The Alliance for Open Media just released the final specification for AV2. This open video codec promises to cut bandwidth demands by roughly 30 percent compared with its predecessor while adding support for emerging formats. Yet real-world hardware acceleration remains years away. And the complexity jump could slow initial rollout.
AV1 arrived in 2018. It delivered meaningful compression gains over VP9 and H.264. Major services adopted it gradually. YouTube started experiments that same year. Netflix followed in 2020. Hardware support lagged. The first smartphone chip with AV1 decode appeared in 2019. Widespread mobile implementation took several more years.
From AV1 Gains to AV2 Ambitions
AV2 builds directly on that foundation. The official AV2 specification site lists superior compression, better AR and VR handling, split-screen and multi-view delivery, improved screen content coding, and operation across a wider quality range. These additions target not only traditional streaming but also immersive experiences and conferencing.
Early tests delivered consistent results. Netflix research scientist Andrey Norkin, who co-chairs AOMedia’s coding group, shared data last year showing 28.63 percent bitrate reduction on PSNR-YUV and 32.59 percent on VMAF versus AV1. An arXiv paper from May 2026 confirmed similar figures. AV2 version 13.0 achieved 29.81 percent BD-rate savings on PSNR-YUV and 33.79 percent on VMAF under random access conditions. The numbers hold across HDR and extended color testing.
But gains come with costs. VideoLAN’s Jean-Baptiste Kempf, lead on the VLC project, noted that AV2 decoding runs roughly five times more complex than AV1. “In practice, that means software running on today’s hardware will struggle to decode AV2 in real time without careful, architecture-specific optimization,” he wrote on his blog. The team responded by open-sourcing dav2d, the successor to the successful dav1d AV1 decoder. It already powers playback in VLC, parts of Windows, and Android libraries.
The final v1.0.0 specification landed on May 28, 2026, according to the AV2 site. It includes the complete bitstream syntax, semantics, decoding process description, and matching AVM reference software. An earlier working draft from January 2026 was superseded. Android Authority reported the milestone hours after it broke, highlighting potential mobile data savings of up to 32 percent.
Adoption surveys from AOMedia members painted an optimistic picture. Fifty-three percent said they planned to deploy AV2 within 12 months of release. Another 35 percent expected to follow in the second year. The Streaming Media article from September 2025 captured those figures and noted parallels to AV1’s path. AV1 moved from announcement to scaled deployment at Netflix, Google, and Microsoft despite similar early complexity concerns.
AV2 stays royalty-free under the same patent policy that defined AV1. Licensees face no fees to AOMedia. That stance contrasts with VVC, also known as H.266, which carries licensing costs and has seen slower uptake. Early benchmarks suggest AV2 matches or exceeds VVC in efficiency for many use cases. Yet hardware matters. Without dedicated silicon, software decoders consume significant CPU resources. Current encoders run slowly. One Hacker News commenter observed that production AV2 encoding often clocks around one frame per second on high-end hardware. Practical real-time performance likely waits for 2028 or later when ASICs appear.
Streaming providers already eye the bandwidth relief. A 30 percent reduction compounds across millions of concurrent viewers. Lower bitrates mean cheaper delivery costs and higher quality on constrained networks. Mobile users stand to benefit most. Data caps remain a reality in many markets. Better compression extends battery life too since less data travels over wireless radios.
AR and VR content presents another driver. Split-screen and multi-view tools let platforms deliver multiple camera angles or layered overlays without doubling bandwidth. Screen content coding improvements sharpen text and graphics in remote desktop or productivity streams. Film grain synthesis carries over from AV1 with refinements. The codec also handles a broader dynamic range, useful for both premium cinema and user-generated clips.
Yet questions linger. The Phoronix report in May 2026 flagged the delayed timeline. AOMedia had targeted end-of-2025. The final release slipped several months. No major smartphone chips advertise AV2 decode yet. Android 17 reportedly adds H.266 support but skipped AV2 because the spec arrived too late for integration. Google previously made software AV1 decode mandatory in Android 14 using efficient libraries. A similar path seems probable for AV2 once dav2d matures.
Encoding hardware presents a steeper climb. Pixel 10 phones became the first to offer AV1 video capture. Most manufacturers still rely on H.264 or HEVC for recording. Samsung and vivo turned to a specialized APV codec for professional capture instead. AV2 could change that equation if efficiency translates to smaller file sizes on device. Consumers might finally store more 4K or 8K footage without sacrificing quality.
Industry reaction split between excitement and caution. AOMedia positioned AV2 as a foundational element for the next decade of video. Member companies include Amazon, Apple, Google, Meta, Microsoft, Netflix, and Samsung. Their collective weight could accelerate software support. But silicon vendors move deliberately. They require stable specifications and clear market demand before taping out new chips.
Developers already experiment. The AVM reference code sits publicly on GitLab. VideoLAN’s dav2d work signals quick progress on playback. Browser makers and operating system teams will likely follow once encoders reach acceptable speed. Full ecosystem readiness could stretch into 2029 for high-volume consumer devices.
Still, the direction looks clear. Video traffic continues its relentless climb. Efficiency gains matter more than ever. AV2 delivers measurable improvements without royalties. Its feature set anticipates future content types. The next few years will test whether the industry can repeat AV1’s adoption success at greater scale and complexity.
One thing feels certain. Services that integrate AV2 early will enjoy cost advantages and quality edges. Those that wait risk higher delivery bills and compressed streams that fail to impress on growing screen resolutions. The codec has arrived. Implementation starts now.
WebProNews is an iEntry Publication