Introducing Private Network Gateway

Databricks serverless compute is fully managed, intelligent, and versionless, allowing users to focus on solving business problems rather than sizing, tuning, and managing clusters and infrastructure. However, for many teams, the high-friction step that remained was networking: giving serverless compute a private path to databases, applications, and systems inside their own networks. In our customer interviews, more than 75% said their current network configuration and the complexity of dealing with network changes was the main reason specific workloads stayed on classic compute instead of moving to serverless.

Today, we are announcing the public preview of Private Network Gateway (PNG) on Azure, and private preview on AWS. PNG gives your serverless workloads one connection into your network, and from there, you control access to everything your network can already reach.

• Simplify ingestion across your private databases: use Lakeflow Connect on serverless compute to ingest from supported databases across your VNet/VPC and connected on-premises environments
• Connect AI agents to private services: An agent you build on Databricks can call a private MCP server you host
• Use your own package repositories: Serverless can install Python libraries from a private PyPI, Artifactory, or Nexus repository inside your network
• Route egress through your own firewall: Send traffic through your existing firewall or content inspection service, so egress is filtered and logged natively with your own tools
• Keep a consistent, customer-owned source IP: Your SaaS apps like Salesforce, can see a source IP that belongs to you

Each of these used to require coordinated deployment between networking, security, and admin teams, and multiple new deployments like load balancers and proxy VMs. With PNG, you configure one connection, and the resources your network can reach become reachable from serverless. PNG scales by network, not by endpoint.

Private Network Gateway has significantly simplified how we connect serverless compute to data sources on Azure and on-premises — no more standing up a private link service and load balancer for every connection. It also brings real flexibility to our serverless operations: we can reuse an existing workspace subnet, which makes firewall approvals easier, while still controlling exactly which destinations are reachable.—Fabiano Seki, Tech Lead, Data Engineering, Banco Bradesco

The challenge with serverless networking

Classic compute runs inside your own virtual network, so reaching your data is straightforward: you peer the networks using standard cloud primitives. In contrast, serverless runs in a Databricks-managed network, which, on one hand removes the operational overhead, but it also means your private resources need an explicit path.

Serverless Private Link provides that path, and it remains the right tool for cloud-managed services like AWS S3, or Azure Data Lake Storage. But for resources you host and manage, like databases, Kafka clusters, or internal APIs, connecting often meant a separate private endpoint, load balancer, and proxy VM for each resource — plus back-and-forth coordination between your Databricks admin, cloud admin, and network security team. For a handful of resources, this may be manageable. But when a team needs serverless to access dozens of databases, APIs, and internal services; across VNets, peered networks, and on-premises environments, the per-resource approach quickly becomes a bottleneck, making it painful, slow, and manually intensive.

How PNG works

The setup is easy and straightforward: You delegate one subnet in your virtual network to Databricks and PNG establishes a managed, encrypted gateway from serverless compute into that subnet. You configure the gateway inside a Network Connectivity Configuration (NCC), the same account-level object you already use for serverless networking. You can then attach the NCC to your workspaces, and supported serverless workloads can reach the private destinations you allow.

There are no gateway VMs for you to deploy, patch, or scale in your network. Databricks creates and runs the gateway for you. You keep control of your network.

Before PNG, enabling private connectivity for Databricks Serverless required significant network setup and coordination. PNG simplified the whole process, reduced operational overhead, and helped us run jobs faster—making serverless much easier to integrate into our existing environment.—Primož Gale, Lead Technical Architect, NLB

You keep control of the network boundary

PNG keeps the network boundary with you:

• Your gateway is yours: it provides a dedicated path into your network, not a shared appliance carrying other customers' traffic. Isolation is enforced via encapsulation and network security rules.
• The gateway enforces workspace-level access controls before any traffic leaves.
• Beyond the subnet, your route tables, network security groups, DNS resolvers, and security appliances decide what serverless can reach, the same way they do for the rest of your network.

You choose how much traffic PNG carries, through two modes:

1. specific_destinations sends only the fully qualified domain names you list through PNG. All other traffic keeps its normal path. Use this when you want precise control over which resources serverless can reach.
2. all_traffic routes outbound traffic through your network, except anything matching a more-specific Private Link or service-endpoint path, which keeps its native route. Use this mode to inspect broad egress through your own firewall, or to send it out through your own IP addresses.

PNG is additive. It sits alongside the private connectivity and egress controls you already use for Databricks and does not change them, so you can enable or remove it without affecting your existing setup.

Before Private Network Gateway, managing secrets and network connectivity for serverless workloads required multiple integrations and ongoing coordination. Today, with PNG, we can securely access internal services directly and centrally manage outbound traffic through our existing controls, making operations simpler, faster, and easier to govern.—Michel Schwandner, Platform Engineer, Julius Blum GmbH

Get started with PNG

Private Network Gateway is now available in Public Preview on Azure, and in Private Preview on AWS. Support for Google Cloud is in development.

Get started today:

• Private Network Gateway - Azure documentation
• For AWS, contact your Databricks account team to discuss enrollment

To learn more about what's new in Databricks Platform security, see What's new in Databricks Platform security and compliance at Data + AI Summit 2026.