Explore the portable communications interface for Earth System Models in this JuliaCon 2024 conference talk by Sriharsha Kandala. Dive into CliMA's Earth System Model (ESM) design, which automatically learns from diverse data sources to reduce and quantify uncertainties in climate predictions. Discover the components of the ESM software stack, including atmosphere, radiation, and land models, as well as a coupler. Learn about ClimaComms.jl, the high-level communications interface and computational backend for various ESM components. Examine the most expensive communication patterns in the atmospheric solver, focusing on Direct Stiffness Summation. Understand how the communication interface is implemented to support various computational backends, including single-threaded and multi-threaded CPUs and GPUs, as well as different transports like Serial and MPI. Gain insights into how this implementation enables users to run distributed simulations across a wide range of architectures, enhancing the portability and efficiency of Earth System Models.