Aims. Submillimetre galaxies (SMGs) are bright sources at submillimetre wavelengths (F850 μm > 2−5 mJy). Made up of mostly of high-z galaxies (z > 1), SMGs are amongst the most luminous dusty galaxies in the Universe. These galaxies are thought to be the progenitors of the massive elliptical galaxies in the local Universe and to reside in massive haloes at early epochs. Studying their environments and clustering strength is thus important to put these galaxies in a cosmological context. Methods. We present an environmental study of a sample of 116 SMGs in 96 ALMA observation fields, which were initially discovered with the AzTEC camera on ASTE and identified with high-resolution 1.25 mm ALMA imaging within the COSMOS survey field, having either spectroscopic or unambiguous photometric redshift. We analysed their environments making use of the latest release of the COSMOS photometric catalogue, COSMOS2015, a catalogue that contains precise photometric redshifts for more than half a million objects over the 2 deg2 COSMOS field. We searched for dense galaxy environments computing the so-called overdensity parameter as a function of distance within a radius of 5′ from the SMG. We validated this approach spectroscopically for those SMGs for which spectroscopic redshift is available. As an additional test, we searched for extended X-ray emission as a proxy for the hot intracluster medium, performing an X-ray stacking analysis in the 0.5−2 keV band with a 32″ aperture and our SMG position using all available XMM-Newton and Chandra X-ray observations of the COSMOS field. Results. We find that 27% (31 out of 116) of the SMGs in our sample are located in a galactic dense environment; a fraction that is similar to previous studies. The spectroscopic redshift is known for 15 of these 31 sources, thus this photometric approach is tested using spectroscopy. We are able to confirm that 7 out of 15 SMGs lie in high-density peaks. However, the search for associated extended X-ray emission via an X-ray stacking analysis leads to a detection that is not statistically significant.