Using a regional GPS data set including ~190 stations in Asia, from Nepal to eastern Indonesia and spanning 11 years, we update the present-day relative motion between the Indian and Sundaland plates and discuss the deformation taking place between them in Myanmar. Revisiting measurements acquired on the Main Boundary Thrust in Nepal, it appears that points in southern Nepal exhibit negligible deformation with respect to mainland India. Including these points, using a longer time span than previous studies, and making an accurate geodetic mapping in the newest reference frame allows us to refine the present-day Indian motion. Our results confirm that the current motion of India is slower than predicted by the NUVEL-1A model, and in addition our India-Eurasia motion is significantly (~5 mm/yr) slower than previous geodetic determinations. This new Indian motion, combined with a refined determination of the Sundaland motion, gives way to a relative India-Sunda angular velocity of 20.2¿N, 26.1¿E, 0.370¿/Myr in ITRF2000, predicting a relative motion of 35 mm/yr oriented N10¿ at the latitude of Myanmar. There, the Sagaing Fault accommodates only 18 mm/yr of right-lateral strike slip, only half of the shear component of motion. We present two models addressing how and where the remaining deformation may occur. A first model of distributed deformation implies convergence on the Arakan subduction (the northern continuation of the now famous Sumatra-Andaman Trench) and wrench faulting in the Arakan wedge. The second model uses localized deformation, where deformation observed west of the Sagaing Fault is entirely due to elastic loading on a faster and oblique Arakan subduction (23 mm/yr). This latter model predicts that a major earthquake of Mw 8.5 may occur every century on this segment of the subduction.