Fabric is a distributed compute network designed specifically for AI and data workloads. It connects idle compute power from devices worldwide with computational needs, allowing users to submit jobs via API or dashboard without managing infrastructure.

The platform supports numerous workloads including AI inference, ML training, data processing, web scraping, media processing, agent simulation, GitHub Actions runners, iOS builds, macOS builds, and custom containerized workloads. It offers specific pricing for embeddings ($0.0001/text), transcription ($0.001/file), and web scraping ($0.001/request).

Fabric operates as a distributed network where an orchestrator automatically splits and distributes work across available devices. The system provides end-to-end encryption with isolated sandboxed environments for secure workload execution, and includes features like autoretry and no cold starts.

The platform offers significant cost savings (80% cheaper than traditional cloud providers) and faster processing times through distributed computing. Use cases span AI development, data analysis, software development workflows, scientific computing, and various business applications requiring scalable compute resources.

Fabric is built for developers who want to ship applications without managing infrastructure, replacing services like Lambda, SageMaker, Colab, and GitHub Actions. It integrates with existing workflows through SDKs and supports containerized workloads.