In October 2017, chip giant Intel (INTC 2.11%) released a new processor family known as Coffee Lake for desktop personal computers. The key selling point of the Coffee Lake chips was that they were now available in configurations with up to six cores, a 50% increase from the number of cores found in prior generation Intel desktop processors.
What's, perhaps, even more amazing is that Intel's mainstream desktop processors (I'm excluding the data center parts that Intel rebadges and sells as high-end consumer parts) have come in configurations with just up to four cores since 2009.
Image source: Intel.
Those cores have gotten smarter and faster over time as Intel has made significant design enhancements, but the company didn't increase the number of cores that it offered for roughly eight years until it faced some serious competitive pressure to do so. There are two key reasons why Intel didn't deliver a core count increase sooner.
10nm delay
A while back, significant evidence surfaced that Intel had planned to launch six- and eight-core versions of its Cannon Lake processor family for desktop computers in the second half of 2016. These chips were intended to be manufactured using the company's 10nm process technology, but since that technology proved (and continues to prove) problematic, the Cannon Lake desktop chips were canceled and new 14nm products were added in their place (Kaby Lake and Coffee Lake).
Intel is expected to release an eight-core variant of Coffee Lake for desktop computers later this year, so in terms of core count increase, the company is about two years off from its original schedule.
Why not earlier, though?
Even if Intel had launched the six- and eight-core Cannon Lake parts on schedule, the company still would've taken seven years to deliver an increase in processor core counts. One common explanation I've read is that Intel simply didn't want to hurt its profit margins -- after all, six-core and eight-core parts, all else being equal, cost more to manufacture than quad-core parts. More cores means bigger chips, which translates into higher manufacturing costs.
The problem with that logic is that Intel could've charged more for the higher core count products that it delivered. In fact, I expect Intel to do just that when it introduces its eight-core Coffee Lake chip within the next few months.
Perhaps, then, Intel didn't think that customers would be willing to pay more for those chips and therefore they wouldn't sell well. In that case, Intel's decision makers could've decided that putting engineering resources on such projects would be a waste of effort and valuable resources.
I think the explanation could be as simple as the following: Intel designed these chips for the laptop computer market (which has been growing as a percentage of total personal computers sold) so what Intel could do in terms of core count was limited by what could fit in the power budget and thermal constraints of increasingly thin and light laptop computers.
Even in that case, the answer seems to ultimately boil down to Intel not thinking it was worth the development costs to build desktop-specific chips with more processor cores. Even the hex-core Coffee Lake part that launched late last year was likely designed with laptops/mobile workstations in mind. The eight-core part that the company is planning to launch soon is the first true desktop-specific processor the company has launched in years.
