Intel shows huge monolithic Xe GPU on Twitter, calls it "The Father of All"

Intel has been teasing Xe-GPU on twitter. Now lets immediately kill speculation, this is about an XE-HP(C) GPU, the stuff that is found in professional servers at data-centers, etc. The chip shown is HUGE, I mean literally bigger than an EPYC processor. 

That big size immediately makes is a candidate for a GPU based on multiple dies, again much like what AMD is doing with Threadripper and EPYC, it seems to be based on an LGA socket design which you can drop onto a motherboard. The photo shows a battery next to the GPU, presumably AA, so you can get a grasp of that size. On the Intel Graphics Twitter account, an update was posted on the Xe graphics processor, highlighting that samples would be ready showing a photo of a rather big GPU package, and big is like 3700 mm² big. Such a gigantic chip with a need for a server-like socket will probably be a Ponte Vecchio chip. The first server chips would be made on Intel's 7nm node.

Ex AMD's Raja Koduri, chief architect posted this in the tweets, "The first GPU with IEEE FP32 support I worked on was in 2005. 321M transistors and 32 FP32 Ops/clk. GPU compute density increases continue to be a shining point for Moore’s law. Tens of billions of transistors and tens of thousands of ops/clk.", which was later deleted saying this processor is a "baap of all",  meaning "father of all", through tens of billions of transistors", when translated from Hindi.