Using a panel of more than 3,100 French corporate groups’ affiliates and parent companies, we estimate a production function model where we enable the productivity of a firm to depend on the knowledge produced by the R&D activities of the other companies in the group. We find indeed that a firm’s productivity may significantly be enhanced thanks to the R&D capital of the other affiliates. This enhancement can be estimated to be, for the corporate group as a whole, between 30% and 40% of the “usual” estimate of the direct impact of firms’ R&D expenses on their own productivity. However, this effect differs depending on whether the firm itself conducts some R&D or not. In case it does, the other affiliates’ R&D does not appear to impact significantly on its own performances: those depend only on its proper R&D activity. At the opposite, the other affiliates’ R&D has a very significant effect on the productivity of firms which do not conduct any R&D. These results emphasize the existence of group spillovers, which differ from the usual industry or geographical spillovers. In particular, they do not seem to require an “R&D based absorptive capacity” to pre-exist and they are clearly the result of an explicit strategy, defined at the group level. Finally, these results might lead to revise upwards our estimates of the private returns on R&D investments.