Netflix remains heavily invested in the Java ecosystem, utilizing it as the primary backbone for its global streaming architecture and studio operations
This 2026 update details the company's "paved road" approach to Spring Boot, its transition to modern JDK features like Generative ZGC, and the integration of agentic workflows using Spring AI to enhance developer productivity and system performance

play_arrow3:08 Netflix manages two distinct types of Java applications: high-scale streaming services requiring low latency and multi-region failover, and traditional enterprise apps used for movie studio operations
play_arrow11:33 The "paved road" is built on Spring Boot, providing standardized modules for security, dynamic configuration, and observability that integrate seamlessly with Netflix's internal infrastructure
play_arrow5:03 The streaming discovery phase uses Federated GraphQL for flexible data fetching, while internal server-to-server communication relies on gRPC for high-performance, method-oriented interaction
play_arrow8:06 Java is chosen as the primary backend language because it offers the best trade-off between runtime performance, developer productivity, and maintainability. While JavaScript is used for UIs and Python for data science, all core microservices are Java-based
play_arrow30:06 Netflix has transitioned to Generational ZGC as its default garbage collector. This change eliminated "stop-the-world" pause spikes, significantly reducing IPC client timeouts and retry-induced cluster load
play_arrow34:39 After a "false start" due to thread pinning issues in earlier versions, Netflix is re-adopting Virtual Threads with JDK 25, integrating them into Tomcat connectors and DGS data fetchers
play_arrow37:25 Utilizing Structured Concurrency requires careful handling of ThreadLocals. Netflix uses a custom thread factory approach to ensure security and tracing contexts are correctly propagated to new virtual threads
play_arrow21:19 To manage thousands of services, Netflix is using agentic workflows (via Cloud) to automate Spring Boot and JDK upgrades. This replaces manual search-and-replace tasks with a multi-step, prompt-based reasoning system
play_arrow41:30 The company is moving beyond simple chat completions to "agentic workflows" where the developer maintains control over the logic loop while using LLMs for specific tasks like data analysis or routing
play_arrow45:40 Netflix built an internal tool using Spring AI that analyzes profiling data to identify slow-starting Spring beans. The agent provides actionable recommendations and analyzes library source code to suggest specific fixes, such as making certain initializations asynchronous
Netflix continues to leverage Java's evolving feature set to maintain its massive scale, prioritizing frequent version upgrades to avoid technical debt. By combining high-performance JVM tuning—specifically Generational ZGC—with modern automation tools like Spring AI and automated migration agents, the platform organization ensures that thousands of microservices remain performant, secure, and developer-friendly