WebAssembly

Imagine you want to run the same program quickly on a web page, on a server, and on a tiny edge device near a user - without recompiling it for each one and without trusting it to roam freely on your machine. WebAssembly, almost always shortened to Wasm, is a way to package compiled code in a single portable, low-level format that any compliant runtime can execute safely and fast. You write in C, C++, Rust, or other languages, compile to a .wasm file, and that one file runs anywhere a Wasm runtime exists.

Wasm is a binary instruction format for a simple, abstract stack machine: instead of registers like a real CPU, most instructions push and pop values on an operand stack (for example, i32.const 2, i32.const 3, i32.add leaves 5 on the stack). It is not the source code and not the final machine code - it is a compact bytecode in between, designed to be validated quickly and then compiled to the host's real machine code just before or while it runs. The design fixes a small set of value types (32- and 64-bit integers and floats, plus references), uses structured control flow (blocks, loops, and ifs rather than arbitrary jumps), and is engineered so a runtime can prove a module is well-formed before executing a single instruction.

It matters because it gives near-native speed with a strong sandbox and genuine portability, which is why it spread from browsers to servers, plugin systems, and serverless platforms. Be honest about the edges, though: Wasm is fast but usually a bit slower than fully native code; the core spec by itself has no access to files, the network, or threads - those come from interfaces layered on top (see WASI); and 'portable' means portable across Wasm runtimes, not that any random binary becomes Wasm for free. It is mature in the browser and rapidly maturing, but still moving, outside it.