Nasdaq, the Russell 2000, and the S&P 500.