There’s a version of the future where the operating system is largely irrelevant to most users, and we’re already living an early draft of it. The browser didn’t conquer the desktop dramatically - it crept in through productivity apps, then communication tools, then developer environments, and now it hosts things that would have required dedicated native software a decade ago: video editors, IDEs, design tools, even 3D modelers running on WebGPU.
Adobe brought Photoshop to the browser in 2021. Figma built an entire professional design workflow inside Chrome without ever asking users to install anything. VS Code runs in the browser through GitHub Codespaces. These aren’t stripped-down web versions of real apps - they are, in most practical respects, the real apps.
The OS Is Now Mostly Infrastructure
What the operating system actually provides today, for a typical knowledge worker, is a bootloader for Chrome. File management, window arrangement, notification routing - yes, those still matter. But the applications themselves, the actual work being done, increasingly lives in a sandboxed runtime that could theoretically run on any OS that ships a standards-compliant browser. That’s why ChromeOS was ever a viable product category. It wasn’t a stripped-down computer; it was an honest acknowledgment of where computing had already gone.
Microsoft understood this threat earlier than they’d probably admit, which explains why Windows 11 integrates so aggressively with Edge. If your OS becomes irrelevant to your application layer, you have two options: own the browser, or lose the user entirely.
The Part Nobody Has Figured Out Yet
Offline capability is the one area where browser-based apps still visibly stumble. Progressive Web Apps have improved the situation - apps can cache assets, sync when reconnected, maintain local state - but the experience of losing connectivity mid-session in a browser-based tool still feels different from losing it in a native app. It’s not catastrophic anymore, but it’s not invisible either.
What’s interesting is that this gap keeps narrowing in ways that don’t generate much coverage. The Storage Access API, the File System Access API, better IndexedDB tooling - none of these are exciting headlines, but they’re methodically eliminating the last reasons to prefer native binaries for general productivity work.
At some point, the argument for native apps will come down to hardware access and performance-critical workloads. Gaming, audio production, machine learning inference at the edge. Everything else is already, quietly, the browser’s territory.