Every language promises easy async. Rust delivers correct async — but that correctness comes with a cost. Before we even write async fn, let's understand why the borrow checker and futures interact the way they do.

A future in Rust is a state machine generated at compile time. The compiler creates an anonymous enum with a variant for every .await point. This means your future's size depends on what it borrows — and if you borrow something across an await point, you need Pin to keep it safe.

Self-referential structs are the crux of the problem. When async fn compiles into a future, it can reference local variables that were created before an .await. That's a self-referencing type. Without Pin, the compiler would have to prevent this entirely — and then async Rust wouldn't work at all.

The Tokio runtime solves this by providing tokio::spawn, which takes your future, wraps it in a JoinHandle, and pins it to the heap. That's why your top-level async functions usually end with .await — because spawned futures are automatically pinned.