Make money from idle GPUs.

Install Lilac once, keep full control of your cluster, and earn from idle capacity.

Terminal

$ kubectl apply -f lilac-provider.yaml
→ Installing Lilac operator...
→ Detecting idle GPUs...
→ Applying your preemption and payout policy...
✓ Idle GPU nodes ready for paid workloads

Stop leaving GPU-hours on the table.

Research shows the average data center runs at just 50% GPU utilization. Enterprise on-prem clusters average as low as 10%. That's thousands of GPU-hours wasted every month.

Revenue potential

  • Small cluster
    $1.4K–4.3K
    est. monthly revenue
    8 GPUs · ~50% idle
    2,880 GPU-hours wasted

  • Mid-size cluster
    $6K–17K
    est. monthly revenue
    32 GPUs · ~50% idle
    11,520 GPU-hours wasted

  • Large fleet
    $23K–69K
    est. monthly revenue
    128+ GPUs · ~50% idle
    46,080+ GPU-hours wasted

Utilization data from LBNL, CRS, and ACM PEARC 2025. Revenue based on $0.50–$1.50/GPU-hr through optimized model hosting and inference routing.

Everything you need to monetize idle compute.

Enterprise-grade infrastructure with zero operational overhead.

Features

Auto-Preemption

Your workloads always come first. The moment internal demand returns, GPUs are reclaimed instantly — no manual intervention, no downtime, no risk to your production jobs.

One Command Install

kubectl apply and you are live. No agents to manage, no sidecars to configure, no complex onboarding. Lilac runs as a lightweight operator inside your existing Kubernetes cluster.

Full Control

Your cluster, your rules. Define exactly which pools to monetize and when.

Clear Payouts

Usage-based revenue with transparent metering and detailed reporting.

Real-time Monitoring

Track GPU utilization, earnings, and workload status from one dashboard.

Enterprise Security

End-to-end encryption with isolated workloads and compliance-ready architecture.

The simplest path from idle to earning.

Purpose-built for GPU operators who want revenue without complexity.

Capacity lifecycle

From idle to earning and back.

Your GPUs cycle between your workloads and monetized capacity.

Step 1
Idle node detected
Finds unused GPU capacity.

Step 2
Workload deployed
Deploys approved workloads to available GPUs.

Step 3
Usage metered
Tracks traffic and earnings for payout reporting.

Step 4
Returned on demand
GPUs are handed back immediately to your workloads.

Your GPUs are yours first.

Lilac only uses capacity you explicitly expose. Reclaim GPUs immediately when internal demand returns.

  • Your workloads always keep first priority.
  • Instant preemption returns GPUs when you need them.
  • You define where and when monetization runs.
  • Transparent metering and payout reporting.

Monetize idle GPU hours without giving up control.

Onboard once, set policy, and generate supplier revenue from capacity you're already paying for.