Welcome back to SavvyMonk, your one-stop for AI and tech news that actually matters.

Today we're talking about the wildest FCC filing you'll read all year. Blue Origin just asked the U.S. government for permission to put tens of thousands of AI data center satellites into orbit. And they're not the only ones with this idea.

Let's get into it.

The Flyover cuts through the noise mainstream media refuses to clear.

No spin. No agenda. Just the day's most important stories — politics, business, sports, tech, and more — delivered fast and free every morning.

Our editorial team combs hundreds of sources so you don't have to spend your morning doom-scrolling.

Join 2.3 million Americans who start their day with facts, not takes.

Start Reading for Free

On March 19, 2026, Blue Origin filed an application with the Federal Communications Commission to launch and operate up to 51,600 satellites in low Earth orbit. The project is called Project Sunrise. Its purpose: run AI compute workloads directly in space.

The satellites would orbit between 500 and 1,800 kilometers above Earth in sun-synchronous orbits. That means they'd stay in near-constant sunlight, generating power from solar panels without needing batteries or a grid connection. Each orbital plane would contain between 300 and 1,000 satellites, connected by laser-based optical links.

The pitch is straightforward. Terrestrial data centers are running into hard limits. They need land. They need power. They need enormous amounts of water for cooling.

Blue Origin argues that space solves all three problems at once.

The Two-Constellation Strategy

Project Sunrise doesn't stand alone. It's designed to work alongside TeraWave, a separate Blue Origin constellation of 5,408 satellites announced in January 2026.

TeraWave is the communications backbone. It handles the data transfer between the orbital compute layer and the ground, supporting speeds up to six terabits per second through its medium Earth orbit satellites.

Think of it this way: Project Sunrise is the brain. TeraWave is the nervous system.

Together, they form a vertically integrated orbital infrastructure stack that no other single company has proposed.

And Blue Origin has one more advantage here. It builds its own rockets. The New Glenn heavy-lift vehicle flew twice in 2025, with the second flight nailing a booster landing on a drone ship.

The company is now producing roughly one full rocket per month and targeting double-digit launches in 2026. That vertical integration, building the rocket, the satellite, and the network, mirrors the playbook that made SpaceX dominant with Starlink.

The Competition Is Insane

Blue Origin isn't even the biggest player in this race. Here's what the orbital data center landscape looks like right now:

SpaceX filed with the FCC in January 2026 for a constellation of up to one million orbital data center satellites. The filing called it a step toward becoming a "Kardashev II-level civilization," one that can harness the full power of its star. SpaceX completed its acquisition of xAI on February 2, 2026, in a $1.25 trillion deal. Musk said the merger was largely about building space-based data centers. An IPO targeting a valuation of up to $1.75 trillion is expected this summer.

Starcloud, a startup based in Redmond, Washington, filed for 88,000 satellites in March 2026. The company already has one satellite in orbit. Starcloud-1, launched in November 2025, carries an Nvidia H100 processor and was the first satellite to run a version of Google's Gemini AI model in space.

Google announced Project Suncatcher in November 2025. It envisions clusters of 81 satellites carrying the company's own TPU chips, connected by optical links. Google plans to launch two prototype satellites with Planet Labs by early 2027. CEO Sundar Pichai has said orbital data centers could become normal within a decade.

And Nvidia is building the hardware for all of them. At GTC 2026 in March, the company unveiled the Vera Rubin Space-1 Module, designed specifically for orbital data centers. CEO Jensen Huang said it delivers up to 25 times more AI compute than the H100 for space-based workloads. Six companies, including Starcloud, Axiom Space, and Planet, are already using Nvidia's space computing platforms.

Why Everyone Wants to Move AI Off Earth

The core argument is simple math. AI workloads are growing faster than the power grid can keep up. Data centers already consume massive amounts of electricity, water, and land. Building new ones requires years of permitting, construction, and grid upgrades. Communities are pushing back against the noise, water usage, and energy demands.

Space, at least in theory, sidesteps all of that. Sun-synchronous orbits provide near-constant solar power. There's no land to buy, no grid to connect to, no water needed for cooling. Heat radiates directly into the vacuum. And you can scale without zoning disputes.

Blue Origin's FCC filing makes the case directly: solar-powered satellites with always-on energy and no land or displacement costs "fundamentally lower the marginal cost of compute capacity compared to terrestrial alternatives."

The Skeptics Have a Point

Not everyone is buying it. Sam Altman called the idea "ridiculous" at a press conference in New Delhi in February 2026. His argument was that the the math doesn't work yet. Launch costs are still too high relative to the cost of power on Earth. GPUs break frequently and you can't send a repair crew to orbit.

Gartner published a report calling orbital data centers "peak insanity," arguing that space-grade solar panels cost roughly 1,000 times their terrestrial equivalents and that cooling in the vacuum of space is far harder than it sounds.

There are real engineering questions too. Blue Origin hasn't launched a single TeraWave satellite yet. New Glenn has only flown twice. Deploying 51,600 satellites would require sustained launch capacity for years.

And there are already roughly 14,500 active satellites in orbit. Adding hundreds of thousands more raises serious concerns about orbital debris and collision risk.

Google's own research acknowledges that orbital data centers won't become cost-competitive with terrestrial ones until launch costs fall below $200 per kilogram, roughly seven times cheaper than current rates.

Google estimates that threshold could be reached by the mid-2030s, but only if SpaceX's Starship achieves high flight rates.

The Bottom Line

This is real paperwork, not science fiction. But it's also not happening tomorrow. Blue Origin, SpaceX, Starcloud, and Google are all placing enormous bets that the economics of space-based compute will eventually work.

The question isn't whether orbital data centers are a good idea in theory. It's whether launch costs, hardware reliability, and thermal engineering can catch up to the ambition.

Right now, the gap between the vision and the reality is still measured in light-years.

But the fact that this many companies are filing this much paperwork tells you something: they believe the gap is closing.

Category: Strategic Analysis

"I'm evaluating a technology that won't be commercially viable for 5-10 years but has significant long-term potential. The technology is [describe technology]. Help me build a framework for deciding whether to invest resources now or wait. Consider: current cost vs. projected cost curves, competitive positioning advantages of early entry, risk of the technology never reaching viability, and opportunity cost of capital. Present your analysis as a decision matrix with clear thresholds for action."

Every generation has its version of "that's impossible." Putting data centers in orbit sounds absurd until you remember that streaming video from space sounded absurd ten years ago, and now Starlink has over nine million subscribers. The line between impossible and inevitable is thinner than we think. Hit reply, I read every response.

See you in the next one.

— Vivek

P.S. Know someone who follows the space industry or AI infrastructure? Forward this their way. They can subscribe at https://savvymonk.beehiiv.com/