5-Month-Old LA Startup Wants 100,000 Satellites in Space
AI is hungry. Not for food, but for electricity and water. As ChatGPT, Claude, and Gemini get smarter, the terrestrial data centers powering them are hitting a wall. They need massive amounts of land, a terrifying amount of power, and millions of gallons of water just to keep the servers from melting.
Enter Orbital, a 5-month-old startup based in Los Angeles that has a solution that sounds like it was ripped straight out of a Christopher Nolan script: If you can't build them on Earth, build them in space.
Orbital isn't just thinking about a few orbiting servers. They have filed with the FCC for a constellation of up to 100,000 satellites designed to function as space-based AI data centers. 🚀
What is an Orbital satellite data center?
Traditionally, satellites are for communication (like Starlink) or observation (like GPS and weather satellites). Orbital is doing something fundamentally different. Their satellites are essentially flying racks of GPUs.
The goal? To provide roughly 10 gigawatts of AI compute power from low Earth orbit (LEO). To put that in perspective, 10 GW is enough to power millions of homes: or one very, very large AI model.
By placing Orbital satellite data centers in space, the company aims to solve the three biggest headaches of the AI industry:
- Power: In space, the sun never sets (if you’re in the right orbit).
- Cooling: The vacuum of space provides a unique environment for heat rejection.
- Real Estate: There are no NIMBY (Not In My Backyard) protests in LEO.
Why 100,000 Satellites?
You might be wondering why they need one hundred thousand of them. After all, SpaceX’s Starlink currently has around 6,000 satellites in orbit and is already considered a massive constellation.
For Orbital, the scale is about the "compute density." AI training and inference require massive parallel processing. By deploying a distributed network of 100,000 nodes, they can create a planet-scale computer that isn't tethered to a local power grid.
According to their FCC filing, these satellites would orbit at an altitude of 500–800 km. While they are only five months old, the team behind Orbital is moving with the typical "move fast and break things" energy that defines the latest stories in the tech world.
The Physics: How do you cool a GPU in a vacuum?
On Earth, we use giant fans and water-cooling systems to keep data centers from overheating. In space, there is no air to blow around. This is one of the biggest technical hurdles for Orbital satellite data centers.
Orbital’s plan relies on passive thermal management. Instead of using water, they use giant radiators to "dump" heat into the cold vacuum of space. Because space is essentially a giant heat sink, they believe they can achieve "zero-cost cooling" once the infrastructure is in place.
Furthermore, being above the atmosphere means their solar panels get hit with unfiltered, high-intensity sunlight. We're talking about 5x the energy density compared to solar farms on the ground. It’s the ultimate "green" energy play for an industry currently criticized for its massive carbon footprint.
Is this "California Weird" or a brilliant business move?
Only in Los Angeles would a 5-month-old company propose a 100,000-satellite constellation. Backed by a16z’s "speedrun" program with around $5 million in pre-seed funding, Orbital is the definition of a high-risk, high-reward venture.
Critics point out several "minor" issues:
- Launch Costs: Even with SpaceX’s Starship, launching 100,000 heavy compute satellites is eye-wateringly expensive.
- Latency: Sending data up to space and back down takes time. For real-time AI like a chatbot, those milliseconds matter.
- Space Junk: Adding 100,000 objects to orbit raises massive concerns about orbital debris and "Kessler Syndrome."
Despite these hurdles, the mission is clear. As terrestrial power grids struggle to keep up with the AI boom, the "final frontier" might be the only place left to grow.
The Roadmap: When do we see it?
Orbital isn't waiting decades. Their timeline is aggressive:
- 2027 (Pathfinder): A single GPU module will hitch a ride on a SpaceX Falcon 9 to test if AI inference actually works in a high-radiation, zero-G environment.
- 2028 (Orbital-1): The first purpose-built satellite with commercial inference capabilities.
- 2030s: Scaling toward that magic 100,000 number.
Conclusion
The idea of Orbital satellite data centers might sound like science fiction today, but so did reusable rockets fifteen years ago. By shifting the burden of AI compute from our strained Earthly resources to the vastness of space, Orbital is positioning itself at the intersection of the two most transformative industries of our time: Space and AI.
Whether they actually reach 100,000 satellites remains to be seen, but the "mission" is undeniable: Unlocking unlimited compute by looking up. For more deep dives into the world of economics and tech, stay tuned to Business Tantra.
FAQ: Orbital Satellite Data Centers
1. Why put AI in space instead of on the ground?
Ground-based data centers are running out of power and water. Space offers 24/7 solar energy and natural cooling, bypassing the limits of the Earth's electrical grid.
2. Who is funding Orbital?
The startup is currently backed by investors like Andreessen Horowitz (a16z) through their "speedrun" program, with initial pre-seed funding of about $5 million.
3. Will this cause more space debris?
This is a major concern. Orbital will need to comply with strict FCC and international regulations regarding satellite de-orbiting and debris mitigation to avoid cluttering Low Earth Orbit.
4. Can I use these satellites for my AI apps?
Not yet. The first commercial prototype is slated for 2028, with early testing starting in 2027.
5. How will the data get back to Earth?
The satellites will use high-bandwidth radio or laser links to transmit AI processing results back to ground stations, which then connect to the standard internet.
Tags:
Share:
Related Post
