What tech stack does Discord use?

Discord's backend is polyglot by design, with each language chosen for a specific purpose. Python handles general API services and business logic — it was the language the team started with in 2015 and remains the default for rapid iteration across most product surfaces. Elixir (via the Phoenix framework and its WebSocket layer) powers the real-time Gateway that relays messages and presence updates across millions of concurrent connections; Discord published extensively on this architecture starting in 2020. Rust was added for the highest-performance-critical components where garbage-collection pauses are unacceptable — most notably the Gateway itself, which was rewritten in Rust for deterministic latency at scale, and custom data structures that were integrated into Elixir services via the Rustler bridge. C/C++ surfaces in audio/video codec and voice-processing layers. Discord described its development environment as a polyglot mono-repo where Python, TypeScript, Rust, Elixir, and C/C++ are all actively maintained.

On data storage, PostgreSQL handles relational data (user accounts, server metadata, permissions), while Apache Cassandra stores the time-series message corpus — well-suited to Discord's append-heavy, chronologically ordered, high-volume write workloads. Redis powers session management, presence tracking, and caching at low latency. Apache Kafka is the event backbone for distributing state changes across services asynchronously. Cloudflare provides DDoS mitigation and global CDN at the network edge. Multi-region Kubernetes clusters on AWS and GCP provide container orchestration, with Terraform managing infrastructure as code.

The web client is built on React with Redux for global state management, written in TypeScript. The desktop application wraps this web client in Electron, giving Discord a cross-platform native-feeling app without maintaining separate codebases. Native mobile apps use Swift on iOS and Kotlin on Android, prioritizing platform-native performance over cross-platform frameworks like React Native or Flutter. Discord moved its entire engineering workflow to cloud development environments (CDEs), as documented on the company's engineering blog — reducing onboarding time and increasing environment parity between local and production.

Observability is a mature and well-documented layer: Datadog handles infrastructure monitoring and APM, Prometheus and Grafana cover metrics and dashboards, Elasticsearch and Kibana aggregate and index logs, Jaeger provides distributed tracing across Discord's microservice mesh, and Sentry catches client-side errors. The in-platform search experience is powered by Elasticsearch. Snyk handles dependency vulnerability scanning as part of Discord's security-first engineering culture. No CRM or marketing-automation vendor has been publicly confirmed — consistent with Discord's historically product-led, B2C distribution model, though commercial build-out under Sakhnini is expected to drive vendor evaluation in this area.

Discord's Rust-and-Elixir backend signals an engineering culture that prizes systems-level correctness, deterministic latency, and ruthless performance optimization. Vendors pitching into the CTO organization should lead with technical credibility, reference architecture fit with Kubernetes/Kafka/Cassandra, and ideally offer self-hosted or private-cloud deployment options for security-sensitive components. Pitching 'easy to use' without equivalent depth on reliability, observability, and scale will not land.

The most significant displacement and greenfield opportunities sit in three areas. First, security and compliance tooling: as Discord accelerates toward a public listing, SOC 2 Type II, ISO 27001, and SEC-mandated financial controls create immediate budget for GRC, identity, and vulnerability management vendors. Second, data warehouse and analytics: Discord's data volume from 259M MAU is enormous, and Cassandra's analytics capabilities are limited — a cloud data warehouse (Snowflake, BigQuery, Databricks) layer is a credible pitch. Third, GTM infrastructure: Discord's commercial build-out under Sakhnini will create first-time demand for enterprise CRM, sales engagement, and ABM tooling that did not exist at meaningful scale under the product-first Citron era.