Gradient is building a decentralized AI infrastructure stack that runs models across phones, laptops, and IoT devices instead of centralized data centers. The company is launching two core protocols, Lattica (a peer-to-peer data communication protocol) and Parallax (which shards large models to run across many devices). Gradient operates a network of Sentry Nodes that it says has facilitated over 1.6 billion peer-to-peer connections across more than 190 regions. The stack uses Solana to coordinate data and payments for devices contributing compute. The startup claims its approach can slash cloud costs and keep user data local, addressing privacy and concentration-of-power concerns. Gradient plans to release additional protocols, publish research papers, and open developer channels to encourage contributions. Critics cited in the article question latency and coordination complexity for large-scale decentralized inference.

Total raised

$10M

Funding rounds

1

Latest round

Seed

Latest activity

Jun 2025