Context. High-energy resolution spectroscopy of the 1.8 MeV radioactive decay line of 26 Al with the SPI instrument onboard the INTEGRAL satellite has recently revealed that diffuse 26 Al has higher velocities than other components of the interstellar medium in the Milky Way. 26 Al shows Galactic rotation in the same sense as the stars and other gas tracers, but reaches excess velocities of up to 300 km s −1 . Aims. We investigate whether this result can be understood in the context of superbubbles, taking into account the statistics of young star clusters and H I supershells as well as the association of young star clusters with spiral arms. Methods. We derived energy output and 26 Al mass of star clusters as a function of the cluster mass by population synthesis from stellar evolutionary tracks of massive stars. Using the limiting cases of weakly and strongly dissipative superbubble expansion, we linked this to the size distribution of H I supershells and assessed the properties of possible 26 Al-carrying superbubbles. Results. 26 Al is produced by star clusters of all masses above ≈200 M , is roughly equally contributed over a logarithmic star cluster mass scale and strongly linked to the injection of feedback energy. The observed superbubble size distribution cannot be related to the star cluster mass function in a straightforward manner. To avoid the added volume of all superbubbles exceeding the volume of the Milky Way, individual superbubbles have to merge frequently. If any two superbubbles merge, or if 26 Al is injected off-centre into a larger HI supershell, we expect the hot 26 Al-carrying gas to obtain velocities of the order of the typical sound speed in superbubbles, ≈300 km s −1 before decay. For star formation coordinated by the spiral arm pattern which, inside co-rotation, is overtaken by the faster moving stars and gas, outflows from spiral arm star clusters would preferentially flow into the cavities that are inflated by previous star formation associated with this arm. These cavities would preferentially be located towards the leading edge of a given arm. Conclusions. This scenario might explain the 26 Al kinematics. The massive-star ejecta are expected to survive ≥10 6 yr before being recycled into next-generation stars.