A zeptosecond multi-MeV laser pulse may either excite a "plasma" of strongly interacting nucleons or a collective mode. We derive the conditions on laser energy and photon number such that either of these scenarios is realized. We use the nuclear Giant Dipole Resonance as a representative example, and a random-matrix description of the fine-structure states and perturbation theory as tools. PACS numbers: 42.50Ct, 24.30Cz, 24.60Dr Purpose. Qualitative Considerations. With the start of the construction of ELI (the "extreme light infrastructure") [1] or with extant ultra-intense laser facilites like NIF (if reconfigured as femtosecond pulse systems) [2,3] nuclear spectroscopy using intense high-energy laser beams with short pulses has become a realistic possibility. Indeed, it is envisaged to generate in the decade ahead pulsed laser light with photon energies of several MeV and pulse lengths of 10 −19 seconds by coherent Thomson backscattering [4][5][6]. This will be possible provided that present intense experimental and theoretical efforts will validate the concept of an electron mirror [6]. These very exciting developments call for a theoretical exploration of the expected nuclear excitation processes. In the framework of the nuclear shell model (a mean-field approach with a residual nucleon-nucleon interaction), two scenarioss come to mind, distinguished by time scales. (i) The time scale for the residual interaction is large compared to the time scale for laser excitation of individual nucleons. Then a single laser pulse containing N ≫ 1 photons excites many nucleons more or less simultaneously. The resulting "plasma" of excited and interacting nucleons (distantly similar to the initial stage of a precompound reaction) is instable. Nucleons excited above particle threshold with low angular momenta are emitted instantaneously. The remainder of the system is equilibrated by the residual interaction. Exciting questions are: What are the mean mass number, the mean excitation energy and the mean angular momentum of the resulting compound nucleus? How big are the spreads of these quantities? Presumably it will be possible to study compound nuclei at excitation energies and spin values not accessible so far. (ii) The time scale for the residual interaction is sufficiently short compared to the time between two successive photon absorption processes. Then the nucleus relaxes after each photon absorption process. Single photon absorption leads to a collective mode (typically the Giant Dipole Mode), and multiple photon absorption within the same laser pulse may lead to the formation of higher harmonics of that mode. Thus, scenarios (i) and (ii) lead to extremely different forms of nuclear excitation.In this Letter we establish the time scales and the resulting conditions on the mean photon energy E L and the number N of photons relevant for scenarios (i) and (ii). We do so by studying scenario (ii) in detail. We show that scenarios (i) and (ii) both occur for realistic choices of E L and N . We also show that scena...