[Lambda][Node.js] Pointer compression / memory-optimized runtime option for large bundles and source maps #90

Hi Lambda team,

We hit a practical memory wall with Node.js Lambda and wanted to ask if pointer compression (or an equivalent runtime-level memory optimization) is on the roadmap for managed Node runtimes.

Our use case is a backend Lambda with a single-file minified bundle. Normal artifact size is ~7.6 MB. If we enable source maps for debugging (NODE_OPTIONS=--enable-source-maps), artifact size jumps to ~31.7 MB (index.mjs + index.mjs.map).

What we observed in real traffic tests:

• Runtime: Node.js 24.x
• Arch: arm64
• Memory: 256 MB
• Without source maps: cold starts are acceptable for our use case
• With source maps enabled: cold start increases, and we started seeing Runtime.OutOfMemory at 256 MB in the editor/debug environment

We can work around this by disabling source maps or increasing memory, but this is a tough tradeoff:

• disabling source maps hurts debuggability
• increasing memory increases cost for workloads that are otherwise lightweight

Request:
Could Lambda provide either:

1. managed Node runtimes with V8 pointer compression enabled, or
2. another runtime-level memory optimization path for Node workloads with large bundles/maps?

Even an experimental/runtime flag path would already help for debugging-heavy environments.

Thanks for considering this.