What WebGPU changed, and what it didn't
Compute shaders and lower draw-call overhead are real wins. Universal availability is not here yet, so every scene we ship still has a WebGL2 path — and will for a while.
WebGPU delivers on the things it promised: dramatically lower per-draw overhead, compute shaders that make GPU-side culling and particle work practical, and a cleaner API that removes a decade of WebGL footguns.
For our workloads the compute path matters most. Point cloud LOD selection, splat depth sorting and instanced culling all move off the CPU, which is exactly where our frame budgets were bottlenecked on mid-range devices.
What has not changed: you still need a WebGL2 fallback. Coverage on the devices that actually visit Indian storefronts is good but not universal, and the failure mode of no fallback is a blank canvas rather than a slow one.
We build both paths from one scene description and test both in CI. The cost is real but bounded, and it will stay necessary for another couple of years. Anyone telling you WebGPU-only is safe today is not looking at their own device analytics.
Written by the engineering team at 3D.MBA, Ahmedabad.
Disagree? Tell us