Design for death.
Build to outlive.
Every AI accelerator in orbit today was designed for a data center on Earth. When a chip dies in space, it stays dead. This is the architecture of a chip designed from first principles around that constraint.
Chips are the reason
space compute costs 4×
The analysis converges: launch cost is falling, cooling is tractable, power is abundant in orbit. The unsolved problem is chip mortality. Dead chips accumulate. Nobody comes to fix them.
Trade FLOPS for lifetime.
“Trade 40% peak FLOPS for 10× better annual chip mortality.”
This trade is obviously wrong for Earth. It is obviously right for space. No existing chip has made it — because every existing chip was designed by an engineer optimizing for a benchmark, not for a satellite that cannot be serviced.
Four decisions.
All wrong for Earth.
Each pillar makes a tradeoff that any Earth-focused chip designer would reject. In orbit, each is the right call. Click any to expand.
What 10× mortality improvement does.
The 18× monthly capital cost gap traces almost entirely to overprovisioning and short satellite lifecycle. This chip attacks both directly.
Google's feasibility study puts cost-competitiveness at $200/kg to orbit. Starship targets $250/kg. The eat-the-sun orbital ring targets ~$50/kg via mass driver. At that launch cost, combined with this chip's mortality profile, orbital compute TCO crosses below terrestrial a decade ahead of any existing model's projection. No other team attacks both dominant cost drivers simultaneously.
Where everyone else stops.
| Player | Hardening | Fab | Unit Price | Volume |
|---|---|---|---|---|
| Ramon.Space | RHBP — Hardening by Process | Specialty gov fabs | $10K–$100K/chip | Low / gov |
| NVIDIA (Earth) | Not hardened | TSMC 4nm | Commercial | Massive |
| →This Chip | RHBD — Hardening by Design | TSMC 5–7nm | Commercial | High |
Ramon.Space proves the market exists. They cannot serve it at scale — defense-process pricing locks them into government contracts. No commercial orbital compute operator can build a business on $100K chips. We fill that gap: commercial fab, commercial pricing, space-native design.
The credibility ladder.
Five rungs from zero to funded, with physical silicon in hand. No VC required to reach Rung 3.