4 space company ideas
In the same vein as YC's Request for Startups, here are some of the space startup ideas that I cooked up on my long walks. I try to keep these ideas as realistic as possible (no warp-space anti-gravity technology) and as economical as possible. Please tweet to @neuralnuance if you have more ideas.
1. Autonomous Space AI Systems
Current space robots face significant communication delays—several minutes for Mars and seconds for Moon operations. In the Apollo missions, one critical part of the mission is when the spacecraft enters the far side of the Moon. The astronauts and ground station would experience a communications blackout as long as 47 minutes because the Moon blocks the line of sight to the vehicle.
The solution? We should develop autonomous systems powered by AI. Using a power-efficient variant of large language models, these systems could operate independently without human intervention. Given an overarching prompt:
"Guide the humans safely on this spacecraft while maintaining the orbit
The LLM would then send out signals that are translated to physical outputs on the spacecraft.
In general, the rule is that local compute is always more expensive than remote compute. This is why we interact with LLMs via remote APIs versus performing computations locally. However, when we are limited by latency (in the minutes) or connectivity, local compute starts to make a lot of sense.
Why use an AI versus hand-written code? Hand-written software is rigid and cannot fully cover every edge case. To make safer and more dynamic software, AI should be the way forward.
2. In-Space Satellite Internet Infrastructure
Right now, the problem with satellites is that you need to have a ground station that you talk to for TT&C. This means that everyone needs to have a custom arrangement with a ground station for every satellite that they launch. To simplify this, satellites should just talk directly to a mesh network of satellites that talk to ground stations. Satellite operators should not have to care about ground stations to begin with.
Using optical laser links and RF communication, this network would function as a secure, private internet for satellites. The system would utilize standardized bands with encryption, allowing satellites to share ground station infrastructure efficiently.
3. Automated CubeSat Manufacturing
The CubeSat industry needs modernization through automated manufacturing processes. Following the chip industry's model of separated design and manufacturing, this approach would focus on producing standardized, SpaceX-compatible parts.
The satellite designer would design their satellites with standardized parts. They would then communicate their designs in some standardized file format. I'm reminded of the Gerber file format that the electronics industry uses.
Currently, there is no easy way for satellite designers to manufacture their custom designs. They either need to DIY or have their satellites built by an aerospace company that would charge them a hefty price.
4. Suborbital Point-to-Point Freight Services
Suborbital point-to-point has been explored by people in the industry, including SpaceX. The idea is that we would send something (human, products etc) on a spacecraft on a suborbital trajectory to their destination. Think New York to Tokyo in 20 minutes.
This is technically and economically possible now. The caveat is that this only works for highly valuable goods (medical supplies, drugs, chips, advanced materials).
How it would work: Put them into a Dragon spacecraft or any spacecraft that is able to perform a deorbit burn and reentry. After they are in orbit, just deorbit and reenter the atmosphere to the destination. The most expensive parts would then be the logistics around the recovery of the spacecraft after reentry. Solving this portion of the transportation would be critical. We would need to solve this by developing a spacecraft with landing capabilities.
The clients for this would value drastically reduced transit times, and current rocket technology can make these hops a reality.