The technology described herein is directed towards an uplink video streaming technology including for private networks that use low-cost mobile devices with reduced hardware capabilities (RedCap in 3GPP). The technology adapts the video quality (bitrate) from the content producer side based on the device capability data of the content consumer, leveraging that both are co-located in the same network. The technology can use machine learning to identify correlated devices with suboptimal video quality, and to propose new uplink data rate limiting data (e.g., PRB quota) for the content provider, to enforce an updated (e.g., reduced) video quality independent of any streaming application. The technology reduces spectrum loss, energy consumption and video rebuffering that otherwise often occur when videos are generated with high quality beyond the capabilities of the video consumers (e.g., including RedCap devices).