ITU and ISO begin standardizing AI-driven video coding beyond VVC

This move signals the official transition toward AI-native video compression, marking the end of the VVC-only development era. For the streaming ecosystem, this indicates that the next leap in efficiency will likely come from hybrid neural architectures rather than incremental signal processing gains. While VVC is only now seeing hardware integration, the 2025 roadmap establishes a clear timeline for the hardware vendors like Qualcomm and MediaTek to begin preparing for AI-assisted codecs. Industry participants should monitor the 2025 Call for Evidence as it will define the computational complexity bounds that will govern the next decade of playback hardware.

The push for a post-VVC standard, often referred to as H.267, coincides with the first waves of commercial VVC adoption. Per PRLog (February 2025), VVC made significant strides in 2024 through Intel's integration of hardware decoding in its Lunar Lake GPUs and the addition of VVC support to the FFmpeg multimedia framework. However, widespread deployment remains hampered by the lack of dedicated silicon in mobile devices. Industry analysts from Streaming Media (March 2025) note that while HEVC remains the dominant standard for 4K/HDR, newer codecs like AV1 and VVC are facing slower adoption curves due to the absence of a singular 'killer app' beyond high-end 8K streaming.

Technically, the Joint Video Experts Team (JVET) is pursuing two primary tracks for the next generation. The Enhanced Compression Model (ECM) represents an evolutionary path refining traditional block-based architectures, while Neural Network-based Video Coding (NNVC) explores end-to-end deep learning frameworks. According to technical reports from Ofinno (October 2025), recent JVET evaluations show that candidate tools for the new standard are already outperforming VVC reference software by 20% to 50% in bitrate efficiency. This performance boost comes at a cost; early versions of the ECM have demonstrated encoding and decoding complexity increases roughly 10 times higher than VVC, necessitating a strong focus on implementation cost-effectiveness.

Global tech leaders are already positioning themselves as 'ecosystem architects' for this new era. Per GreyB (2025), companies like ByteDance, Huawei, and Alibaba are among the top contributors to the JVET proceedings, with ByteDance alone holding hundreds of patent families related to the upcoming standard. This participation suggests a shift toward codecs that are inherently optimized for cloud-native and mobile-first platforms. The standardization process is projected to reach completion by late 2028 or 2029, with January 2027 serving as a critical milestone for assessing the collaborative phase of the new standard.