The phytoplankton spring bloom constitutes the majority of annual Arctic primary production. In the Eurasian Arctic it is usually composed of various diatoms and the haptophyte <em>Phaeocystis</em>; recently, <em>Phaeocystis</em>-dominated blooms seem to be expanding. Low silicic acid (Si) availability has been hypothesised to promote <em>Phaeocystis</em> blooms, though the key drivers of <em>Phaeocystis</em> and diatom competition are not yet well understood. To investigate how Si availability controls bloom succession and biogeochemical fluxes we conducted a 3-week mesocosm experiment in Kongsfjorden, Svalbard in May 2024 on the spring phytoplankton community. The initial phytoplankton community was composed predominantly of <em>Phaeocystis</em> and the diatom <em>Chaetoceros</em>. Eight 450-L mesocosms received an Si gradient (0–40 µM) with nitrogen and phosphorus additions every 48 hours and we tracked nutrient uptake rates, community composition, elemental cycling and export. The nutrient additions stimulated phytoplankton growth across all Si levels. Phase I (Days 0–14) was dominated by a <em>Phaeocystis </em>bloom, terminating synchronously due to nitrogen and phosphorus depletion despite variable Si availability. In phase II (Days 16–22) <em>Chaetoceros</em> emerged as the dominant species with clear Si gradient effects: <em>Chaetoceros spp. </em>sustained exponential growth (µ = 0.40 d<sup>-1</sup>) and achieved higher abundances in >3 µM Si treatments, while collapsing in Si-depleted mesocosms. Overall, particulate organic carbon export fluxes were low, but were strongly correlated with biogenic silica export fluxes, and increased exponentially in phase II of the experiment reflecting increasingly diatom-dominated sinking material. Our results demonstrate that Si availability does not govern initial <em>Phaeocystis</em>–diatom competition, but is the primary determinant of diatom bloom magnitude. The nutrient storage capacity of <em>Chaetoceros</em> confers resilience under transient depletion, while micrograzing and viral infection further mediate community successions. This study underscores the importance of integrating cell-specific and ecosystem-level measurements over extended timescales when predicting climate-driven shifts in Arctic primary production and carbon export.