Buyers
Use available capacity.
Buy compute or inference from listed providers, then receive SSH access, workload delivery, or a streamed model response.
Backers openCPX is live on EasyA Kickstart. Back the provider capacity market.
CapIX / Bare-Metal Capacity Exchange
Providers list spare compute, GPU windows, and inference routes. Builders pay CPX to use them. Smart Route splits workloads across cheaper reliable capacity to cut run costs.
$ capix route workload.py --mode smart > deposit: CPX receipt verified > split: CPU prep + GPU kernels + endpoint tasks > result: lower-cost route returned before execution
Genesis Grid
Join the Genesis Grid for early buyer credits and priority provider onboarding as new compute and inference capacity comes online.
Market activity
Buyers
Buy compute or inference from listed providers, then receive SSH access, workload delivery, or a streamed model response.
Sellers
List idle GPU windows, CPU lanes, lab machines, model endpoints, or reserved provider capacity for buyers to use.
Why the exchange works
The exchange shows provider supply and CPX-priced quotes for compute, inference, reserved windows, and job delivery. Smart Route compares price, latency, reliability, task shape, and availability, then returns a cheaper reliable path to execution.
Sellers expose spare machines, GPU windows, CPU lanes, and endpoint capacity with clear CPX terms.
Builders pay CPX to use that supply as SSH leases, routed workloads, or streamed inference.
CapIX's proprietary router decomposes scripts and prompts before matching work to cheaper reliable lanes.
The route engine compares each subtask against available capacity before execution to reduce workload cost.
CapIX network
capix.network uses CPX as the settlement rail for listed compute, inference, and routed workload capacity. The whitepaper explains provider listing, buyer redemption, and the proprietary Smart Route layer.