Lambda network speed jumps up to 4.8x — scaling from 625 to 3,000 Mbps

Hi, I'm Shii-chan! Today I've got a small but potentially very impactful update from AWS. Lambda functions can now scale their network bandwidth up to 3,000 Mbps! When I saw that number, I could tell a lot of people have been waiting for this.

AWS What's New aws.amazon.com

What was announced?

AWS's What's New announced scalable network bandwidth for Lambda functions outside a VPC. True to the What's New format, it's a straightforward announcement of a new feature going live.

This applies to Lambda functions running outside a VPC with 2 GB of memory or more. Previously, the bandwidth cap was a flat 625 Mbps no matter what. Now, bandwidth scales proportionally with memory configuration — 625 Mbps at 2 GB, up to 3,000 Mbps at 10 GB. That's up to about 4.8x faster!

The story so far

Until now, Lambda functions outside a VPC were capped at 625 Mbps of network bandwidth per execution environment, regardless of how much memory you configured.

Lambda is commonly used for latency-sensitive data processing workloads that pull in data from external sources — sometimes several terabytes' worth — for processing. As data volumes and performance requirements grew, that 625 Mbps ceiling became a real constraint on transfer speed, even if you'd cranked up memory (and CPU) for the function.

What changes

For data-heavy workloads, you can expect shorter execution times, since network transfer is no longer stuck at the old cap. Since Lambda billing is tied to execution duration, faster transfers can also mean lower cost per invocation. And naturally, this should translate into a better end-user experience too.

If your workloads pull large files from external APIs, move a lot of data to and from S3, or do anything with a streaming-like data flow, this is a genuinely useful upgrade if you've been network-bound.

Dive Deep

This isn't enabled by default — you need to request it through AWS Service Quotas, specifically the "Network bandwidth per execution environment" quota. Once enabled, bandwidth scales automatically based on each function's memory configuration, for all functions outside a VPC in your account.

The two anchor points given are 625 Mbps at 2 GB and 3,000 Mbps at 10 GB, scaling proportionally in between. There's no additional charge, and it's available in all commercial AWS Regions. For more detail on the limits, AWS points to the Lambda quotas page.

Wrap-up

• Lambda functions outside a VPC (2 GB+ memory) can now scale their network bandwidth based on memory configuration
• Bandwidth ranges from 625 Mbps at 2 GB up to 3,000 Mbps at 10 GB — up to about 4.8x faster
• Great for latency-sensitive, data-heavy workloads: expect shorter execution times, lower per-invocation cost, and a better end-user experience
• You need to request the "Network bandwidth per execution environment" quota via AWS Service Quotas to enable it
• Free of charge, available in all commercial Regions

If you've been running data-heavy Lambda functions and suspected network was your bottleneck, this update is worth checking out!