Swarming Resources
On-Demand Distributed Compute
Lease compute from distributed providers and swarm across them. Any node can operate as a provider. Orchestrate inference swarms across distributed nodes for parallel processing, model sharding, or distributed training. No central scheduler required.
Lease, Accept, Settle
Compute leasing on XE follows a three-phase lifecycle recorded on the block lattice. Consumers create lease blocks, providers accept and provision VMs, and settlement mints XE emission rewards. Request 100 VMs across multiple providers – you get a swarm.
What You Can Lease
Specify vCPUs, memory, and disk with durations from 60 seconds to 365 days. Providers run Ubuntu 24.04 VMs via QEMU with your SSH key injected for direct access through the libp2p tunnel gateway.
LLM Inference
Distribute large model inference across multiple leased VMs. Model-parallel and pipeline-parallel strategies across your swarm.
Distributed Storage
Replicate data across geographic regions for redundancy and low-latency access from any network point.
Compute Workloads
Containerised workloads deployed across leased VMs. SSH access via the libp2p tunnel gateway for direct control.
Bandwidth Sharing
Distribute content delivery and API traffic across provider nodes for low-latency global reach.
XUSD Pricing & XE Rewards
Resource costs are denominated in XUSD. XUSD is burned when providers accept leases (deflationary). Providers earn XE emission rewards on settlement (inflationary). Collateral stakes align provider incentives.
- vCPU: 20 milli-XUSD/hour · Memory: 10 milli-XUSD/GB/hour · Disk: 1 milli-XUSD/GB/hour.
- Provider collateral: cost / 5 (minimum 1 XUSD), recovered on settlement.
- XE emission reward equals the XUSD cost of the lease.
- Performance certificates required – 60-second benchmark validates provider capabilities.
Harness the Swarm
Start deploying distributed workloads on the XE network. Linear scaling, trustless leasing, no central scheduler.