Copy pasting some AI summary from Google since I’d never heard of it:

Key Improvements of AV2 over AV1

Higher Compression Efficiency: Early tests indicate a 30%–40% improvement in coding efficiency, enabling lower bandwidth usage for high-quality streaming (VMAF/PSNR improvements).
Larger Block Processing: AV2 supports superblocks up to
pixels, up from AV1’s
, allowing better handling of high-resolution video.
Enhanced Partitioning: Features fully recursive partitioning with improved flexibility (semi-decoupled luma/chroma).
Advanced Prediction Techniques: Includes Temporal Interpolation Prediction (TIP), data-driven intra modes, and improved chroma-from-luma modeling.
Newer Coding Features: Enhanced support for HDR10+, 360-degree video, and split-screen viewing.
Improved Screen Content Coding (SCC): Offers improved handling of screen content, crucial for application sharing and gaming. 

Technical Advancements

Transform Kernel Redesign: The codec uses new 8-bit multiplier-based transform kernels for better coding efficiency.
Data-Driven Transforms (DDT): Implements DDT for inter-predicted blocks to adapt to varying image textures. 

Development and Adoption

Timeline: The final specifications are expected late 2025, with hardware implementations anticipated for 2026.
Status: While 53% of AOM members planned to adopt within 12 months, the codec remained in draft status as of early 2026.
Licensing: Like its predecessor, AV2 is designed to be royalty-free. 

AV2 is designed to compete with VVC (H.266) while providing a significant boost to streaming efficiency, which will help reduce CDN costs and improve buffering in high-definition video applications.