[01.00] KUBERNETES + UNIKRAFT

Cut your K8s infra bill by 50-90%.

Unikraft drops into your existing Kubernetes cluster, on any cloud or on-prem. Your pods run as microVMs that boot in 10 milliseconds and scale to zero when idle. One Helm install, nothing else changes.

01 · Savings
50-90%
off your infra bill
02 · Cold Start
<10ms
all workloads
03 · Setup
1 command
helm install, done
Talk to an engineer → Watch the demo

[02.00] SEE IT IN ACTION

From helm install to 14ms boot times

Your existing cluster. Your existing YAML. Pods now boot in milliseconds instead of seconds.

[03.00] IN PRODUCTION

Browser Use cut their infra costs and ditched Kubernetes.

80K+ GitHub stars. $17M seed. Hundreds of concurrent Chrome instances for AI agents. This is what their switch looked like.

Cost per browser
~$0.30/hr
Bare-metal economics
Cold start time
8ms
Out-of-the-box scale-to-zero
Concurrency
100s
of Chrome instances on bare metal
"We tried Vercel Functions, Kubernetes, Docker, third-party browser providers. None of them could handle hundreds of concurrent Chrome instances at the cost and performance we needed. With Unikraft on bare metal, we finally had infrastructure that could actually keep up."
Gregor Zunic — Co-Founder & CTO, Browser Use
Before After
Expensive per-invocation pricing ~$0.30/hr per browser on bare metal
Container isolation (shared kernel) VM isolation per session
No scale-to-zero Scale-to-zero out of the box
Browsers too expensive at scale Hundreds of concurrent Chrome instances
Read the full case study →

[04.00] HOW IT WORKS

Works on your infra. Takes 5 minutes.

You install Kraftlet via Helm. It registers as a node in your existing cluster. Pods scheduled to it run as microVMs instead of containers. Your cloud, your on-prem, your VPC. We don't care.

  • Runs on AWS, GCP, Azure, on-prem, or dedicated hosts. Your choice.
  • Idle pods scale to zero. You stop paying for stuff that's not running.
  • Same kubectl, same YAML, same CI/CD. Nothing breaks.
Read the docs →
~/cluster
# Install it
$ helm install kraftlet \
  oci://ghcr.io/unikraft-cloud/helm-charts/kraftlet

# It shows up as a node
$ kubectl get nodes

NAME        STATUS   ROLES
Kraftlet    Ready    agent

# Deploy like normal. Pods boot in ~14ms.
$ kubectl apply -f my-app.yaml
$ kubectl get pods

NAME                      STATUS    BOOT TIME
my-app-78c766fb67-k2dgk  Running   14.06 ms
my-app-78c766fb67-tkkxq  Running   14.43 ms

[05.00] NEXT STEP

How much could you save?

15 minutes with an engineer. Bring your infra bill, we'll tell you what changes and what doesn't.

Talk to an engineer →