The brief
- AWS leads enterprise cloud with 1M+ deployed robots. Prime Air delivers to 500 US cities, hundreds of thousands of packages in 2026.
- Custom silicon: Trainium/Inferentia accelerators, Graviton ARM CPUs, cat-qubit chips. Project Rainier (500k Trainium2) trains Claude.
- Pieter Abbeel leads frontier model work, due re:Invent 2026. Kiro agentic IDE inverts coding: specs are source, code is artifact.
Technical approach
Custom accelerators for AI
Trainium (training), Inferentia (inference) and Graviton (ARM servers) reduce GPU dependence, lower costs and tailor silicon to AWS workloads.
Cat qubits for error correction
Ocelot cat qubits suppress certain errors, reducing physical qubits needed for error-corrected logical qubits by up to 90%.
Drone delivery at scale
Prime Air drones deliver up to 5 lbs in under 30 minutes using proprietary detect-and-avoid for safe autonomous flight.
Consolidating to one flagship model
Sunsetting Nova portfolio (Premier, Omni, Reel, Canvas) to focus on single frontier model due at re:Invent 2026.






