Realtime messaging platform

Realtime messaging, hosted your way.

Power the live parts of your product — chat, notifications, presence, live data — with delivery you can count on. Run it on your own infrastructure, or let us host it for you.

Self-hosted available now Managed coming soon
one message · every subscriber
Built for scale

Realtime you can put on the critical path.

Fan-out that scales to tens of millions of messages a second and millions of connections — with delivery guarantees that hold under load.

40M+
messages / second
Fleet throughput, linear from the measured unit.
Millions
concurrent connections
Scale by adding nodes and regions.
<250ms
p99 delivery
Tens of milliseconds typical, coordination-free.
Zero
messages lost silently
Exactly-once and in order, or an explicit GAP.

Throughput and connection figures are measured on a single capacity unit and scale linearly across the fleet — re-baseline on your hardware. 875,000 messages/second sustained from one node; delivery guarantees verified in fault drills (zero sequence forks).

Hosting

Two ways to run it.

Same platform, same guarantees. The only question is who runs the infrastructure.

Available now

Self-hosted

Run NextSpark in your own Kubernetes cluster, alongside your other services.

  • Your data stays in your network
  • Full control, no per-message fees
  • License by subscription, usage, or one-time
See pricing
Coming soon

Managed by us

We host, scale, and operate it for you, so your team ships features instead of running servers.

  • No infrastructure to manage
  • Upgrades, scaling, and on-call handled
  • Usage-based pricing
Join the waitlist
Use cases

One engine, many realtime products.

NextSpark is the connection and fan-out layer under whatever you are building.

Live chat & messaging

Rooms, direct messages, presence, and typing — delivered in order at chat scale.

Live dashboards & data

Stream metrics, prices, and events to thousands of viewers without polling.

Multiplayer & game state

Fan out moves and state to every player with low, predictable latency.

Telemetry & IoT

Ingest and broadcast device streams, with durable capture for replay.

Notifications & feeds

Push activity, alerts, and updates the moment they happen.

Collaborative apps

Keep documents, boards, and cursors in sync across every session.

Get started

Running in your cluster the same day.

Ships as signed container images with a published Helm chart. Install it, prove delivery with the built-in test, and point your clients at it.

# deploy into your own cluster
helm install nextspark oci://ghcr.io/xunholy/charts/nextspark

# prove delivery end-to-end (subscribe → publish → receive)
helm test nextspark --timeout 5m

# your clients connect over standard WebSockets
const ws = new WebSocket("wss://edge.your-domain.com");
How it compares

The short version.

Against the realtime services teams ask us about most. Honest about where they lead, too.

yes   ~ partial   no   coming soon

Security & control

Built for teams who own their stack.

Your data path, hardened by default — and entirely yours when you self-host.

Your data, your network

Self-host and your message path never leaves your VPC. The only egress is a documented daily version check — off with one flag.

Encrypted end to end

TLS 1.3 to clients, mutual TLS between services, and a zero-trust network policy.

Verified supply chain

Distroless, non-root images — signed, SBOM'd, and CVE-scanned in the release pipeline.

Pricing

Buy it the way that suits you.

Self-host free for development; enterprise licensing (annual or capacity) is available today. Managed hosting is coming soon.

Common questions

The things teams ask us.

Can you host it for us?

Self-hosting is available today. A fully managed service, where we run and scale NextSpark for you, is on the way — join the waitlist and we'll let you know when it's ready.

How is this different from Pusher or Ably?

Those are good managed services with a global edge network. NextSpark gives you a choice: host it yourself to keep your data in-house and your costs predictable, or (soon) let us run it. Either way, messages arrive in order by default.

What does support include?

Onboarding, a direct support channel, and security patches on every plan. Subscription and perpetual add an SLA and an architecture review.

Can we try it first?

Yes. We'll help you run a proof of concept against your own workload before you commit to anything.

Get started

Let's see if it's a fit.

Tell us what you're building and how much traffic you expect. We'll help you figure out the right setup and what it would cost.