Managed Postgres, search, secrets, message queue, chat and app updates, all from one cloud with one API. You pay only while your services are in use.
Create a service with one API call and you get an endpoint. We run it, scale it, back it up and upgrade it. When nobody is using it, it scales down to zero and you pay nothing for compute. The next request starts it again, and your data is still there. Or define the usage contract and run at defined times for as mnay requests as you wish.
1. You don't want to run infrastructure.
You have a product to ship and a small team. Running Kubernetes, databases, backups and upgrades yourself takes engineers away from that product. On Hyperion Cloud there is nothing for you to operate. You call the API and use the endpoint.
2. You pay for services that sit idle.
Staging, preview branches, internal tools and side projects run all day and night on most clouds. Non-production environments or idle time in prod often costs more than 30–45% of a cloud bill. Hyperion services scale down to zero and start again on the first request or contract defined with API. An idle environment costs you sleeping storage only.
3. "Serverless" databases still charge a minimum, or make you wait.
Some serverless Postgres offerings keep charging a minimum at all times. Some take many seconds to resume. On Hyperion Cloud you choose for each service: keep one instance always running, or scale fully to zero. Either way, no data is lost.
NOTE. We use our version of Postgres, because original Postgres is not designed for autoscalability. The downside is that we don't use extensions. We need to intgerate extensions by hand. You can open the issue on this repo to get support for some extension.
NOTE. Our search is running on Postgres, do not request pg_search integration (integrated, fully featured form of hsearch, open source in this org)
4. Managed services are expensive.
Most managed database providers run on full-price cloud machines and pass that cost on to you. Hyperion Cloud partly runs on spot instances or Hetzner, which cost a fraction of full price. Our platform keeps data off the machines themselves and moves services to another machine when one is reclaimed, so running on spot does not risk your data. We pass the savings on in our prices.
5. Every backend piece is a separate vendor and a separate bill.
Postgres from one provider, search from another, a queue from a third, secrets from a fourth. Each has its own pricing tiers, dashboard and account. On Hyperion Cloud every service is in one account, created the same way and billed on the same model.
6. Traffic between clouds is hard to secure.
When your services run with more than one provider, the traffic between them crosses the public internet. Securing it usually means setting up VPNs, peering or certificates yourself. On Hyperion Cloud, services talk to each other over WireGuard tunnels, which are encrypted, across every cloud and data center we run on. You do not set anything up. We'll be multicloud if you wish.
7. You have to guess how big your servers should be.
On most managed services you pick an instance size or pricing tier up front. Pick too small and a traffic spike slows or stops your service. Pick too large and you pay for capacity you don't use. Moving to a bigger tier is often a manual step, and sometimes it means downtime. On Hyperion Cloud, services autoscale: they add servers as load rises and remove them as load falls, with no action from you.
| Service | What it is for |
|---|---|
| Postgres | Primary database, with replicas |
| Search | Full-text search (on Postgres) |
| Vault | Secrets and credentials |
| Message queue | Events and background jobs between services |
| Chat | Real-time messaging for your users (MQTT) |
| Updates | Delivering app and service updates (MQTT) |
- One API. One request creates a service; another scales it or deletes it.
- Autoscaling. Services add servers as load rises and remove them as load falls. No instance sizes or tiers to pick.
- Scale to zero, one, or many. Each service can sleep when idle, keep one instance always running, or run on as many servers as the load needs.
- Starts on first request. A sleeping service wakes when traffic arrives. Your code does not need to handle it.
- No data loss. Data is stored separately from compute, so scaling down and machine changes do not lose it.
- Nothing to operate. Backups, upgrades and failover are done by us.
- Encrypted traffic between clouds. Services connect over WireGuard across every provider we run on, with no VPN setup on your side.
- Startups that want production services without an operations team.
- Teams with many environments (staging, previews, per-customer) that sit idle most of the day.
- Teams tired of paying for five vendors to run one backend.
Will be live 2027.