For the first time in history, your web browser can tap into your GPU with native-level speed. Three out of four browsers now support WebGPU. It’s being called a revolution. And it is—unless you’re running Linux.
You’ve probably seen the celebration: “WebGPU crosses 75% availability!” The tech press is giddy. Developers are already shipping 3D tools, AI inference pipelines, and games that run inside a tab. It feels like the web has finally grown up.
But something’s rotting under the hood.
The 75% number is a trap.
Here’s what nobody’s saying: that metric comes from a survey that is biased by default—web3dsurvey.com samples visitors who already care about 3D. And it hides a brutal truth: on Linux, WebGPU availability sits below 20%. Not 70%. Not even 40%. Twenty percent. If you’re building for Linux, you might as well be targeting a different planet.
I saw this firsthand at a WebGL meetup last month. A developer showed off a real-time fluid simulation that ran butter-smooth on macOS and Windows. Then someone asked, “How does it run on Fedora?” Silence. The answer: it doesn’t. The driver simply doesn’t exist for that hardware-software combo.
WebGPU isn’t upgrading the web. It’s quietly bifurcating it.
Here’s the pattern that makes me uneasy: Every leap forward in browser performance has come with a silent cost. First, it was video codecs (H.264 vs. VP9). Then it was WebGL (NVidia vs. AMD). Now it’s WebGPU. Each time, the “universal” web becomes a little less universal. The barrier isn’t intentional malice—it’s the grinding reality of hardware fragmentation. Linux users don’t get first-class GPU drivers from the major vendors. They never have. And now, the web is about to punish them for it.
You might be thinking: “But Linux is only a tiny slice of the market!”
That’s exactly the attitude that will kill the open web. Of course, it’s a minority platform. So is the open web itself. Remember the original promise? “Write once, run anywhere.” WebGPU should be the fulfilment of that dream. Instead, it’s becoming a tool that lets the majority platform pull ahead, while the minority gets stuck with a degraded experience.
The next great web app won’t work on your Linux laptop—and you’ll be told it’s your fault for choosing the wrong OS.
I watched this happen with WebGL. Then with WebVR. Now it’s happening again. We keep building bridges that only half the people can cross.
So what do we do?
First, stop quoting the 75% number like it’s gospel. It’s a self-selected audience of enthusiasts. Real-world penetration is lower. And for Linux, it’s a catastrophe.
Second, demand driver support. If you’re a developer, test on Linux. If you’re a corporation, push your GPU vendors to ship Vulkan/WebGPU drivers for open-source stacks. Don’t accept “it’s not a priority.”
Third, call out the lie. The web is not a level playing field. WebGPU is amazing for some, and invisible for others. That’s not a revolution. That’s a partition.
We need to build a web that serves everyone—even the 3% running Linux. Because if we don’t, the web will become just another walled garden, this time made of silicon and shaders.
The next time you see someone celebrate WebGPU’s “75% milestone,” ask them: what about the other 25%? And then ask yourself: are you okay building a future that leaves a quarter of the web behind?
FAQ
Q: Is the 75% WebGPU number accurate?
A: It's from a self-selected survey of web3D enthusiasts, not a random sample. Real-world availability is lower, especially on Linux where it's under 20%.
Q: What does this mean for developers?
A: If you rely on WebGPU, your app won't work reliably on Linux. You need to either polyfill, fall back to WebGL, or accept a broken experience for open-source users.
Q: Is Linux really that important?
A: To the spirit of the open web, yes. If we let hardware fragmentation dictate compatibility, the web loses its 'write once, run anywhere' promise. It becomes just another platform.