Darkbloom – Private inference on idle Macs
5 hours ago
- #apple-silicon
- #privacy-inference
- #decentralized-ai
- Darkbloom is a decentralized inference network that utilizes idle Apple Silicon Macs to reduce AI compute costs by up to 70% compared to centralized alternatives.
- Operators retain 95% of revenue, with near-zero marginal costs for idle hardware, leading to lower prices and profit from electricity costs as low as $0.01–0.03 per hour.
- The network ensures verifiable privacy through four layers: end-to-end encryption, hardware-verified keys, a hardened runtime, and traceable hardware attestation, preventing operators from accessing user data.
- It offers an OpenAI-compatible API for chat, image generation, and speech-to-text, allowing seamless integration with existing SDKs and tools.
- With over 100 million Apple Silicon machines often idle, Darkbloom addresses a marketplace problem by connecting distributed capacity to demand, similar to models like Airbnb and Uber.