At its Technology and Manufacturing Day, Intel (NASDAQ:INTC) went into some detail about its plans for future chips. Intel executive Murthy Renduchintala explained that it will build chips going forward by combining several discrete silicon dies connected using the company's exciting new Embedded Multi-Die Interconnect Bridge (EMIB) technology.
In my initial article covering Renduchintala's comments, I mainly went over about the potential for Intel to build a wider range of products to better serve different subsegments of the overall personal computer market. However, there's another reason for investors to be excited about this new product development methodology.
More efficient research and development spending
Intel spent $12.74 billion in R&D during 2016, a figure that dwarfs the R&D budgets of virtually all its peers in the chip industry.
However, just throwing money at product development -- as we saw with Intel's failed attempts to become a major player in the tablet and smartphone applications processor market -- isn't enough. That spending needs to be focused to build the right technologies and products and it needs to be efficient to maximize shareholder returns.
"Clearly, one of the key things on my mind is how I most effectively use Intel's [R&D] budget," Renduchintala began. "And what I don't want to be doing is spending a large amount of [research and development] porting [intellectual properties] from node to node that doesn't get any inherent performance benefit." Instead, Renduchintala wants Intel's R&D efforts to be focused on "absolutely competitive advantage-generating capabilities."
To that end, Renduchintala indicated, the company would move its CPU cores and its graphics and media technology to its newest manufacturing technologies, but other pieces of intellectual property -- ones that Renduchintala says don't benefit from newer manufacturing technologies and can even see performance degradation from such moves -- can stay behind on older technologies.
Not only does Intel not need to spend the R&D dollars to bring those intellectual properties to newer manufacturing technologies, but it sounds as though there will be product performance/quality benefits, too.
When will we see such chips?
Although Intel has talked up the importance of this new product development methodology, what the company has yet to publicly disclose is when we'll start to see it releasing products that mix and match different intellectual properties built on different manufacturing technologies.
If I had to guess, I'd say that we should see this technology in action -- at least in Intel's personal computer processor lineup -- with its Ice Lake architecture. Products based on the Ice Lake architecture aren't expected to launch until, at the earliest, the second half of 2018, with perhaps a broader product rollout happening in the first half of 2019.
What I think will be interesting to see is how Intel ultimately splits up these chips and what kinds of different product variants the company builds. My hypothesis is that Intel will be able to build a wider variety of chips that are specifically tuned for each of the subsegments of the personal computer market, which should help improve Intel's margins, product mix, and even afford it the flexibility to put out new products to respond to market demand/conditions at a quicker pace.
I hope that it won't be too long before the general public gets to see the impact of this development methodology on Intel's products and financial results.