Reduce allocations & fix hashing in turbopack trace-server bottom-up pass

Executive Summary

Many observability and build-tool teams running turbopack's trace-server experience excessive allocations and a suboptimal hashing pass in the bottom-up aggregation step, which inflates CPU, memory and downstream storage costs during trace processing. This is particularly painful for enterprise dev orgs building edge-first web apps and global CDNs — roughly 40,000 potential enterprise accounts in a $12.0B market where organizations spend on the order of $300K ACV for observability and trace-processing optimizations. You could build a Rust-based reimplementation of the bottom-up pass that eliminates unnecessary allocations, fixes the hashing algorithm, and ships as a drop-in native plugin or FFI module for turbopack trace-server; realistic targets are a 2–5x reduction in peak memory and CPU during aggregation and commensurate reductions in ingest/storage spend. The product would include reproducible benchmarks, compatibility shims for existing trace schemas, an open reference implementation and an enterprise support package. The timing is favorable: edge-first architectures are increasing trace volume and cardinality, Rust adoption for performance-critical components is growing, and rising cloud/ingest costs are pushing teams to optimize trace processing — factors reflected in a market score of 88/100 and revenue potential of 76/100. With medium competition and a $12B addressable market, securing 10–100 pilot customers could validate the value proposition and support commercialization. This approach can stand out by marrying low-level systems optimization (allocation elimination and correct, fast hashing) with pragmatic integration and support, but it faces honest challenges: proving reliability across heterogeneous trace pipelines, keeping up with turbopack internals, and doing the sales and engineering work needed to turn technical wins into enterprise contracts.