Brain-computer interfaces (BCIs) enable users to interact with the environment using their brain activity alone, i.e., without any movement. Current technology developments suggest a large range of possible applications for BCIs, both in the clinical (e.g., control of smart wheelchairs, stroke rehabilitation) and in the non-clinical (e.g., video-games, home control) domains. Despite this potential, and the increasing number of BCI start-ups, BCIs remain barely used outside laboratories due, at least in part, to their lack of reliability and usability. The latter is in turn due to i) a low signal/noise ratio of the recorded brain activities (the sensors being of limited quality), ii) signal processing algorithms that cannot always extract relevant and reliable information from this brain activity due to high within- and between-subject variability in the signals, and iii) user training procedures that are often long and tedious. A lot of resources have been, and are still devoted to overcoming these challenges, which has enabled substantial progress the last years. Nonetheless, both human and machine learning performances remain modest. Their improvement requires data bases far larger than the ones that are currently available in order to understand and model within- and between-subject variability, and then to adapt artificial intelligence algorithms and user training procedures accordingly. Collecting such data bases itself requires an interdisciplinary and international collaboration: it cannot be done by a single lab. This is the reason why we have gathered a large European consortium (20+ labs, 30+ researchers). This consortium is a unique opportunity to provide an international, interdisciplinary and intersectoral training program that will enable the emergence of the next generation of BCI specialists. At present, such a program does not exist, most of the PhD students being trained in a unique and disciplinary lab. Yet, to emerge, BCIs require experts who are able to speak with each other and to understand the all the challenges associated with BCIs, be they related to different disciplines (neuroscience, psychology, engineering, artificial intelligence, ehics, ...) or to different sectors (fundamental research, clinics, industry, ...). With our consortium, we will train a PhD student network from both the theoretical and experimental standpoints. We will provide a common core curriculum to help them apprehending the different aspects of the field, as well as specialy courses that will enable them to acquire high quality skills corresponding to their carreer plan. They will apply their skills by contributing to the data collection for the open database and then use the latter to innovate through different research projects. This innovation will result in drastically improved efficiency and usability of BCIs, and will favour their democratisation, through the improvement of hardware (brain activity measures), software (signal processing) and user training. Regular meetings between the researchers, clinicians, industrials and PhD students wille create an emulation in the consortium. The PhD students will also follow training sessions specifically on ethics, open science, communication and scientific outreach. Through this approach, we aim to train a generation of specialists able to fully understand the challenges associated with BCIs (be they scientific, technical or societal), open-minded, honest and transparent in their use/development of neurotechnologies, and having a wide range of professional integration possibilities.