The study of nuclei far from stability is an increasingly important part of a nuclear physics portfolio [1,2,3]. As radioactive beams gradually expand the borders of the nuclear landscape, theoretical modeling of the nucleus is changing in significant ways. The crucial question for the field [2], namely ``What binds protons and neutrons into stable nuclei and rare isotopes?," nicely underlines this point: indeed, the data on rare isotopes with the large neutron-to-proton imbalance indicate that there are many gaps in our present understanding.

Short-lived exotic nuclei offer unique tests of those aspects of the nuclear theory that depend on neutron excess [4,5]. The major challenge is to predict or describe in detail exotic new properties of nuclei far from the stability valley, and to explain the origins of these properties. New ideas and progress in computer technology have allowed nuclear theorists to understand bits and pieces of nuclear structure quantitatively.

The new experimental developments inevitably require safe and reliable theoretical predictions of nuclear properties throughout the whole nuclear chart in two main directions: (i) along the isospin axis, i.e., going outwards from the beta stability line to the neutron and proton drip lines, and (ii) towards the uncharted territory of super-heavy elements at the limit of mass and charge. The tool of choice is the nuclear density functional theory (DFT) based on the self-consistent Hartree-Fock-Bogoliubov (HFB) method. The key component is the universal energy density functional, which will be able to describe properties of finite nuclei as well as extended asymmetric nucleonic matter. The development of such a universal functional, including dynamical effects and symmetry restoration, is one of the main goals of the field.

By employing various criteria (agreement with measured masses, radii, low-lying excited states, giant vibrations, rotational properties, and other global nuclear characteristics), one aims at adjusting the coupling constants of the functional. By finding correlations between parameters, one hopes to reduce their number and to understand physical reasons why different parametrizations yield similar results. One may also want to expand the parametrizations to cover aspects dictated by physics arguments and/or motivations coming from the effective field theory and QCD. The main challenges in this quest have been nicely summarized through five questions [6]:

- What is the form of the nuclear energy density functional?
- What are the constraints on the nuclear energy density functional?
- What is the form of the pairing functional?
- How to account for quantum correlations and symmetry-breaking effects?
- How to optimize computational techniques and error analysis?

The aim of this paper is to briefly review the present state of the large-scale microscopic nuclear mass calculations and to discuss improvements needed. Section 2 introduces the DFT and Skyrme-HFB method. Some details concerning global mass calculations are given in Sec. 3. The long-term program is outlined in Sec. 4. Finally, the summary is given in Sec. 5.