Lambda's new strategic partnership with Nous Research: decentralized artificial intelligence

We’re pleased to announce our partnership with Nous Research to help develop Psyche, a decentralized AI training network. The system is designed to allow anyone to contribute to model training using idle compute, making AI development more open, efficient, and verifiable.

This initiative addresses a long-standing problem in AI: the high barrier to entry caused by the cost of training. Psyche is built to enable experimentation, lower infrastructure requirements, and distribute control away from a small number of centralized actors.

What is Psyche?

Psyche is a Rust-based decentralized training system that uses peer-to-peer networking to coordinate multiple training runs across devices. Instead of relying on centralized data centers, it allows individual users with idle machines—such as gaming PCs—to contribute compute to model training.

All coordination between nodes happens on the Solana blockchain, providing a fault-tolerant and censorship-resistant system.

The Core Technology: DisTrO

Psyche is made possible by DisTrO, a set of training optimizers developed by Nous Research. DisTrO reduces the amount of data exchanged between nodes during training by several orders of magnitude, enabling training over standard broadband connections.

The idea is conceptually similar to image compression (like JPEG): much of the essential information in a model’s gradient can be retained by transmitting only a few low-frequency components. DisTrO goes further by transmitting just the sign of each frequency amplitude, quantizing it down to one bit. This results in roughly a 3x further reduction in data transmission.

Additionally, nodes can start training without immediately applying the updates from the previous training step. This means that network latency does not become a bottleneck, improving resource utilization and allowing decentralized training to approach the efficiency of centralized systems.

The P2P Layer

Networking for Psyche is handled by Iroh, a protocol designed for decentralized applications:

• Each peer is identified by a 32-byte Ed25519 public key, not an IP address.
• Communication is end-to-end encrypted and authenticated.
• Nodes behind NAT or firewalls connect using UDP hole-punching in approximately 90% of cases, with relays used as fallback.

Nodes participating in training runs share training metadata using iroh-gossip, which builds on the HyParView and PlumTree protocols. Training results are shared using the iroh-blobs protocol, which bundles gradient information into binary blobs and references them via content-addressed tickets.

Training Lifecycle

Training in Psyche occurs in epochs (groups of training steps). Nodes can join or leave the network at the start or end of an epoch, reducing the opportunity cost for contributors.

At the beginning of each epoch, nodes download the current model (either from a HuggingFace repo or from other peers directly) and begin training. Some nodes act as witnesses, verifying received results using Bloom filters. If too few nodes remain active or witness quorum is lost, training is paused and checkpointed until new nodes join and resume the process.

Verification

To verify that nodes are training correctly, selected nodes should recompute the training performed by another node and check that the resulting gradient is accurate

Due to the non-deterministic nature of training (from rounding errors, hardware differences, etc.), the system must find a balance between accepting minor differences in output and detecting actual faults or adversarial behavior. Various similarity metrics—such as Jaccard index, Manhattan distance, and Hamming distance—are being explored.

Why This Matters

The current landscape of AI is dominated by a small number of entities with access to significant compute resources. This centralization limits who can participate in developing and steering the future of AI.

Our work with Nous Research on Psyche represents a meaningful step toward more open and equitable participation. It allows:

• Efficient use of idle compute
• Lower-cost training of custom models
• Greater experimentation and model diversity
• More transparency and less reliance on opaque corporate infrastructures

We believe AI should be owned by everyone. This partnership is a move in that direction. Lambda will work as hard as possible to build the new networks that make decentralized, open, and verifiable AI development practical, scalable, and accessible to all.