The general idea of Nanotechnology is not new – it has been studied since Nobel laureate Richard Feynman outlined the idea in a speech in 1959 – but it's only recently that the progress carried out in the various fields of material, optic, physic and engineering have allowed scientists to make Feynman's vision comes true. A large part of tools used to interact with nanoscale structures are based on scanning probes technology founded with the invention of the scanning tunneling microscope (STM) by Gerd Binnig and Heinrich Rohrer of IBM's Zurich Lab in 1981. The working principle is based on a micro-lever carrying a probe on its end which is brought in proximity of the sample to be manipulated/studied. The deflection of the lever induced by the atomic forces in the vicinity of the sample is measured by a sensor, allowing the reconstruction of the desired information, be it topographical information or information about the nature of the sample (DNA molecule identification, etc.). Today, the scanning probe microscope (SPM) technologies based on individual probe have reached their limit, principally by the slow data acquisition frames rate due to the intrinsic sequential readout of these systems. This last decade, great deals of efforts have been carried out for developing multiprobes SPM in order to increase range and acquisition speed. One obvious path is to make multiple cantilever probes on one chip which are controlled independently and coupled with a multiplexed cantilever detection method. Developed in linear or in 2D configurations, manufactured starting from very diverse materials and coupled with a large range of sensors, the scanning probes devices extend nowadays their applicability to many different domains, starting from the observation to the direct interaction with nanoscale structures. However, an important limitation appears when using multiprobes SPM with multiplexed integrated sensors. Indeed, as density of probes increases, number of contacts and interconnections wires increase and space for the leads decrease as much. Moreover, this contributes to enhance the complexity of the parallel electronics interface which is preferably wanted based on standard CMOS process. With regards to these limitations, current studies and development depict a limit up to 100 × 100 probes/mm2. This thesis presents an entirely new approach with the promise to break several of theses barriers and is to our knowledge the first successful experiment which shows how to overcome the interconnection limit. The setup proposed here separates physically an entirely passive probes array, without any electrical correction, from massively parallel CMOS standard readout electronics. The key idea is the readout of a Scanning Probes Microscope (SPM) array by Digital Holographic Microscopy (DHM). This technology directly gives phase information at each pixel of a CCD array. This means that no contact line to each individual SPM probe is needed. The data is directly available in parallel form, at the acquisition rate of the camera. Moreover, the optical setup needs in principle no expensive components, optical (or, to a large extent, mechanical) imperfections are compensated in the signal processing, i.e. in electronics, given the system the potential for a low cost device with fast Terabit readout capability. The present contribution describes the working principle of the whole device. The acquisition sequence start with the recording of the holograms containing the cantilevers bending angular deflection information, a spatial filter derived from the intensity image allowing us to keep only the functional information from the whole area of the device. Thus, the phase information is reconstructed in real-time, reproducing the topographical information of the sample below the SPM array. The study of the requirements and the design rules leading to the micro-fabrication process of such a functional SPM array is also reported in this document. This process is constituted of four main steps: In the first two steps, a direct tip method which defines the shape of the probes is coupled with the structuring of the cantilevers in the silicon membrane of a silicon-on-insulator (SOI) wafer. A Third step consists in a thin-film nitride deposition which is exploited for the functional out-of-plane bending of those passive cantilevers. Finally, the release of the devices is operated after a backside plasma etching, leading to perfectly functional 27 × 27 SPM array on a 4 × 4 mm square area. Using a commercial DHM for the acquisition of the phase information of this first 729 two-dimensional free bending SPM cantilever array, the real-time topography reconstruction of varied samples brought in contact is validated. Regarding to current storage technology's vertical resolution and scanning acquisition time, the approvement of the concept enlarged to a high density SPM array readout is finally discussed.