Evolution of the spatial pattern of ocean surface warming impacts global radiative feedbacks and climate sensitivity, but different climate models have yielded different estimates of the spatial pattern in future climate change. Paleoclimate data, particularly those from distant past warm climates would provide constraints on the future, equilibrium warming pattern. Its comparison with model simulated warming pattern under equilibrium conditions would also help shed light on which ones of these predictions have all necessary physics. Here we used a novel regression-based technique that removes time variations and therefore allows past climate data to be directly compared with modern and future climates. Long (10 million years), globally distributed paleoclimate sea surface temperature timeseries were regressed onto the record from the Western Pacific Warm Pool (WPWP) – the warmest endmember of the global ocean, which was used as a benchmark to define the amplified warming of non-WPWP regions relative to the WPWP. This led us to identify a spatial pattern of amplified warming similar to some equilibrated model simulations under high CO₂ conditions. Paleoclimate and model output data collectively help us to define an equilibrium warming pattern, which turns out to be different from the transient pattern of warming that characterizes the past 170 years, illuminating our potential future path of the “pattern effect”.