SPICE is a Bittensor subnet where competing AI models analyze satellite imagery, delivering enterprise earth observation at a fraction of centralized costs.
SPICE (Satellite Processing & Intelligence Compute Engine) is a Bittensor subnet that democratizes satellite intelligence. Every five days, ESA's Sentinel-2 satellites photograph the entire planet at 10-meter resolution for free. Most of that data goes unanalyzed because existing providers like Planet Labs and Maxar price out 90% of potential users.
SPICE fixes this by creating a decentralized marketplace where hundreds of independent AI teams compete to deliver the best satellite imagery analysis. Miners run computer vision models across five task types (classification, segmentation, change detection, object detection, and regression) and are scored against objective ground truth from ~1 million labeled satellite tiles. Better models earn more TAO. Pure meritocracy.
The timing is driven by regulatory pressure: the EU Deforestation Regulation (EUDR), CSRD, and SEC climate disclosure rules now require tens of thousands of companies to prove their supply chains are deforestation-free using satellite evidence. These companies need affordable monitoring, not sub-meter resolution.
The long-term vision goes beyond software. SPICE starts as a Bittensor subnet analyzing free Sentinel-2 data, then scales toward launching a proprietary satellite constellation for sub-meter, daily-revisit imagery. From decentralized analysis to owning the entire data pipeline end-to-end.