On another note, though fairly uncommon in France, the option of a call for corporate sponsors should also be explored, particularly among successful AI startuppers. The needs in terms of budget of the measures recommended for setting up the 3IA institutes and the RN3IA (and more generally of all of the recommendations in this chapter) are very modest in comparison with the expenses that would need investing in the other sectors, when the return on investment, over the medium—and long- term admittedly, will be immense if the aim of creating a thriving fabric for entrepreneurs is achieved. Integrating this network in the European AI research area Via its national coordination, the RN3IA will be able to become the lead correspondent for our European partners to ensure French research in AI connects with the main European AI centers (DFKI and MPI in Germany, Alan Turing Institute in Great Britain, IDSIA and Écoles Polytechniques Fédérales in Switzerland, CWI in the Netherlands, IRIDIA in Belgium, Sapienza Roma and the other robotics and AI research centers in Italy, etc.), not least amid the emergence of a major European AI network, within which this network of French institutes will naturally be expected to represent the French ecosystem. Initially, precedence will be (and already is being) given to the Franco-German partnership. The form that such a network might take is not yet known, but it could be modelled on the EMBL (European Molecular Biology Laboratory), which has been operating successfully since 1974. There are also hopes for other European partnerships, with the instruments of the H2020 programme8, like the current public-private partnerships in robotics and Big Data. But each 3IA institute will, of course, be given scope to forge cooperations with other partners, whether or not European, based on its specific features and the personal relationships of its researchers. We have partnerships in place with our counterparts in Quebec, for example, which would be worth nurturing with respect to the momentum in AI that is gathering at the moment. A stronger researcher presence alongside French Tech entrepreneurs must also be encouraged at flagship European or international events (Consumer Electronic Show, Web Summit and Founders Lisbon for example). 2. Computing Means for Research The 3IA institutes must have computing tools at their disposal that can rival the almost unlimited means of the leading private stakeholders. There are, however, a number of needs, of different types, which cover the different stages of research, development and life of products. Indeed, in some areas such as machine learning, the development cycle entails two key stages: learning and inference. The speed and performance of the learning stage depend on the scale of physical means devoted to them, particularly in terms of dedicated processors (mainly GPUs today). 8. Horizon 2020, the current EU Research and Innovation programme. 74