Today's high-performance OLAP systems are either implemented fully in-memory (and are thus expensive), or employ SSDs as their primary storage medium, which have a much better price/performance ratio. In this work, we suggest a novel GPU-in-data-path architecture for analytical query processing of SSD-resident data, that leverages a GPU to accelerate the SSD I/O path and thus can achieve almost in-memory bandwidth. The main idea is to stream data in heavy-weight compressed blocks from SSDs directly into the GPU and decompress it on-the-fly as part of the table scan to inflate data before processing it by downstream query operators. Furthermore, we employ novel GPU-optimized pruning techniques that help us further inflate the perceived read bandwidth and design hybrid GPU-CPU co-processing plans for situations where full GPU execution of a query is not feasible.