We had some in the 1990s like 3dfx (GPU) and Cyrix (CPU):
http://en.wikipedia.org/wiki/3dfx
http://en.wikipedia.org/wiki/Cyrix
We had a discussion about it (as off-topic discussion) today: https://news.ycombinator.com/item?id=7658275
Simply ramping up the clock speed is not going to happen much; the problem of waiting for your data to traverse the huge distance from the socketed DRAM all the way over to the processor is already severe.
Founded by experts in machine learning, neuroscience, processor architecture, and chip design, Nervana Systems is bringing unprecedented scale and simplicity to the application of brain-inspired algorithms. Deep learning has emerged as the leading strategy for making sense of a wide variety of data but is very computationally intensive. We are developing a scalable hardware solution to solve these types of problems. By making unsupervised learning dramatically faster and more scalable, customers at all levels can derive meaningful insights from their data, an ability previously available only to an elite few companies.