Carl Lerche created Tokio. He's been deep in async Rust longer than almost anyone. So when he talks about where async is going, it's worth listening.
The Fragmentation Problem
For years, Tokio was the answer. Need async? You got Tokio. It became so dominant that "async Rust" and "Tokio" nearly became synonyms.
But that's changing fast.
The new wave of async runtimesāglommium, axum, async-stdāaren't fighting Tokio for the same use case. They're optimizing for different constraints: embedded systems, WASM, simplicity over features, single-threaded execution.
The fragmentation isn't a bug. It's a sign that async Rust has matured enough that one-size-fits-all no longer works.
What Tokio Got Right (and Wrong)
Tokio nailed reliability. It powers the most demanding Rust systems in productionādatabases, distributed systems, hyper-scale services. When you need guarantees, Tokio delivers.
But those guarantees come with complexity. For many applications, you don't need everything Tokio offers. You need "good enough" async with a smaller footprint.
This is the gap new runtimes are filling.
The High-Level Shift
The more interesting shift isn't the runtimeāit's what's built on top of it.
Frameworks like axum (by the Tokio team) are moving async Rust up the stack. You don't think about futures or task spawning. You define handlers, return types, and the framework handles the rest.
This is where async Rust is going: less boilerplate, more application logic.
What This Means for You
If you're starting a new Rust project in 2026:
-
Don't default to Tokio unless you need it. For APIs, web services, moderate concurrencyāaxum with its default runtime is probably enough.
-
Consider your deployment target. If it's WASM, look at glommium or pure async-std. If it's embedded, smol might fit better.
-
Think about what you're optimizing for. Throughput? Memory? Compile times? Simplicity? The "right" runtime depends on your constraints, not community consensus.
The async Rust ecosystem is growing up. That's a good thingāeven if it means more choices.