MoQ Monthly, Issue #2

May 2026

The London interim is coming up soon! We'll be hosting the IETF MoQ Working Group June 9-12 at the Cloudflare London office, and I'm very much looking forward to seeing everyone in person again. There's yet again been a lot of activity since the last issue of MoQ Monthly. Let's get into some of it.

London Interim and draft-18

This is the last planned WG hybrid interim before IETF 126 in Vienna in July, and there's a concerted push among some key members of the working group to see if we can reach Working Group Last Call on the base MoQT draft after that plenary. Whether that aggressive timing is fully realistic remains to be seen, but the intent is there and it's shaping agendas and prioritization. A lot of heavy lifting on the spec will happen in the next few weeks and especially at the London interim.

Setting the stage for London: draft-ietf-moq-transport-18 was published May 12. Sergio Garcia Murillo (Dolby.io) has a good implementer-focused summary on LinkedIn. Key changes include a unified moqt:// URI scheme covering both native QUIC and WebTransport; separation of SUBSCRIBE_NAMESPACE and SUBSCRIBE_TRACKS into distinct messages; a new .session reserved namespace for Session-Level Tracks, which lets extensions reuse the subscription and delivery machinery without requiring new control message types; REDIRECT support via REQUEST_ERROR; updated GOAWAY semantics for session migration; PUBLISH_OK merged into REQUEST_OK; and various wire format updates including delta encoding in FETCH responses. The interop target for the London interim is this newly released draft-18 and I've updated the automated interop test runner accordingly.

In line with the (aggressive but possible) push towards WGLC, the WG milestone on Datatracker now shows December 2026 as the target for the IESG publication request of the base MoQT spec. That doesn't mean the spec will necessarily be done by then, but it does mean that the WG is signaling that it's aiming to be done by then and the next few months are going to be a critical period for getting there.

MoQ Working Group chair Martin Duke posted a preliminary agenda for London to the list last week. Day 1 targets Object Range Filters, Subscription Location Filters, Request Blocking, Concurrent Subscribe, Joining FETCH, and SWITCH. Day 2 brings the Interop Report, DoS Design Team readout, Top-N, MSF/CMSF, Privacy Pass, Secure Objects, and LOC. As with previous hybrid interim events, remote participation will also be available - I'll send connection details to the mailing list closer to the date.

Dan Rayburn Town Hall

On May 12 - the same day draft-18 was published - Dan Rayburn hosted a MoQ industry town hall. I wasn't able to attend live due to other meetings, but I listened to the recording and there's some good discussion in there.

Dan Rayburn covers the commercial streaming industry mostly from an analyst's perspective, not often venturing into the technical details of IETF protocol work - his decision to convene this event, and the fact that this many people showed up for it, says something about where MoQ sits in the industry conversation right now.

Streaming Tech Sweden

Streaming Tech Sweden ran May 21 in Stockholm. I wasn't there this year - I spoke about MoQ at STS last year and can say firsthand that it's a great event. The session lineup this year looks like it was excellent again and had a strong MoQ showing.

Per Mafrost (Vindral CTO) presented their end-to-end MoQ implementation experience and first production numbers. Implementing a specification that's still being written is non-trivial, and this presentation was pitched as an experience report on laboring under the complex constraints that entails.

Christoffer Ainek and Andreas Lejonlycka (SVT) presented on software-based production workflows for the Milano Cortina Winter Olympics, which, as I understand, included some use of MoQ as part of SVT's NEO initiative.

Will Law (Akamai) also presented on MSF and CMSF, covering media and event timelines and use cases. MSF/CMSF is also on the London agenda and represents kind of the "typical" streaming format for broadcast style live streaming use cases and similar scenarios.

Recordings are up on plus.streamingtech.se already, but it looks like a code is needed to access them right now. I'm personally looking forward to watching the presentations as soon as I'm able.

MoQ Boy

Luke Curley published MoQ Boy on April 16. We're covering it a little late because we had too much to cover last time and ran out of time to give this a proper write-up. Essentially, MoQ Boy is a "Twitch Plays Pokemon"-style demo running homebrew Game Boy games on emulators, streamable and controllable by anyone over MoQ. It's fun to play, and it's a great demo of MoQ's low latency capabilities, but the architecture is the real story and it's not hard to see how the technology involved could be adapted for much more serious use cases as well.

The on-demand resource management story is particularly clean. Each encoder - audio, video - only runs if there's an active subscription for that track. No subscribers? The emulator sleeps. That behavior is built into MoQ: publishers only transmit when there's a subscriber. Luke walks through the implications: if nobody wants captions, don't run Whisper; if nobody wants H.265, don't transcode it. The savings compound as workloads get more expensive. The controls implementation is also a nice illustration of bidirectional communication via two unidirectional tracks and live namespace discovery, with each player publishing their own controls track and the emulator using namespace discovery to find and subscribe to all of them. No separate signaling layer needed.

Software Mansion are working on React Native bindings for the moq.dev Rust library - there's a screenshot in the post of MoQ Boy running on a real phone.

One thing worth noting: Luke is available for MoQ pilot consulting (see the post). I also know of a couple of other people in the community who'd be glad to help as well - if you're trying to get a MoQ project off the ground and could use a hand, either reach out to Luke or to me and we'll try to connect you with the right person for the job.

CDN Traction

Two announcements (also from mid-April) are worth reading together:

Fastly published "Media over QUIC: Can Streaming Finally Have Both Scale and Low Latency?" on April 15, signaling intent to play an active role in the ecosystem as it evolves.

CacheFly and Red5 announced a MoQ beta on April 14, with commercial rollout slated for summer 2026. The stack supports WHIP, SRT, RTMP, and Zixi ingest. Chris Allen (Red5) called it "one of the first end-to-end production-ready MOQ streaming solutions that operates at global scale."

Cloudflare launched a MoQ relay service some time ago. Akamai has Will Law running MSF/CMSF work in the WG. Now Fastly is publicly signaling interest and Red5 and CacheFly are building commercial products on it. The CDN layer is forming up.

Jan Ozer on MoQ

Jumping ahead to this month again, Jan Ozer published a really solid overview and explainer at Streaming Learning Center on May 26 - how MoQ works, where it fits relative to HLS/DASH and WebRTC, production readiness as of now. It's well-researched, has a ton of links to additional resources, and would be a good thing to send a colleague who's asking what MoQ is and why they should pay attention.

Layering Drafts

MoQT doesn't particularly care what exactly you're transporting with it. Two recent drafts illustrate the range:

Paul Gregoire (Red5) and Gwendal Simon (Synamedia) published draft-gregoire-moq-msfts, which registers an "m2ts" packaging value for MSF to carry MPEG-2 Transport Stream content over MoQ without repackaging. It covers SCTE-35, PCR continuity, random access, and multi-program TS. The draft hit Hacker News; some of the discussion there missed the point. The value isn't "MPEG-TS is new" - it's backward compatibility with the enormous installed broadcast infrastructure that runs on it today. The draft is on the agenda to discuss in London.

Erik Herz also published draft-herz-moq-nmsf -01 back on April 7, a Neural Media Streaming Format for carrying neural video codec representations over MoQ. The two-track model separates hyperprior and latent representations as distinct MoQ tracks with explicit dependency and priority signaling, with per-frame timestamps and per-frame quality parameters for E2E latency measurement and rate-distortion signaling. Neural video codecs are still largely research territory, but MoQT can transport the relevant data efficiently and Erik's spec demonstrates the flexibility of the layered MoQT + streaming format approach.

MMSys '26

Three MoQ papers at ACM Multimedia Systems Conference 2026 that happened in Hong Kong in April, all from Zafer Gurel, Deniz Ugur, and other colleagues in Ali Begen's group at Özyeğin University:

"MOQtail: Open-Source, IETF-Compliant MOQT Protocol Libraries" describes their draft-16-compliant Rust and TypeScript implementation, and reportedly won the Best Open Dataset and Software Award. The project is at moqtail.dev.

"Time Travel in MOQ Conferencing" demonstrates sub-100ms end-to-end videoconferencing with relay-cached rewind, so participants can replay a missed segment from another person's stream without interrupting the live session.

"Scaling The Cheer: Co-Viewing with Dual-Mode MOQ Transport" covers a watch-party architecture with a dual-mode design for high-fidelity main content combined with synchronized real-time group interaction.

Özyeğin continues to be one of the most prolific research groups in the ecosystem.

Coming Up

Besides the IETF MoQ Working Group itself meeting in London soon, there are a few other events coming up in the next few months with MoQ content:

FOKUS Media Web Symposium - June 16-17, Berlin. The week after London. nanocosmos is presenting "one year of MoQ in production" - architecture, failure modes, real operational numbers. They introduced MoQ at MWS 2025; this is a follow-up which should have some actual production data.

CommCon 2026 - June 9-11, now in Düsseldorf. Running concurrently with the start of the London interim. MoQ is on the topics list alongside WebRTC, SRT, and voice AI. An unfortunate scheduling conflict for anyone who has to choose. At least three talks deal explicitly with MoQ: Dear WebRTC Devs: It's Time to MOQ On by Ali C. Begen, Looking at MoQ with WebRTC eyes by Lorenzo Miniero, and How to live-stream your conference with MOQ by Marco Vidonis.

RTC.ON 2026 - September 16-18, Kraków, hosted by Software Mansion. Luke Curley is running a hands-on MoQ workshop: build a MoQ-powered audio/video room call from scratch, with an advanced track extending it with real-time speech-to-speech translation. Connects naturally to Software Mansion's React Native MoQ binding work mentioned above.

Resources

• IETF MoQ WG: datatracker.ietf.org/group/moq
• Mailing list: mailarchive.ietf.org/arch/browse/moq
• MoQ.dev community Discord: discord.gg/FCYF3p99mr

If you're working on something MoQ-related - a draft, an implementation, a deployment, a talk, a blog post - I'd love to hear about it and help share your news with the broader community.

Subscribe: buttondown.com/moqmonthly

See you in June!

-Mike