Multimodal Licensing Premiums Rise as Text Supply Saturates
Foundation model developers have largely exhausted freely available web text suitable for training, shifting demand toward multimodal academic content including figures, chemical structures, and data tables that models still generate unreliably from prompts alone. Publishers have responded by pricing multimodal licenses at roughly 3.2 times the rate of plain text agreements, a premium several major publishers have specifically highlighted in recent earnings calls as their fastest-growing licensing revenue category. This dynamic is pulling licensing negotiations toward archive completeness rather than text volume alone, rewarding publishers with strong figure and dataset coverage over those with text-only digital archives.
Market Impact: Peer-reviewed content earns 4x higher rates








