Well, don't forget the huge shift in the business industry of turning towards to RDC, terminal services and cloud computing with cheap PC's.
A lot of office computers are being replaced as a simple terminal computer with the PC actually only needing to remote connect to a server or mainframe under a VM environment to run applications and windows, server side. The PC there in becomes nothing more than a TV with input devices. For office applications that don't require a lot of internet traffic for RDC broadcast, its very doable over most internet bandwidth. In theory, you could use a simple $100 nettop using a low end atom cpu or even just an ARM cpu while you stream in everything from you main i7 rig in the other room or in another state.
These desktop CPUs are far more powerful than what the average person needs, which is why the tablet market is booming
You cannot correlate the two directly. You are using logical fallacy to connect a very distant dot.
What's really been happening is that the CPU industry has essentially plateaued for the last 3-5 years with the mainstream markets in terms of performance needs for the average user. Like you said, most people don't need powerful ivy bridge cpu's to run simple everyday apps like office, excel or just general web surfing and video playback. Shit, most cpu's in the past 5-10 years is more than powerful enough to do this.
What shifted the demand was actually the rapid expansion of Intel's new product line, the netbooks with the Asus EeePC's like 5 years ago. It showcased that a simple, small cpu could easily run most people's stuff without hindrance. This only translated within the mobile lineup, not the desktop; for that, that's when 2009 came as nettops exploded as low powered desktop replacements. Then all of a sudden Apple took industry by storm with this growing change in PC needs with the iPad a year later, showing that you could mirror a netbook and nettop with a relatively simple interface for general things that don't need to be done on a full laptop or desktop device. It was $500 iPad vs. sub $400 netbooks at the time.
Had Apple released their ipad before the growth of the netbook/nettop market, I'm not exactly sure if it would have been as successful, but because of the consumer culture mindset was rapidly changing to simple, easy and with just enough performance; they struck gold again. Plus you also have to include the rapid expansion of mobile media consumption with streaming services like pandora, netflix, and hulu were becoming vastly popular around this time as another reasoning to changing computing demands. Not to mention the invention of "the cloud". Amazon is credited for paving this way towards more mainstream usage in the industry with their EC2 data centers.
Will Intel kill off PC's? No way.
Will Intel focus away from main stream cpu's? I think so.
The enthusiast PC market is way bigger than it was 10 years ago, but in the grand scheme of things, we have lost a significant market-share in the computing industry if you include cell phones, mobile devices and mainstream desktops. We live in a very niche market segment where performance minded people will spend more money on a certain product to achieve a level of expectation that they desire; its the definition of marketing and could be applied to anything and everything.
You have to remember that nvidia has been doing very well, mainly because of their niche market of making performance chips. They have diversified in the last few years to cover a wider range of niche markets such as ARM, mobile and parallel computing, not just gaming. What Intel needs to do is follow what nvidia has done and move away from general computing and more towards low powered or performance based processing. Performance per watt is the new game.
As mobile computing increases, Intel should really focus more towards server products as data centers are going to continuously grow. Intel knows this and the reason why they are working hard on ARM and x86 cpu's focused solely for data centers. All we are doing is shifting where the computing will happen; away from the consumer end.
What also isn't being said in that article is that Broadwell chips maybe using a new form of MCM or multi-chip module packaging technique called chip stacking. This could mean the chip would be multi-layered vertically to reduce the overall package surface area and lower the space requirement on the motherboard; best for mobile or compact solutions. It could also mean because of the increased complexity that the only way to connect the cpu is through a much stronger surface connection using BGA or ball grid array like they do for memory chips.
Remember, with something like the i5 or i7's currently, there are 1155 connection points throughout the bottom of the pcb board of the actual cpu. LGA 2011 socket is almost double that and you can physically see that its a huge package/pcb board despite the cpu not being that much bigger relatively. Not to mention the 2011 x79 motherboards are godly huge. So we get to the point of how big is too big for a socket to be effective and practical?
BGA connection points are much much smaller than the normal flip-chip land grid array of the modern sockets. This would leave a much more limited upgrade approach and limited availabilities when buying motherboards & cpu's, but that shouldn't limit your choices too much if motherboard builders are given the chance to solder whatever cpu you want per the request of the consumer. This would probably increase the consumer's cost of the two parts together, but then again, a lot of this is based on speculation and not a lot of facts from Intel.
