Compare Go with Java, Rust, and Python for backend services: concurrency model, memory management, error handling and deployment. When would you not choose Go?

• Concurrency: Go has goroutines on an M:N scheduler plus channels. Blocking-style code scales to 100k+ tasks without function coloring. Java 21+ virtual threads get close. Rust uses async/await with an executor (Tokio): zero-cost and very fast, but colored functions, Send/'static bounds and a steeper learning curve, with data races ruled out at compile time. Python's asyncio is colored and CPU work is limited by the GIL (the free-threaded 3.13+ builds are still maturing), so it scales mostly with processes.
• Memory: Go has a concurrent, non-moving GC tuned for low latency (sub-ms pauses; GOGC/GOMEMLIMIT), value types and escape analysis. Java has strong generational/ZGC collectors but a higher baseline footprint and JIT warm-up. Rust uses ownership with no GC, which gives predictable latency and the lowest memory. Python uses reference counting plus a cycle GC, with high per-object overhead.
• Errors: Go returns explicit error values (verbose but visible). Java uses exceptions (checked and unchecked). Rust's Result with ? is explicit, concise, and checked for exhaustiveness. Python uses exceptions.
• Deployment: Go builds a single static binary, cross-compiles trivially, starts fast, and fits in small containers. Java needs a JVM (or GraalVM native-image). Rust also produces a static binary, but compiles slowly. Python needs an interpreter plus a dependency environment.

When not to choose Go: hard real-time or ultra-low-latency systems where any GC pause matters, or embedded/no-OS targets (Rust, C++). Heavy data science or ML work that depends on the Python ecosystem. Domains with rich type-level modeling needs such as compilers or complex business rules (Rust, Kotlin or Scala sum types). Teams heavily invested in JVM libraries and frameworks. Prototypes where time to first demo matters most. Interviewer angle: give a balanced answer. Go's strengths are simplicity, fast builds, operability and good-enough performance, not raw speed or expressiveness.