Salad is an orchestration cloud for AI inference.
We schedule containerized workloads across distributed GPU capacity in 191
countries, 16,000+ GPUs running every day, and deliver production throughput
at a fraction of hyperscaler cost. Push a container image, get an endpoint back.
No instance types to reserve, no capacity queue, no commitment.
Read more
Teams run image generation, speech-to-text, computer vision, molecular dynamics,
and large-scale batch inference on SaladCloud in production today.
The hard part isn't the GPUs. It's making heterogeneous, residential-network
hardware behave like a dependable inference substrate: node trust scoring,
availability forecasting, cohort selection, and failure-tolerant rescheduling
fast enough that your users never see a node drop. Individual nodes are
unreliable. The fleet is not.
That orchestration layer is the product, and it isn't tied to our hardware:
- AI Gateway — routed inference across models and providers
- Container Engine — fully managed container orchestration
- Community Cloud — flexible, low-cost GPU access on distributed capacity
- Salad Dedicated — reserved enterprise GPUs on contract
- Private Cloud — Salad orchestration running on your own fleet
Salad began in 2018 by rewarding people for the idle time on their gaming PCs. More
than two million machines have joined since; that supply engine is why the
capacity exists and why it costs what it costs.
SOC 2 Type II. Backed by Left Lane Capital, Origin Ventures, Kickstart, Royal
Street Ventures, and Carthona Capital.
Deploy at salad.com.
Earn with your PC at salad.com/download.
