Firefox 141 enables WebGPU on Windows — stable cross‑browser GPU for the Web
Key update
Firefox 141 (stable) has enabled the WebGPU API on Windows, making the standardized GPU compute and graphics API available to mainstream Firefox users (support is not yet available in service workers). (firefox.com)
Why it matters
With Firefox stable joining Chrome’s long‑running implementation, WebGPU is now widely available across major desktop browsers, removing a major cross‑browser blocker for shipping web apps that rely on GPU compute or next‑gen rendering. Practically, that means real‑world apps — games, interactive visualizations, compute‑heavy tooling, and even browser‑side ML acceleration — can target WebGPU without falling back to WebGL or heavy native components for a large portion of your user base. Firefox’s implementation is based on the wgpu stack and the team notes some remaining performance and feature gaps (for example, importExternalTexture and certain IPC/timing optimizations) that are being iteratively fixed, so test on representative hardware and drivers before a wide rollout. (mozillagfx.wordpress.com, firefox.com)
Source
Read Next
Node.js v25 scheduled for 2025‑10‑15 — semver‑major release imminent
September 30, 2025Node.js v25 is scheduled for October 15, 2025 (commit cutoff 2025‑09‑15). Teams should run CI against the new major, validate native modules, and prepare canary deployments.
Azure Functions Proxies: community support ends 2025‑09‑30 — migrate off Proxies now
September 29, 2025Azure announced Azure Functions Proxies will be unsupported after 2025‑09‑30; teams still using Proxies must inventory and migrate to a supported API surface (APIM, Front Door, or a lightweight reverse proxy) immediately.
NodeShield: runtime SBOM enforcement (CBOM) for Node.js limits supply‑chain attacks with negligible overhead
September 28, 2025A new paper introduces NodeShield, a runtime enforcement system that uses SBOMs extended with per‑dependency capabilities (CBOM) to prevent supply‑chain abuses in Node.js with ~98% effectiveness and <1ms overhead.