An FPGA is like computer chip Playdoh, it's a special type of CPU circuitry that is completely re-programmable, it can be a high speed audio signal processing chip one day, a high speed hardware compression chip the next. These FPGA's have been around for a long time, they were initially used to prototype silicon designs before they were cast into permanent chips and mass produced, but Altera (now owned by Intel & the world's leaders at FPGA's) have made these chips so fast that you can run commercial applications on them. Microsoft and other cloud platforms + providers are leveraging this, adding in FPGA's that can be pre-programmed to run A.I code as if though it was a specialized circuit, but being able to patch and self-patch a piece of circuitry every day - you can imagine the speed and benefits that has. The real question is, if we let an A.I. chip self-improve at such pace, should we just call it Skynet and get ready for the war (clearly... I'm joking)?
Today’s Project Catapult combines an FPGA integrated into nearly every new Microsoft datacenter server with a unique distributed architecture. The distributed architecture deploys FPGAs as an addition to each datacenter server, rather than a bolt-on isolated cluster, to create an “acceleration fabric” throughout the datacenter. This elastic reconfigurable acceleration fabric provides the flexibility to harness an individual FPGA or up to thousands of FPGAs for a single service.