Media over QUIC · QUIC Development
We build live streaming on the transport that's replacing the old stack — Media over QUIC, raw QUIC, and WebTransport — for sub-second latency that holds at scale. Built on quic-go, Cloudflare quiche, and WHIP/WHEP ingest, deployed in your cloud. First working build in 1–2 weeks, from $8K.
Who we build for
The transport decision
Media over QUIC (MoQ) is an IETF transport, in active draft as draft-ietf-moq-transport, that carries live media over QUIC. It aims to combine the sub-second latency of WebRTC with the CDN-scale fan-out of HLS — the two things you previously had to choose between. Here's how the four transports that matter actually compare for a production build.
The pipeline
A MoQ system replaces the brittle parts of the old low-latency stack — the TCP segment fetches of HLS, the bespoke SFU mesh of WebRTC at scale — with one QUIC-based path from contribution to player. Here's the route a frame takes, and where the latency budget goes.
Figure 1: Media over QUIC delivery path — ingest to player, with per-hop latency budget.
The first mile arrives over WHIP (WebRTC ingest), SRT, or RTMP. We normalize it and hand it to the QUIC layer.
~50–150 msQUIC (RFC 9000) carries media as independent streams, so one lost packet never stalls the rest — no head-of-line blocking, and connection migration survives a network switch mid-stream.
~10–40 ms/hopA Media over QUIC relay (moq-rs, moxygen, or a Cloudflare relay) forwards objects to subscribers and downstream relays. Fan-out happens at the relay tier, the way a CDN scales HLS — but at sub-second latency.
~10–30 ms/hopRelays sit at the edge over the QUIC/HTTP-3 path, so delivery rides existing CDN economics instead of a bespoke real-time mesh.
~5–20 msThe client subscribes over WebTransport and renders through MSE or a custom decode path, with a jitter buffer tuned to your latency-vs-smoothness target.
~50–100 msA tuned MoQ path delivers glass-to-glass latency under one second to a CDN-scale audience — the combination WebRTC and HLS each gave you only half of. We've shipped 0.4–0.5s at 10,000 concurrent viewers on a custom WebRTC + Kurento build (Worldcast Live); MoQ is how we now take that latency to a far larger audience without the mesh. For the protocol-level detail, see how QUIC works.
Why now
For a decade, live streaming forced a trade-off: WebRTC for sub-second latency but painful past ten thousand viewers, or HLS/DASH for CDN-scale reach but two-to-thirty-second delay. Media over QUIC collapses that choice. Three things made 2026 the year to build on it.
QUIC is RFC 9000 and HTTP/3 is RFC 9114; the IETF MoQ Transport draft (draft-ietf-moq-transport, now at revision -17, May 2026) is far enough along that production implementations track it closely.
Cloudflare launched a production MoQ relay network in August 2025, running on every server across 330+ cities (open-source moq-rs), with Meta, Google, and Cisco building interoperable implementations. Browser WebTransport support is broad enough to reach real audiences.
quic-go, Cloudflare quiche, moq-rs, and moxygen are production-grade — you no longer write a QUIC stack from scratch to ship a MoQ product.
Being early is the advantage. The teams that build correct MoQ products in 2026 own the latency-sensitive use cases — live commerce, in-play betting, real-time auctions, interactive sports — before the field crowds in. We've spent twenty years on the hard half of this problem: the transport, the jitter buffers, the congestion control, the fan-out. The protocol is new; the engineering underneath it is what we've always done.
What we build
Sub-second video so a viewer taps “buy” while the product is still on screen. We built Sprii's RTMP + WebRTC multistreaming on Cloudflare and Mux — 12.3 million products sold through live shopping in a single year.
In-play interactive overlays only work if every viewer sees the same moment at the same instant. MoQ keeps the whole audience inside a one-second window.
Real-time desktop and chart streaming where a half-second of lag is a missed trade. We built Tradecaster — live trade streaming for 46,000+ users, auto-scaling through market-hour spikes.
Low-latency live channels inside a large VOD platform. We build for OTT scale: Mangomolo serves 1B+ streams a month to 30M+ daily viewers at up to 4K.
Concerts and shows where remote performers play together and audiences talk back. Worldcast Live runs full-duplex HD at 0.4–0.5s latency for 10,000 concurrent viewers.
Input-to-photon budgets where QUIC's stream independence and connection migration keep a session alive across a network change.
When custom wins
A low-latency SaaS (Millicast, Ant Media, Red5) is the right call when its feature set fits and you're happy renting the transport. Custom wins when latency-at-scale is the product itself, when you need to own the relay tier and the roadmap, or when you're early enough that being first with the right stack is the moat. It wins at any audience size — a thousand viewers or a million.
Figure 2: Build vs Buy — sub-second-at-scale requirement × control and future-proofing. Custom wins the top-right at any audience size.
How we work
A latency target, an audience size, no system yet. We pick the transport mix, build the relay and player path, and ship a working low-latency stream.
UpgradesYou're on HLS or an SFU and the delay or the cost is hurting. We move the latency-critical path to MoQ/QUIC and tune the budget hop by hop.
TakeoversYou inherited a half-built real-time stack. We stabilize it, document it, and extend it — the way we took over and rebuilt Rafiky's real-time pipeline.
Pricing
Fixed-scope starting points. Final scope depends on ingest mix, audience scale, relay topology, and player targets — run the calculator for an instant estimate.
Free for qualified projects
Before any contract, we'll give you something useful. Pick the one that fits where you are.
Competitor analysis, core feature definition, monetization modeling, and a full launch blueprint — delivered within a week. Written by engineers who'll build what they plan.
An independent review of your system's technology choices, structural components, and workload fit — with a plain verdict on what's working, what's a liability, and exactly what to change to reach your goal. Delivered within a week.
A full audit of your code with every issue documented, evidenced, and located — exact file, exact line. Plus a system architecture review and a prioritized fix roadmap. Not a consultant's opinion. A case file. Delivered within a week.
A specialist review of your video or streaming product covering latency, media server architecture, WebRTC, playback reliability, real-time chat, and scalability. Every finding is specific, located, and fixable. Delivered within a week.
Why Fora Soft
Worldcast Live runs 0.4–0.5s glass-to-glass for 10,000 concurrent viewers on a custom build. The hard part of MoQ is the part we've done for years.
quic-go, Cloudflare quiche, WHIP/WHEP, SRT, WebRTC, mediasoup, CMAF, LL-HLS — shipped in real products, not slide decks. Sprii, Mangomolo, Tradecaster, Worldcast Live.
We track the IETF MoQ draft, Cloudflare's production relay network, and the Meta/Google/Cisco implementations so your build is correct against where the standard is going, not where it was.
Senior engineers, no offshore handoffs, 250+ products since 2005, and a 100% job-success score on Upwork. We finish and hand over clean.
FAQ
What is Media over QUIC, in one line?
How is MoQ different from WebRTC?
How is MoQ different from HLS/DASH?
What latency can you actually hit?
Is MoQ production-ready in 2026?
Will it work in browsers and on CDNs?
Can you migrate our existing HLS or WebRTC stack?
How much does a build cost?
How long does it take?
Do we need QUIC and MoQ, or just one?
Keep reading
What Media over QUIC is
Read article →Knowledge BaseHow QUIC works
Read article →ToolEstimate your build
Get instant quote →Related ServiceScalable video streaming
See related service →Related ServiceWowza streaming development
See related service →Related ServiceWebRTC development
See related service →Within 48 hours you'll get a realistic estimate, a technical recommendation, and an outline of next steps. No obligation. NDA before any access to your code, recordings, or operator dashboards.
Specialist software house for video, real-time and AI products. Founded 2005. 50 in-house engineers.
