Motivated by (i) the widespread reliance of industrial aerodynamics on computationally inexpensive, low-fidelity models—typically accurate only near nominal operating conditions and prone to degradation when nonlinear and turbulent effects become important—and (ii) the maturity of discontinuous Galerkin (DG) high-fidelity discretizations for nonlinear, compressible, and turbulent flows, yet with runtimes that remain challenging due to their large number of degrees of freedom even for relatively simple models, we present an in-house solver based on the Hybrid Discontinuous Galerkin (HDG) method and implemented on GPU architectures for the ultimate goal of aerodynamic simulation of floating offshore wind turbines. HDG has been shown to retain DG-level accuracy while reducing the globally coupled unknowns to roughly one third (see left figure; mesh 1 corresponds to the coarsest mesh), which can translate into up to a twofold reduction in runtime. Moreover, for turbulent flows, HDG combined with the one-equation Spalart–Allmaras model has demonstrated accuracy comparable to finite-volume approaches while using approximately one quarter of the elements, and it preserves its characteristic optimal convergence (order p+1 for polynomial degree p) and superconvergence (order p+2). Our GPU-accelerated HDG framework, initially developed and validated for benchmark second-order elliptic problems, achieves up to 15× speedup in wall-clock time when compared to a CPU implementation—particularly for high-order discretizations on fine meshes (see right figure, finest-mesh results).