We argue that the relative yields of Υ states observed at the LHC can be understood as bottomonium states coming to early thermal equilibrium and then freezing out. The bottomonium freezeout temperature is approximately 250 MeV. We examine its systematics as a function of centrality. We remark on the interesting differences seen by the CMS and ALICE experiments in the charmonium sector.PACS numbers: 12.38. Mh, 11.15.Ha, 12.38.Gc A thermal medium can disrupt the formation of a bound state of a heavy quark, Q and antiquark,Q [1]. This is a multi-scale problem, involving the temperature T and the quark mass M . The quark is heavy, i.e., M/T ≫ 1. For the charmonium this ratio is about 5-10 and for the bottomonium it is around 15-30; so both these flavours can be considered to be heavy in this sense. Also, M/Λ ≫ 1, where Λ ≃ m ρ /2 is the typical scale of QCD [27]. This second comparison implies that in theQQ bound state the quarks are slow, with velocity v 2 ≃ 0.23 for charm and v 2 ≃ 0.08 for bottom [2]. In NRQCD counting, the binding energy B ≃ M v 2 ≃ Λ in both cases [3]. For the temperature range of relevance, we find that B/T ≃ Λ/T ≃ 1. As a result, thermal effects can drastically modify the bound state.Since a thermal medium can be formed in PbPb collisions, but not in pp collisions, there should be certain systematic differences between the yields in these two cases [4]. Such effects were first seen at the SPS [5], but there were important backgrounds to the signal which came from the initial state via parton density effects [6] and the final state through comover interactions [7]. Furthermore, the observed effects switched on slowly with centrality and nuclear size, and so were difficult to interpret clearly [8].At the LHC the experimental situation has changed drastically. Initial state effects cannot be the explanation, since the relevant values of the Bjorken variable are almost equal for the three bottomonium states, and a different common value for the two charmonium states, implying that parton density effects would be the same. Final state comover interactions were invoked even at the LHC in order to explain the observed suppression of J/ψ [13]. The comoving material which is thermalized gives rise to the signal. The comovers responsible for the background are unthermalized and relatively cold spectators from the initial PbPb collision. However, the data are taken at central rapidity and separated from the spectator fragments by ∆y ≃ 6. So comover interactions cannot be the explanation for the observations. In this cleaner environment it would be interesting to check whether the data allow a thermal interpretation.We begin by noting that the equilibrium density of heavy quarkonia, with massis small. As a result, the mean distance between quarkonia in thermal equilibrium, λ, is large. In fact, we find that T λ = z −1/3 T /M exp(2M/3T ). For z = 1 this dimensionless number is over 100 for charmonia and over 10 million for bottomonia. So, for both the charm and bottom systems, the exponential dominate...