
Thinking Elixir Podcast 286: A NextJS Escape and 2025 in Review
18 snips
Jan 6, 2026 A rapid NextJS to Phoenix and Ash migration delivered in two weeks sparks a discussion on developer workflows. Big Credo speedups on large codebases get attention. Time-series forecasting with Soothsayer and Torchx running on Apple Metal highlight ML progress. An Erlang DNS case study shows extreme-scale reliability. A wide 2025 retrospective covers interoperability, tooling, and ecosystem trends.
AI Snips
Chapters
Transcript
Episode notes
Apple Metal Support For Torchx
- Nx now enables Torchx to run on Apple Metal, opening efficient ML on M-series Macs.
- This expands Elixir ML workflows to use local Apple hardware for GPU-accelerated models.
Erlang Powers Massive DNS Infrastructure
- The EEF case study shows DNSimple uses Erlang to handle hundreds of billions of DNS queries monthly with sub-60s zone propagation.
- This validates Erlang/BEAM for globally distributed, high-throughput, low-latency services.
2025: Types And AI Integration
- 2025 saw major progress on set-theoretic types in Elixir, moving toward near-complete type coverage.
- The ecosystem also embraced AI/ML integrations like Livebook, Nx, and Python X to connect to other toolchains.
