SVC encodes a video so that decoders can pick how much to consume. The base layer is independently decodable at low quality; each enhancement layer raises spatial resolution, temporal frame rate, or bitrate fidelity. An SFU can drop enhancement layers per-receiver — sending base + temporal-1 + spatial-1 to one receiver, base + temporal-1 + spatial-1 + spatial-2 to another — and each receiver decodes only the layers it gets. The bitstream is roughly 10–20 % bigger than the equivalent unenhanced encode, but much smaller than encoding 3 simulcast variants separately.
H.264 had an SVC extension (Annex G, 2007) but it was rarely supported in hardware and never widely deployed. VP9 has solid SVC support and was Google's preferred codec for WebRTC SVC through 2019–2024. AV1 has native SVC primitives that are increasingly used in 2025–2026 deployments — particularly in Google Meet, Discord and Cloudflare Calls.
SVC's appeal vs simulcast is total upload bandwidth: a 3-layer SVC stream is roughly 50 % the bytes of 3-layer simulcast at the same per-layer quality. The downside is encoder complexity — SVC encoders are more sophisticated, browser support is asymmetric (some browsers don't support AV1 SVC encode but do support decode), and falling back to simulcast remains the safe default for cross-browser deployments. By 2026 SVC is winning in pure-browser WebRTC deployments and simulcast persists in mixed-codec environments.
