Not discussed here is the OCP initiative which I believe to be the biggest upcoming disruption beyond today's ongoing aggregation at data centers. In addition, server count may be held in check while node count per server will soar higher.
It is very shocking news that Loongson Technology Corp. Ltd. is sampling its first commercial microprocessor, this might become a strong competitor for Inter and AMD, looking at the market share of China. This processor might an other flavour of ARM based designs.
Are MIPS designs really a viable means of competition in this market? It seems like a two horse race between ARM and Intel---or maybe its between all the different companies who are licensing ARM and Intel, but a MIPS-based Chinese company would be coming a bit out of left field.
1) OCP is promoting an alternative larger form factor that will support > node count
2) OCP also has aspirations to make CPUs interchangeable across server chassis, in this case you just replace the CPU module instead of the whole server sled. In this case it may disrupt Intel Tick/Tock dependency and creates additional server OEM challenges
3) It's all about higher density, lower cost, more tuned to high density configurations. Existing XEONs don't scale well given they typically must be augmented with so much support circuitry. I beleive the biggest configuartion is 4 XEONs per PCH. Competitive suppliers are architecting SOCs that can be more easily scaled.
At a first glance it seems strange considering Loongson Tech as in competition of Intel, but the demand of processors is of such a high volume that if Chinese govt change the import strategy all the Chinese companies will turn towards Loongson processors, And technically The Loongson 3 has the specific benefit of speeding up Intel x86 CPU emulation. So it can be easily replaced in x86 based devices.