Magnetoencephalography (MEG) is generally considered as the complicated and expensive counterpart to electroencephalography. Therefore, its use is not widespread with close to 200 locations worldwide, nevertheless a string of important insights in neuroscience have been achieved with it. MEG is currently experiencing a hardware innovation cycle since the cryogenic sensor, the SQUID, is challenged by a new type of sensor, the optically pumped magnetometer, OPM. A comparison on the individual sensor level will not immediately identify the winner as, e.g., noise is at least five times lower in SQUIDs, but OPMs can be brought much closer to the brain yielding several times higher signal amplitudes. Due to the very different operating principle of these sensors the complete system design is different between MEG systems based on either of the sensor types. Here OPM-MEG is the winner in terms of flexibility and unobtrusiveness for the participant, but the noise budget of OPM-MEG is not yet that well understood, and flexibility might come at the cost of increased noise. For multimodal applications such as combined EEG-MEG or fNIRS-MEG again OPM-MEG appears to be the winner as ongoing research indicates. Interestingly, manufacturing of individual sensors and systems has rapidly attracted funding particularly in the USA, UK, France, and China. Three companies are established in the USA, one each in UK and France, and it is expected that at least four companies are operating in China. This commercialization indicates that the medical device industry sees a clear potential in OPM-MEG.