Media over QUIC, abbreviated MoQ, is the IETF's attempt to design a single protocol that serves both the cacheable-CDN use case (where HLS and DASH rule) and the sub-second live use case (where WebRTC rules). The core idea is a publish-subscribe model: publishers send media into "tracks" identified by URI, subscribers fetch tracks by name, and intermediate relays cache and fan out the data exactly as a CDN does for HLS but with QUIC streams instead of HTTP responses.
The architecture is documented in draft-ietf-moq-transport (latest draft as of 2025) plus draft-ietf-moq-streaming-format and draft-ietf-moq-warp for media-specific framings. The substrate is QUIC plus WebTransport for browser access. The killer property is per-track loss recovery: a single relay can serve thousands of subscribers with each subscriber getting QUIC's per-stream behaviour, eliminating the head-of-line and per-subscriber state problems that limit WebRTC SFU scale.
By 2026 MoQ is at proof-of-concept stage in several major operators: Cloudflare, Meta, Twitch and Akamai have all shipped MoQ experiments. The protocol is not yet ready for general adoption — drafts are still moving, browser support is partial, and tooling is sparse. The expectation is that 2027–2028 will see the first standardised RFCs and the first wave of production deployments in interactive-live applications that today choose between LL-HLS (too slow) and WebRTC (too operationally heavy).
