The data center in the sky

A few days ago, during my usual morning walk, I looked up and saw something strange: a long-tailed light in the sky that made an almost perfect triangle.

It wasn’t a plane. Living close enough to an airport, I know what an aircraft looks like. Besides, this object, whatever it was, wasn’t moving as far as I could tell.

I’ll admit that, for a moment there, I thought I might be seeing a UFO. But I’m not really a sci-fi kind of guy, so I turned to what used to be considered sci-fi instead…

Artificial intelligence (AI).

Still walking, I told ChatGPT what I was looking at. And its explanation ended up being much more fascinating than ET phoning home:

That cargo mission, simple though it might seem, has enormous implications for investors. Today, SpaceX (SPCX) is supplying the space station with solar arrays.

Tomorrow, it could be delivering the building blocks for data centers orbiting the earth.

With that realistic potential on the table, let’s discuss how income investors should react.

I’ve written about this before, but the artificial intelligence revolution is draining our power resources at an alarming rate.

AI requires an enormous amount of computing power, and computing power requires an enormous amount of electricity. Considering all the electricity this country – and the larger world – uses without Big Tech’s new passion factored in, that puts everyone in an increasingly enormous bind.

That’s why AI giants and data center operators are getting more and more innovative in managing their needs and expectations.

Take Alphabet (GOOG) with its Google Project Suncatcher, which is exploring solar-powered satellites. As of October 1, it’s in active-stage testing – as in circling the earth even now – equipped with Tensor Processing Units (TPUs) that are interconnected using optical communications.

If everything goes as planned, this high-tech solar panel will be up to 8x more productive than it could be on land.

Project Suncatcher is also testing out AI chips in orbit. Though, as advanced as that mission is, it’s still not as groundbreaking as you might think. Starcloud, a company wholly devoted to launching data centers in space – already demonstrated last year that it can work.

There’s also Lonestar, which is developing space-based data storage and computing infrastructure. And Star Catcher, which is putting together an orbital power grid that can transmit energy between spacecraft.

Again, this is all activity that was relegated to the realm of science fiction not that long ago. So it’s pretty awe-inspiring what we’re accomplishing and what we’re still driving (or launching) toward.

Considering all the investment opportunities emerging, I understand why financial headlines are so focused on Big Tech. But having a background in real estate rentals, I’m more geared toward looking past the “shiny new building” to whoever owns the dirt beneath it.

When it comes to AI, that means identifying what it absolutely can’t function without, such as electricity and data-center capacity. It’s all about finding established necessities that are hard to replicate… and then capitalizing on them.

For most of human history, infrastructure investment has been earthbound: roads, railroads, power plants, cell towers and, most recently, data centers.

But as my morning walk earlier this week showed, we’re in a whole new world of possibilities. And I, for one, want to keep up.

Yet I also have to recognize how volatile this world is despite the massive amounts of money and effort we’re expending to explore it. Space-based data centers and supporting technologies are still working through a whole list of problems like:

The common denominator in solving these issues is compute: the power, capacity, memory, networking, central processing units (CPUs), and graphic processing units (GPUs) necessary to make our digital lives function.

In which case, even the most intrepid, out-of-this-world, sci-fi pursuing geniuses remain exceptionally reliant on data centers. So I’m not even close to being stumped by the question of where investors should look if they want to be involved in technological breakthroughs – without putting their money on a limb.

The answer is right here on earth… as it has been for decades and will no doubt continue for decades to come. In which case, there are three real estate investment trusts (REITs) that need to be mentioned.

In the lineup of data center landlords, Digital Realty easily stands out with over 300 facilities located across six continents. Its long list of customers includes hyperscalers, enterprises, and cloud providers in their constant and increasing pursuit of computing capacity.

I particularly like Digital Realty’s PlatformDIGITAL setup. This ecosystem keeps those customers locked in through operations-wide connectivity that then creates significant switching costs.

Plus, this REIT is creatively effective in adding capacity even while land, electricity, and permits become increasingly scarce.

Add in Digital Realty’s investment-grade balance sheet, and you’ve got an attractive way to participate in AI infrastructure – without gambling on groundbreaking technological efforts that may or may not succeed.

Equinix may be smaller with its 270+ data centers worldwide. But it’s still a great company to consider.

Its greatest competitive advantage is the interconnected ecosystem it created. Equinix’s exact setup helps the thousands of businesses, cloud providers, and network operators it serves exchange data both securely and efficiently.

That, in turn, creates powerful network effects that are hard to replace. Essentially, Equinix’ customers are well-incentivized to stick with it, both by quality and scope of service and switching costs.

Like Digital Realty, this data center powerhouse boasts an investment-grade balance sheet. And I also appreciate its historically notable recurring cash flow growth.

All told, it’s hard to beat Equinix’s competitive moat, which is why I’m impressed by it today.