Table of contents
Back to resources
SpaceX just merged with xAI to put a million data center satellites in space | In The Loop Episode 44
Published by
Published on
February 16, 2026
February 17, 2026
Read time
7
min read
Category
Elon Musk merged his AI company with his rocket company, creating a $1.25 trillion entity aiming to put a million data center satellites into low Earth orbit. In this episode, I dig into the financials behind the deal, the skeptics who think it's a bailout dressed up as a big vision, and the believers who think it's transformative.
I also want to use this as a lens to explore the historical patterns it highlights---how industry and production have always followed the cheapest energy source, from textile mills moving to coalfields to data centers chasing hydroelectric dams. We'll look at the Kardashev scale, the idea behind the Dyson sphere, the concept of a civilization harvesting its star, and whether AI's growing energy appetite is what finally pushes us to move infrastructure off-planet.
The deal and why it matters
Many of you will have heard about these companies. xAI is Musk's AI venture---the company behind Grok, accessible via X (formerly Twitter) and its API. It's the one behind the scandals enabling people to upload photos and generate manipulated images. SpaceX, of course, is the rocket company.
The merger matters because of energy. A single query to a large language model consumes roughly ten times the energy of a Google search. Scale that to billions of queries a day and the picture gets uncomfortable fast. The International Energy Agency found that global data center energy consumption hit about 415 terawatt-hours in 2024---around 1.5% of all electricity generated on Earth. By 2030, the same agency projects that figure will more than double to nearly a thousand terawatt-hours, slightly more than Japan's entire national electricity consumption today. Goldman Sachs estimates the surge in power demand is equivalent to adding another top-ten energy-consuming country to the planet. And that's before you account for growth trajectories everyone keeps revising upward.
It's in this context that Musk announced the merger of SpaceX and xAI, creating what the company called "the most ambitious, vertically integrated innovation engine on (and off) Earth." The combined entity is valued at about $1.25 trillion, with SpaceX at roughly a trillion and xAI at $250 billion. By several measures, it's the largest merger in history. All of it driven by the demand for AI.
The financials are bonkers
SpaceX generated about $15 billion in revenue last year with an estimated $8 billion in profit. Bloomberg reported that xAI recorded a loss of about $1.46 billion in Q3 2025 alone, while pulling in only about $107 million in revenue. That revenue did double last quarter, but it would need to double about four more times to catch up with xAI's burn rate. The quality of that revenue is also questionable, with many observers noting that what xAI calls LLM revenue looks a lot like advertising revenue from the X platform.
Days before the merger announcement, SpaceX filed an FCC application to put one million AI satellites into low Earth orbit. For context, there are about 14,500 active satellites around Earth right now. That's a staggering leap.
The application describes a first step toward becoming a Kardashev Type Two civilization, one that can eventually harness the power of its star to fuel energy demand. SpaceX has argued that orbital data centers would be cheaper and more environmentally friendly than terrestrial ones, with satellites venting heat into the extreme cold of space and harvesting solar energy without clouds or atmosphere in the way. Musk himself has claimed that "within two to three years, the lowest cost way to generate AI compute will be in space." Looking longer-term, he's described lunar manufacturing as a way to deploy 500 to a thousand terawatts of AI satellites into deep space orbit every year---remember, Japan's national consumption sits at about 900 terawatts.
The skeptics and the optimists
Not everyone's buying it. Investor Ross Gerber offered one of the most-shared reactions on X: "X was out of money, so it merged with xAI. xAI was out of money, so it merged with SpaceX. SpaceX is going to run out of money, so it merges with Tesla. And when they're all out of money..." SpaceX didn't explain how xAI would contribute to orbital data centers beyond being a customer, which fuels the argument that the move was primarily financial.
Musk with xAI faces the same funding challenges as any AI startup. You're competing against Meta and Google and their advertising-fueled war chests. As Gary Marcus wrote on his Substack, trading xAI at roughly a thousand times revenue when the company sits in an increasingly crowded field, where everyone is being forced into a price war, seems unreasonable.
Behind much of this sits the IPO question. SpaceX is reportedly targeting a mid-2026 IPO that could be one of the largest ever. Many analysts argue this is less about orbital data centers and more about packaging an attractive entity for that public listing.
The optimists have their arguments too. Finance commentator Tom Nash pointed out that when SpaceX talks about infrastructure, history suggests people should pay attention---even when the initial ideas sound unrealistic. A good example is Starlink. When SpaceX began launching internet satellites, the economics didn't work. But since then, launch costs have dropped dramatically. Starlink now has over ten million global subscribers generating roughly $10 billion in revenue. The thing everyone called impossible became routine.
Energy always wins: a 250-year pattern
Here's where the historical comparison gets compelling. Before anyone dismisses orbital data centers, the pattern has held for 250 years: industry follows the cheapest energy source. Every time.
In the late 18th century, textile mills in England clustered around rivers. Water wheels were the only viable power source, and the Pennine valleys north and east of Manchester were ideal---fast-flowing streams driving cotton production. Then the steam engine changed everything. Mills no longer needed rivers; they needed coal. By 1802, Manchester and Salford had more than 50 steam-powered mills, drawn by ready access to Lancashire's coalfields. The same pattern played out in the 20th century with aluminum smelting.
Look at data centers. Microsoft has operated a huge campus in Quincy, Washington, since about 2006. The town has no particular reason to exist as a technology hub except that it sits downstream of a massive dam where hydroelectric power is among the cheapest in the country. Yahoo and Vantage followed Microsoft to the exact same location. In 2024, Microsoft signed a 20-year power agreement to restart a nuclear power plant. Amazon and Google are doing the same---trillion-dollar companies doing what governments used to do, just to power data centers.
This is the context that makes the space argument interesting. Solar energy in orbit is about eight times more productive per unit of panel than on Earth---no atmosphere absorbing radiation, no night cycle, no weather, no seasonal variation.
But it won't be easy
The engineering challenges are immense. Latency is real. Even in low Earth orbit, you're adding 20 to 50 milliseconds of round-trip delay, which rules out real-time applications. Radiation degrades electronics, meaning maintenance requires robots in space, not technicians. And the manufacturing scale to produce a million satellites doesn't exist yet. SpaceX currently launches about 2,000 satellites per year---imagine what a million would require in production and deployment. It's close to science fiction, but it's also credible as a vision.
The pattern holds. Industry has always followed cheap energy, every single time, without exception. The question isn't whether that continues---it will. The question is whether and when the next migration moves upward.
I believe that if we're going to survive as a species, we need to take that leap. Even when COVID shut down most of the world, we were producing far too much CO2 to hit the Paris climate agreement targets. I don't believe it's possible to reduce our emissions fast enough. That's not me saying we shouldn't try---I've been a vegetarian for over ten years and I don't own a car. But we need solutions that are extraordinarily ambitious, we need them fast, and I believe AI has a huge role in that process.
The Kardashev scale and the Dyson sphere
This is where the Kardashev vision comes in, and the fact that SpaceX explicitly references it makes things even more interesting.
In 1964, Soviet astrophysicist Nikolai Kardashev published a paper proposing that civilizations be classified by the amount of energy they consume. A Type One civilization harnesses all energy on its planet. A Type Two harnesses the entire output of its star. Type Three captures all the energy of its galaxy. According to Carl Sagan's formula, we currently sit at about 0.73 on that scale---not even Type One. Most predictions place us reaching Type One status in about a hundred years.
What makes the SpaceX filing so striking, regardless of what you think of the company or Musk, is that it explicitly cites this framework as motivation for moving toward Type Two.
What they're alluding to is a Dyson sphere. Freeman Dyson described what Type Two infrastructure might look like: a swarm of objects orbiting a star to capture its total energy output. He later clarified that a rigid shell was impossible, but independent orbiting objects beaming energy back to civilization could work. That's SpaceX's vision: a million solar-paneled AI satellites orbiting Earth as a first step, eventually extending the concept to orbit the Sun itself.
It feels exciting. It feels right. Whether it's overly ambitious, whether it's a financial play, whether the companies are struggling---hard to say. But taking a step toward unlimited clean energy is probably always a good thing.
Closing thoughts
Every query you send to an AI model, every image generated, every agent running in the background consumes energy. Collectively, these queries are creating a demand curve that will double by 2030---the equivalent of adding a top-ten country to the global energy grid.
The SpaceX filing and the xAI merger speak to either a vision of the future or complete delusion. The honest answer is we won't know for years. You don't have to like the person proposing it. You don't have to like the company---I'd never own a Tesla, for example. But I believe in the future of this planet, and I believe the ambition is worth paying attention to.
Thanks for listening. I hope I'll see you next week.
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.png)
.jpeg)
Book a demo today.
