The field of data-intensive computing is witnessing a significant shift towards leveraging GPU acceleration to improve performance. Recent developments focus on optimizing data loading, processing, and analytics on large-scale datasets. Notably, advancements in deep learning models trained on Earth observation data have led to increased throughput and accuracy. Additionally, innovative approaches to data indexing, compaction, and transformation are being explored to reduce costs and improve efficiency. GPU acceleration is also being applied to SQL analytics, enabling fast query execution on compressed data. These advancements have the potential to revolutionize various applications, from geospatial analysis to data centers. Noteworthy papers include: Optimizing Cloud-to-GPU Throughput for Deep Learning With Earth Observation Data, which presents a novel approach to optimize data loading for deep learning models. Mycelium: A Transformation-Embedded LSM-Tree, introduces a new data structure that embeds transformations into the compaction process, reducing IO costs and amplification. GPU Acceleration of SQL Analytics on Compressed Data, proposes a set of methods for running queries directly on compressed data, achieving speedups of an order of magnitude compared to state-of-the-art commercial CPU-only analytics systems.