The temporal variation of the carbonate system, air-sea CO2 fluxes and pH is analyzed in the Southern Indian Ocean, south of the Polar Front, based on in-situ data obtained from 1985 to 2021 at a fixed station (50°40’S–68°25’E) and results from a neural network model that reconstructs the fugacity of CO2 (fCO2) and fluxes at monthly scale. Anthropogenic CO2 (Cant) was estimated in the water column and detected down to the bottom (1600 m) in 1985 resulting in an aragonite saturation horizon at 600 m that migrated up to 400 m in 2021 due to the accumulation of Cant. In subsurface, the trend of Cant is estimated at +0.53 (±0.01) µmol.kg-1.yr-1 with a detectable increase in recent years. At the surface during austral winter the oceanic fCO2 increased at a rate close or slightly lower than in the atmosphere. To the contrary, in summer, we observed contrasting fCO2 and dissolved inorganic carbon (CT) trends depending on the decade and emphasizing the role of biological drivers on air-sea CO2 fluxes and pH inter-annual variability. The region moved from an annual source of 0.8 molC.m-2.yr-1 in 1985 to a sink of -0.5 molC.m-2.yr-1 in 2020. In 1985–2020, the annual pH trend in surface of -0.0165 (± 0.0040).decade-1 was mainly controlled by anthropogenic CO2 but the trend was modulated by natural processes. Using historical data from November 1962 we estimated the long-term trend for fCO2, CT and pH confirming that the progressive acidification was driven by atmospheric CO2 increase. In 59 years this leads to a diminution of 11 % for both aragonite and calcite saturation state. As atmospheric CO2 will desperately continue rising in the future, the pH and carbonate saturation state will decrease at a faster rate than observed in recent years. A projection of future CT concentrations for a high emission scenario (SSP5-8.5) indicates that the surface pH in 2100 would decrease to 7.32 in winter. This is up to -0.86 lower than pre-industrial pH and -0.71 lower than pH observed in 2020. The aragonite under-saturation in surface waters would be reached as soon as 2050 (scenario SSP5-8.5) and 20 years later for a stabilization scenario (SSP2-4.5) with potential impacts on phytoplankton species and higher trophic levels in the rich ecosystems of the Kerguelen Island area.