Postgres for serverless apps
With thousands of connections, real-time autoscaling, and usage-based pricing
Why use Neon for serverless architectures:
- CPU, memory, and storage scale automatically with your workload. 
- Usage-based pricing eliminates overprovisioning. 
- Supports up to 10,000 connections via its built-in connection pooler. 
- It’s 100% Postgres, without the server management. 
Sign up to get the Free Plan, plus $100 in credits
...or talk to our team if you have any questions.
Sign upThousands of connections and real-time autoscaling
Neon autoscales CPU and memory in real time based on your workload. Our autoscaling algorithm dynamically allocates compute resources, reducing your bill — you no longer need to provision for peak — while ensuring good performance.
Neon integrates a pooler built on PgBouncer directly into its architecture: every endpoint has connection pooling enabled. Supports up to 10,000 concurrent connections.
Neon worked out of the box, handling hundreds of Lambdas without any of the connection issues we saw in Aurora v2. On top of that, it costs us 1/6 of what we were paying with AWS.
 Cody JenkinsHead of Engineering at Invenco
Cody JenkinsHead of Engineering at InvencoTrue serverless Postgres
 - Separation of compute and storage- Neon’s architecture decouples compute and storage layers. Compute nodes handle query execution, while the storage layer persists data on distributed, highly durable object storage. - Elastic scaling 
- Durable storage 
- Optimized performance 
 
 - Instant provisioning- You can spin up Postgres databases in seconds, a big win when integrating Neon into database-per-user platforms and AI agents. - Fast deployment 
- Ready in seconds 
- AI & SaaS-friendly 
 
 - No server management- In Neon there’s no provisioning or manual resizing of servers. The platform automatically scales resources based on workload demands, not only up but also down. - Hands-free scaling 
- Auto-scaling 
- Zero maintenance 
 
  - Typical CPU utilization pattern in a production database in RDS. Traffic peaks once per day up to 60% capacity, going down to 10% capacity for the rest of the day. Based on a real use case. - Pay for what you use- Your costs are directly tied to the resources your workload consumes — both compute and storage. There's no need to over-provision or pay for idle capacity. - Cost efficiency 
- Peak-aware scaling 
- Optimized compute 
 
Instead of having to overprovision our servers to handle peak loads, which leads to inefficiencies and higher costs, Neon’s autoscaling handles it. We get more performance when we need it.
 Julian BenegasCEO of BaseHub
Julian BenegasCEO of BaseHub