Understanding the carbon dioxide emission rates under different agricultural practices is a critical step in determining the role of agriculture in greenhouse gas emissions. One of the challenges in advocating for an intercropping system as a sustainable practice in the face of climate change is the lack of information on how much CO2 is emitted by the system. A factorial randomized complete block design study was set up at two distinct agroecological locations (Syferkuil and Ofcolaco) in the Limpopo Province of South Africa to investigate carbon dynamics in sorghum-cowpea intercropping and sole cropping system over two seasons. Intercropping system emitted less CO2 compared to sole cropping system. In 2018/19 at Syferkuil and 2020/21 at Ofcolaco, intercropping systems emitted 11% and 19% less CO2, respectively, than sole cropping systems. In both agroecological regions, low cowpea density consistently resulted in higher CO2 emissions than high density. During the 2018/19 cropping season, sorghum emitted more CO2 of 5.87 t·ha−1 than cowpea with 5.14 t·ha−1 in a sole cropping system at Syferkuil. Cowpea, on the other hand, emitted more CO2 of 6.5 t·ha−1 and 10.18 t·ha−1 than sorghum during the 2020/21 cropping season at Syferkuil and Ofcolaco, respectively. Furthermore, intercropping improved the carbon emission efficiency (CEE) of the individual crops in the system. The treatments used in the intercropping and sole cropping systems had a significant impact on the strength of the relationship between carbon stocks and CEE. Our results revealed that sorghum-cowpea intercropping system at a relatively higher cowpea density in a no-till system reduces the amount of CO2 lost to the atmosphere. The system can thus, be promoted as one of the sustainable farming practices to reduce emissions and improve carbon storage in the soil.