Carbon nanotubes (CNTs) and graphene are currently attracting the largest investments by firms focussing on the development of nanotechnologies. However, firms are heterogeneous in their ability to develop nanotechnologies that present different levels of radicalness and a different potential to become basic building blocks of future developments. We address this issue by analysing the patent portfolios of the largest players in the field. For each patent, we explore how the probability to generate a radical knowledge recombination is affected by the diversity of knowledge bases on which such a patent has been built. We thus explore to what extent science and technology have contributed to technology development, and to what extent the diversity among firms’ patent portfolios is explained by a different ability to combine together different scientific and technological sources. Results show that firms are heterogeneous in their ability to recombine knowledge. Albeit there are no companies neglecting neither science nor technology, some of the top inventors tend to cite more science than technology, while others tend to prefer the alternative innovation pattern of technology-driven recombination. From these results we derive implications for both managers and policy makers.