This is clearly not that.
Edit: Intelligence Processing Unit. https://www.graphcore.ai/products/
I assume that the Bionic neural engine in Apple's A11 and later is an example of this?
It’ll likely end up just like GPU though, as part of the industry’s nomenclature.
Caveat emptor: the published numbers (_any_ numbers, not just Graphcore's) are mostly bullshit unless code is also published and hardware is available for independent measurement. There's no way you're getting the claimed 13TFLOPs out of your shiny new NVIDIA GPU. Take an off the shelf resnet50 (4GFLOPS) and witness it run at about 700 samples per second, which pencils out to 2.8TFLOPs not 13. Still amazing (and still easily 10x the high end CPU throughput), but perf claims are often exaggerated by as much as 10x even by big names, let alone startups.