Low latency that survives the concurrency peak

Any CDN can hit two seconds with a thousand viewers. The question is what happens at a million, on mobile networks, during the only three minutes of the event that anyone remembers. We built the packaging into the edge so the answer stays the same.

Talk to a media engineer See pricing

2.1 s

Glass-to-glass, LL-HLS

0.19%

Rebuffer ratio at peak

1.4 M

Peak concurrent viewers served

31%

Egress saved, per-title encoding

The pipeline

One ingest. Every format. No storage fan-out.

Contribution arrives once. Transcoding, packaging, DRM and delivery all happen downstream of that single mezzanine — so adding a format costs you CPU, not another copy of the catalogue.

Live

Latency is easy. Latency and stability is the product.

Chasing the live edge is what breaks players. Our edge holds partial segments in memory and answers blocking playlist requests locally, so a viewer on a congested mobile network steps down a rung instead of stalling and resyncing.

LL-HLS with blocking playlist reload and preload hints, and LL-DASH with chunked-CMAF delivery, both answered at the edge

Redundant ingest with automatic failover between contribution sources

Server-guided ABR so the edge can steer a struggling client before the buffer drains

Server-side ad insertion stitched per viewer without a client-side latency penalty

Quality of experience at 1.4 M concurrent

Startup time (p75)0.72 s

Rebuffer ratio0.19%

Viewers on top rung86%

Exits before first frame0.4%

Championship final, Meridian Live, 2026. Illustrative figures for this design.

Per-title encoding

A talking-head interview and a 60 fps action sequence do not deserve the same ladder. We analyse each asset and build the rungs it actually needs — typically 22–38% fewer bits at matched VMAF.

Studio-grade protection

Multi-DRM with a single licence API, forensic session watermarking, concurrent-stream limits and geo/token gating enforced at the edge rather than in your player.

Catalogue at the edge

Thumbnails, sprites, subtitles and trick-play generated on request and cached, so a back-catalogue refresh does not mean re-uploading a petabyte.

Formats

What we package, from one mezzanine

Delivery format	Codecs	DRM	Typical use
LL-HLS (CMAF)	H.264, HEVC, AV1	FairPlay, Widevine	Live sport, auctions, esports
LL-DASH (CMAF)	H.264, HEVC, AV1	Widevine, PlayReady	Live, Android & smart TV
HLS (TS & fMP4)	H.264, HEVC	FairPlay	VOD catalogue, legacy iOS
DASH	H.264, HEVC, AV1, VP9	Widevine, PlayReady	VOD catalogue, web
Progressive MP4	H.264, AV1	Signed URL	Social clips, downloads, previews

Point a test stream at us this afternoon

Free ingest endpoint, real PoPs, real numbers. Compare startup time and rebuffer ratio against your current delivery before you commit to anything.

Get an ingest endpoint See the network