Comparing AWS Lambda Arm64 vs x86_64 Performance Across Multiple Runtimes in Late 2025

Summary

The article compares the performance of AWS Lambda functions on arm64 and x86_64 architectures using various runtimes. Rust on arm64 is identified as the most performant and cost-efficient combination overall.

Key Takeaways

1. Rust on arm64 is the most performant and cost-efficient combination overall.
2. Node.js 22 on arm64 was consistently faster than Node.js 20 on x86_64.
3. Across the board, arm64 delivered roughly 30–40% lower compute costs with equal or better performance than x86_64.

Statement Breakdown

• Claimed Facts: 70% of statements the article presents as facts
• Opinions: 20% of statements classified as editorial or subjective
• Claims: 10% of statements surfaced for additional reader evaluation

Credibility & Bias Reasoning

Credibility assessment: The article presents a well-defined benchmark methodology and open-source code, enhancing transparency. The author discloses potential biases and limitations, such as the focus on single-threaded workloads. The results are based on multiple test runs and statistical analysis, increasing confidence in the findings.

Bias assessment: Performance Optimization Focus. The article is primarily focused on optimizing performance and cost-efficiency within AWS Lambda. While the author acknowledges potential limitations and biases, the overall tone is geared towards identifying the best configurations for specific workloads. This focus may lead to overlooking other factors relevant to Lambda usage.

Note: The article presents benchmark results with a clear methodology. Consider your specific workload when applying these findings.

Credibility flag: Data-Driven Insights

Claimed Facts (7)

• This is a factual statement about AWS's introduction of arm64 support.
• This describes the scope of the benchmark.
• This provides release dates for the runtimes tested.
• This specifies the sample sizes used in the benchmark.
• Describes the components of the light workload test.
• Describes the CPU-intensive workload.
• Describes the memory-intensive workload.

Opinions (6)

• This is a personal endorsement of serverless services.
• This expresses the author's personal interest in Lambda's development.
• This is a subjective assessment of the value of generic benchmarks.
• This is a recommendation based on the benchmark results.
• This is a subjective decision regarding the choice of runtime.
• This is a personal reason for including Rust in the benchmark.

Claims (5)

• While potentially true, this statement lacks context and specific data to verify the claim's relevance to the benchmark.
• This is a subjective assessment without concrete evidence.
• The term "free" is an oversimplification and could be misleading.
• The surprise element suggests a potential bias or unexpected result that requires further scrutiny.
• This is a speculative explanation for potential cold-start penalties.

Key Sources

• Chris Ebert — Author
• Amazon Web Services (AWS) — Cloud Computing Platform