According to ExtremeTech, nVidia has chosen the "Tesla" moniker for their line of GPGPUs. There will apparently be boards & entire servers based on 'em, too.

"The S870 server will contain between 48 to 52 GPUs, with each Tesla GPU producing about half of its theoretical 518 gigaflops at peak performance while running real-world applications said Andy Keane, the company's general manager of the GPU computing group at Nvidia."