LHC@home was launched as a BOINC project in 2004 as an outreach project for CERN's 50 years anniversary. Initially focused on the accelerator physics simulation code SixTrack, the project was expanded in 2011 to run other physics simulation codes on Linux thanks to virtualisation. Later on the experiment and theory applications running on the LHC@home platform have evolved to use containers and take advantage of the CVMFS file system as well as content delivery networks. Furthermore, a substantial part of the contributed computing capacity nowadays is provided as opportunistic back-fill from data centers with spare capacity, in addition to enthusiastic volunteers. The paper will address the challenges with this distributed computing model, new applications to exploit GPUs and the future outlook for volunteer computing.