The AI Stack
Sign in

Neon

Neon is a serverless Postgres platform with branching and autoscaling for developers and AI agents.

Updated April 2026

Overview

Website
neon.tech
Segment
Data Platforms & Warehouses
Posture
Specialized DB

Product overview

Neon provides serverless PostgreSQL with autoscaling compute, instant git-like database branching via copy-on-write storage, read replicas, and pgvector extension for vector similarity search, enabling AI applications like RAG. Used by developers, AI agent platforms (e.g., Replit, Databutton), and codegen tools for provisioning fleets of databases; acquired by Databricks in 2025. Distinct for separating compute/storage, scale-to-zero idle costs, API-driven fleet management, and no platform fees on enterprise features like HIPAA/SOC2.,

Revenue model

Usage-based pay-as-you-go: Free tier (0.5GB storage, 100 CU-hours/project); Launch ($0.106/CU-hr compute, $0.35/GB-month storage, $0.20/GB-month history); Scale ($0.222/CU-hr, same storage); CU ≈1 vCPU+4GB RAM, scale-to-zero, no minimums/platform fees. Extra branches $0.002/hour.

Moat

  • Cost Advantages
  • Scale Advantages
  • Data Flywheel
  • Brand
  • Network Effects

Neon, the leading Brazilian neobank, has a competitive moat primarily built on efficiency in operations, particularly in emerging market banking, enabling ~$1B run-rate revenue and 60% YoY growth while serving underserved lower- to middle-income customers. Additional strengths include customer intelligence for proactive retention and product development, data flywheel from millions of users, and scale advantages from major investors like General Atlantic and BlackRock.