The End of JS

A 2014 prediction about “the end of JavaScript” hinges on a single bet: if the browser gets an “assembly for the web” that compiles code into a low-level, type-safe execution format, then developers won’t need to write JavaScript directly—JavaScript will still power the runtime, but it will stop being the default language. The forecast matters because it frames today’s web platform shift as more than a tooling trend: it’s a potential change in what “native” web development looks like.

The talk’s technical backbone is Azam.js, described as JavaScript with enforced types—essentially a web-targeted assembly layer. The transcript illustrates the idea with simple numeric functions: inputs are explicitly converted to numbers, operations like multiplication are performed, and results are converted back. That structure is meant to let a JIT compiler collapse high-level intent into efficient machine-like instructions (the example compares it to turning a function into a single “multiply” operation). The point isn’t that the code is pretty; it’s that the runtime can optimize aggressively because the types and operations are predictable.

A key historical anchor comes from the early era of WebAssembly-style experiments, when high-performance engines could run in the browser at “playable” frame rates. The transcript recalls Unreal Engine 3 running inside Firefox, with a comparison to Chrome demos that lacked native support at the time. The broader takeaway: once a compiler can target the web assembly format, “anything that compiles” can run in the browser—C code, game engines, and system-like libraries—so the language barrier starts to fall.

From there, the prediction expands in stages. The first major milestone is around 2025, when extremely large applications would be ported to the assembly format. The transcript then challenges that timeline by pointing to AI-driven code generation and the 2020–2025 disruption period, arguing that the original “JavaScript frustration” catalyst didn’t arrive as expected. Still, the assembly ecosystem has evolved: the successor to Azam.js (referred to as Wom) is described as reaching a 3.0 stage with features like garbage collection, improved exception handling, JavaScript string built-ins, custom text format annotations, and deterministic profiling—changes that make it more practical for real-world apps.

Adoption is also framed as real, if uneven. Cloudflare Workers are cited as a path where Wom can be used via Rust, C++, and Go (with Python mentioned as less certain), because those languages can compile down to the same web execution target. Notable usage includes Figma.

The most radical final prediction is “metal”: placing a JavaScript VM in the kernel so every program runs in an isolated VM environment, with compiled code targeting the assembly format throughout. Selling that idea in 2014 was hard; selling it in 2026 is still difficult, but the transcript notes that hosting providers appear to be moving in that direction. The conclusion is less “JavaScript dies tomorrow” and more “JavaScript stops being the default authoring language”—with AI blamed as the wildcard that derailed the original timetable.