Editor’s Note: On Tuesdays we turn the spotlight outside of Porter & Co. to bring you exclusive access to the research, the thinking, and the investment ideas of the analysts who Porter follows.
When Porter finds investment ideas of high value from an analyst who isn’t part of the Porter & Co. family, we will send them to you.
These are analysts Porter has known and followed for years, if not decades, whose ideas and insights have withstood the test of time, and who have helped Porter make better, more profitable investments.
Below, you’ll find a guest essay from Matt Milner, the founder and Chief Investment Officer at Crowdability on why the pre-IPO opportunity in SpaceX could just be getting started.
*Disclaimer: Porter & Co. may receive compensation, referral fees, or other benefits if you purchase products or services mentioned in this email.
Have you ever looked up and seen a giant data center flying around in space?
If Elon Musk has his way, you will soon.
Last week, Elon announced that SpaceX is acquiring xAI. The deal is valued at $1.25 trillion, which is the largest M&A transaction in history.
The combined entity is a bet that “space-based computing” — literally, massive data centers floating around in space — will unleash the vast potential for artificial intelligence.
So today, I’ll be doing two things:
First, I’ll introduce you to the wild world of space-based computing.
Then I’ll point you to a way that any investor can play this trend today — for as little as $500, and for potential profits that could reach 41x or even more over the coming years.
The Limiting Factor: Compute
AI requires an extraordinary amount of computing power — or “compute” as it’s called.
For example, generating a single response from ChatGPT takes 151 times more compute than the sum of everything you can do on your phone or your laptop.
The demand for compute explains why Nvidia has become the world’s most valuable company, and why AI companies are scrambling to build massive data centers, like OpenAI’s $500 billion “Stargate” project with Microsoft, Softbank, and Oracle.
Sam Altman of OpenAI said: “Compute is going to be the currency of the future. It’ll maybe be the most precious commodity in the world.” It’s even been reported that the Middle East is shifting its focus — from oil to data centers. As Mohammed Soliman, Senior Fellow at the Middle East Institute in Washington DC said about this shift, “Compute is the new oil.”
The problem is, building out massive data centers filled with advanced AI chips is unbelievably expensive. For example, a single Nvidia Grace Blackwell 200 “superchip” costs as much as $70,000. AI data centers need thousands of them, and sometimes they need tens or even hundreds of thousands of them.
But now Elon is coming to the rescue.
The Wild World of Space-Based Computing
Data centers have suddenly become central to everything.
They sit at the intersection of artificial intelligence, energy, geopolitics, and economic growth — and they’ve turned into a critical asset. In fact, data centers might be one of the best assets in the world. They feature such characteristics as:
-
Long-term contracts from 10 to 20 years
-
Investment-grade tenants including Big Tech
-
Inflation-linked pricing
-
High switching-costs
-
Scarcity due to power constraints
These characteristics explain why private equity is investing heavily into this opportunity, and why data centers are being valued more like monopolistic utilities than real estate.
And now that SpaceX has acquired xAI, this is the opportunity that Elon’s going after.
Simply put, he believes that space-based computing — literally, massive data centers floating in space — will unleash the vast potential for artificial intelligence.
Here are the top reasons why building data centers in space might make so much sense — in general, and specifically for SpaceX.
1. Nearly Unlimited Solar Power
Power and cooling are the two biggest operating costs for AI data centers. But in orbit, solar panels get continuous, unobstructed sunlight. This dramatically reduces one of Earth-based data centers’ biggest constraints: energy cost and grid capacity. Furthermore, cooling can be done with radiative cooling into space, rather than with energy-hungry chillers.
2. Reduced Terrestrial Constraints
Space-based data centers avoid challenges such as land acquisition, zoning fights, and water usage, which is a growing political issue for AI data centers.
3. Ultra-Low Latency
For space-native or global applications — for example, satellite networks, Earth observation, defense & intelligence, or communications routing — orbital compute can be faster. Why? Because it enables computing to take place where the data sits already.
4. Strategic And Defense Advantages
Space-based data centers are physically harder to attack, potentially agnostic to jurisdiction, and attractive for military, intelligence, and national security workloads. This alone could justify early deployments, even if the economics are marginal at first.
5. Starship Changes The Math
Historically, the idea of space-based data centers was impractical because launch costs were prohibitive. But SpaceX’s Starship can change the math. It has massive payload capacity, it’s reusable, and marginal launch-costs drop dramatically.
And Now, Back To The Real World
Of course, space-based computing isn’t a slam dunk. There are many challenges, some of them daunting, others seemingly existential. Here are the main challenges:
1. Launch Costs And Logistics
To set up even one large data center, you’d need thousands of tons of hardware, thousands of launches (or a massive increase in rocket cadence), and the ability to do robotic assembly in orbit. Many analysts today consider this far from realistic.
2. Radiation Destroys Electronics
Space is brutal on semiconductors. Cosmic rays flip bits, chips degrade faster, and error-correction and redundancy add cost and weight. Furthermore, commercial hardware needs significant modification or shielding to survive. This is one of the least appreciated constraints in investor discussions.
3. Maintenance Is Extremely Hard
On Earth, a technician can swap out a failed GPU in minutes. In orbit, you either need robots or replacement satellites, or you accept high failure rates. Even underwater data centers — a much easier environment — struggled with maintenance economics and were eventually discontinued as a program. Orbit is far harder.
Other challenges include communication and latency limits, the difficulty of networking between satellites, the risk of space debris creating insane collisions, and of course, the fact that the consensus timeline for orbital data centers is a whopping 10 to 15 years.
Ouch.
Now, as you’ve heard Porter say many times before:
Never. Bet. Against. Elon.
But if you’re looking for an investment play in this arena that might come to fruition far sooner than ten or fifteen years — for example, later this year — here’s an idea for you…
Solving The Same Problem — Today, On Earth
I’ve identified a company that’s aiming to solve the compute problem for AI on Earth — and it’s doing so now.
This company’s solution is better, faster, and cheaper than existing options:
-
First, it works with Nvidia to get access to massive quantities of its advanced AI chips — chips that are almost impossible to buy, even if you’re Larry Ellison or Elon.
-
Then it builds out massive “AI factories” that are filled with these chips.
-
And finally, it rents out this infrastructure — its super-powerful AI “stack” — on demand.
So, instead of an AI company being forced to spend millions of dollars, even hundreds of millions of dollars, figuring out how to build its own AI stack from scratch, it can just pay this company to handle everything.
Every company in the AI sector today, and every U.S. government agency, needs specialized infrastructure so they can build, train, and run their AI. And now, instead of being forced to do it all on their own, they have the option to pay this company to handle everything.
Can you imagine how much that could be worth?
Obviously I can’t make any guarantees. I don’t have a crystal ball. No one does. Furthermore, as Porter has said about pre-IPO opportunities: “As you know, this is not the type of investment approach I usually take. It’s speculative, it’s high risk, and it’s not for everyone.”
But I believe getting in on this AI company — before its IPO — is a great bet. In fact, I believe it could potentially hand investors gains of 41x, or even far more, in the coming years.
Here are a few key things to know about this company:
-
It’s already bringing in $500 million in annualized revenues.
-
It’s already backed not just by Nvidia, but by InQTel, the venture-capital fund created by the CIA. One of the last pre-IPO companies backed by InQTel was Palantir (PLTR) — and at this point, InQTel has made an estimated 2,000x its money on it.
-
It’s already hired bankers to take it public this year. And for pre-IPO investors who get in now, the potential profits could be extraordinary.
I’ve discovered a way that all investors can get exposure to this company, pre-IPO, for as little as $500.
If you’re interested in learning more, Porter and I recorded a video for you.
We hope you enjoy it.
Happy investing,
Matt Milner