For GPU Providers
Turn your idle GPUs into recurring revenue
VectorLay is a distributed inference network. You bring the hardware — GPUs, CPUs, or AI accelerators; we bring the customers, the orchestration, the isolation, and the billing. Plug in your nodes and get paid on every inference request they serve — we handle the software, sales, and billing so you don't have to.
Per transaction
Get paid on every inference request your hardware serves — CPU and GPU alike.
Monthly
Clean payouts based on verified utilization — no invoicing to do.
No lock-in
Drain and pull your nodes from the network at any time — your hardware is always yours.
You provide the compute. We run everything else.
Being a provider on VectorLay isn't the same as running your own GPU cloud. We take on the hard, expensive parts of the business so your hardware can simply earn.
We bring the customers
Your hardware joins a single distributed network that developers deploy to in minutes. You never have to find, onboard, or support tenants — demand is routed to you automatically.
We schedule the work
Our control plane places workloads on your node via Nomad, manages VM lifecycle, health checks, and failover. No manual ops — the orchestrator keeps your hardware busy.
We isolate every tenant
Customer workloads run in hardware-isolated QEMU VMs with direct VFIO passthrough to your GPUs or accelerators. Their code never touches your host filesystem, drivers, or other tenants — and all VM data is wiped on termination.
We handle billing & payouts
Metering, invoicing, fraud, chargebacks, and collections are all on us. You get a clean monthly payout based on verified utilization — no payment infrastructure to build.
We give you visibility
A provider dashboard shows real-time utilization, earnings, and node health so you always know what each node is producing.
We secure the network
Every node joins an encrypted WireGuard mesh to the control plane. Traffic between your hardware, the network, and customers is encrypted end to end.
How we monetize your compute
The model is simple: customers pay per inference request, and you earn a revenue share of every transaction your hardware serves — CPU and GPU workloads alike.
- 1
Customers send inference to the network
Developers route their inference workloads to VectorLay and pay per request at market rates.
- 2
Our control plane routes requests to your nodes
After benchmarking, the scheduler sends matching inference traffic to your hardware and keeps it busy.
- 3
We meter every transaction
Every inference request your hardware serves — CPU or GPU — is metered and attributed to your node.
- 4
You earn a revenue share, paid monthly
You earn a revenue share of every transaction your hardware serves. VectorLay's cut covers customer acquisition, orchestration, isolation, billing, and support.
Getting started takes three steps
No driver wrangling, no BIOS spelunking. Install the OS, run one script, and we take it from there.
Step 1
Install Ubuntu 25.04
Do a clean Ubuntu 25.04 install on your node — no GPU drivers, no CUDA, no manual configuration. The onboarding script handles all of it for you.
Step 2
Run the onboarding script
Generate a one-time token in your provider dashboard and paste it into our one-line onboarding script. It installs everything, joins the encrypted network, and registers your node automatically.
Step 3
We benchmark & schedule traffic
We automatically benchmark your hardware, place it in the right tier, and start routing inference traffic to it. From there it earns on every request — no manual ops, no babysitting.
Hardware requirements
Provider FAQ
How do I make money as a VectorLay provider?
You get paid on every inference request your hardware serves — both CPU and GPU workloads. When customers run inference on your node, you receive a revenue share of what they pay for each transaction. Rates vary by hardware — higher-demand accelerators like H100 and A100 earn more. Earnings are tracked in real time in your provider dashboard and paid out monthly.
How much of the revenue do I keep?
Providers keep the majority of the revenue their hardware generates. Your exact share depends on hardware type and commitment — VectorLay's portion covers customer acquisition, orchestration, billing, isolation, networking, and support. Contact us for the rates for your hardware.
Is it safe to run other people's workloads on my hardware?
Yes. Every customer workload runs inside an isolated QEMU virtual machine with its own kernel. GPUs and accelerators are passed through with VFIO and memory isolation is enforced at the hardware level by IOMMU. Tenants have no access to your host filesystem or to each other, and all VM data is wiped when a workload ends.
What hardware can I contribute?
A wide range of compute: NVIDIA GPUs (RTX 3090/4090/5090, A100, H100, H200), AMD Instinct and Radeon GPUs, Intel data-center GPUs, other AI accelerators, and high-core x86_64 CPU nodes for CPU-only workloads. You'll need a CPU with IOMMU for passthrough, at least 32 GB of RAM, a 500 GB+ NVMe SSD, a stable 1 Gbps+ connection, and a clean Ubuntu 25.04 install — no drivers required, the onboarding script sets everything up.
Can I pull my node off the network whenever I want?
Yes. There's no lock-in or long-term commitment. You can drain and remove a node at any time — it stops receiving new workloads, in-flight jobs finish or migrate off, and you keep everything you've earned up to that point. Your hardware is always yours.
Do I have to manage customers or support?
No. VectorLay handles customer acquisition, onboarding, billing, and support. You provide the compute; we run the marketplace, the control plane, and everything customer-facing.
When do I get paid?
Payouts are issued monthly based on verified utilization. Your dashboard shows live utilization and accrued earnings between payouts.
Become a provider
Put your hardware to work
Tell us about your hardware and we'll reach out with next steps to get your nodes onto the network and earning.