VPC peering vs Transit Gateway cost: a deep dive

Quick answer

VPC peering has no hourly charge and bills only for data transfer between VPCs, so it is cheapest for a small number of VPCs connected point-to-point. Transit Gateway adds an hourly charge per VPC attachment plus a per-GB data processing fee, but it replaces a tangle of peering connections with a hub-and-spoke design that scales cleanly as VPC count grows. The crossover is the number of VPCs and connections: few VPCs favor peering on cost, many VPCs favor Transit Gateway because full-mesh peering becomes unmanageable and its connection count explodes, even though each peering link itself has no hourly fee.

Connecting VPCs is a networking decision with a real cost dimension, and the two main options price very differently. VPC peering is nearly free to establish and charges only for the traffic that flows. Transit Gateway charges an hourly fee per attachment and a per-GB processing fee, but it tames the complexity of connecting many VPCs. The cheaper choice depends heavily on how many VPCs you are wiring together.

How each one bills

Factor	VPC peering	Transit Gateway
Hourly fee	None	Per VPC attachment
Data transfer	Per GB between VPCs	Per GB, plus data processing fee
Topology	Point-to-point, full mesh to connect all	Hub-and-spoke
Scales to many VPCs	Poorly, connection count explodes	Cleanly

VPC peering has no hourly charge, you pay only for data transfer between peered VPCs. Transit Gateway charges an hourly fee for each VPC attachment plus a per-GB data processing fee on traffic through it, on top of data transfer. So on a pure per-connection basis, peering is cheaper, its cost advantage is having no fixed hourly fee at all. The catch is what happens as the number of VPCs grows.

Why peering does not scale

VPC peering is point-to-point and not transitive: to fully connect N VPCs, you need a mesh of connections that grows roughly with the square of N. At a handful of VPCs this is fine and cheap. At many VPCs it becomes an unmanageable tangle of connections to create, route, and maintain, and the operational burden (and route-table complexity) outweighs the lack of hourly fees. Peering is cheapest only while the number of VPCs stays small.

Why Transit Gateway scales

Transit Gateway is a hub: each VPC attaches once, and the gateway routes between all of them, so connecting N VPCs needs N attachments, not a full mesh. That hub-and-spoke model scales cleanly and stays manageable as VPC count grows, which is worth its hourly and processing fees once you have enough VPCs. It also simplifies connecting on-premises networks and centralizing egress, reducing the NAT gateway sprawl that per-VPC internet access causes.

Choosing the topology on cost

For a few VPCs, peering is cheaper and simple. As VPC count grows, the mesh becomes unmanageable and Transit Gateway's clean scaling justifies its fees, and its centralization can reduce other costs like duplicated endpoints and NAT gateways, echoing the PrivateLink consolidation argument. Cross-region or cross-AZ traffic between VPCs also carries data-transfer fees, one of the unexpected charges to watch. Estimate your VPC count, connection needs, and inter-VPC traffic, then price both topologies. Price your network design against the resource catalog so the topology cost is clear before you build it.

FAQ

Is VPC peering or Transit Gateway cheaper?

For a small number of VPCs, VPC peering is cheaper because it has no hourly charge and bills only for data transfer, while Transit Gateway adds an hourly fee per attachment plus a per-GB processing fee. For many VPCs, Transit Gateway is usually better because full-mesh peering becomes unmanageable as the connection count explodes, so its clean hub-and-spoke scaling justifies the fees despite peering's lack of hourly cost.

How is Transit Gateway priced?

Transit Gateway charges an hourly fee for each VPC attachment plus a per-GB data processing fee on traffic flowing through the gateway, on top of standard data transfer charges. So its cost has a fixed component (per attachment) and a variable component (per GB processed), unlike VPC peering which has no hourly fee. These fees buy clean hub-and-spoke connectivity that scales to many VPCs.

Why does VPC peering not scale to many VPCs?

Because peering is point-to-point and not transitive, so to fully connect N VPCs you need a mesh of connections that grows roughly with the square of N. At a handful of VPCs this is cheap and manageable, but at many VPCs it becomes an unmanageable tangle of connections to create, route, and maintain, and the operational and route-table complexity outweighs peering's lack of hourly fees.

When should I use Transit Gateway?

When you connect enough VPCs that a full mesh of peering connections becomes unmanageable, typically beyond a handful. Transit Gateway's hub-and-spoke model needs only one attachment per VPC and routes between all of them, scaling cleanly. It also simplifies connecting on-premises networks and centralizing egress, which can reduce duplicated endpoints and NAT gateway sprawl, adding cost savings beyond just connectivity.

Can Transit Gateway reduce other networking costs?

Yes. By centralizing connectivity and egress, Transit Gateway can reduce NAT gateway sprawl (fewer per-VPC internet paths) and duplicated VPC endpoints across many VPCs, consolidating traffic through a shared hub. These savings can offset its hourly and processing fees at scale, similar to how consolidating PrivateLink endpoints via shared-services networking reduces per-VPC duplication. The centralization value grows with VPC count.

How does C3X help choose a network topology?

C3X prices your networking infrastructure, including VPC peering, Transit Gateway attachments, and their fees, from Terraform before you deploy, so the cost of each topology is visible in the pull request. That lets you compare peering against Transit Gateway for your VPC count and traffic at design time, choosing the topology on total cost and scalability rather than discovering the fees after building the network.

What to do next

Compare network topologies on cost before you build. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.