Shares of Intel (NASDAQ:INTC) rallied 23% this year as the chipmaker impressed investors with solid earnings growth and the aggressive diversification of its portfolio beyond PCs and data centers. But can Intel continue rising next year? Let's see where Intel will focus its efforts to find out.
New x86 chips
One of the biggest changes to Intel's business model was its shift from a two year, two-stage "tick-tock" upgrade cycle to a 30-month three-stage "process, architecture, optimization" cycle last year.
Intel made that change because it was becoming increasingly difficult to shrink its die for next-gen chips and keep up with Moore's Law, a five-decade claim that the number of transistors per dense integrated circuit would double every two years.
For the 14nm generation, Intel released the last-gen Broadwell in 2014, the first current-gen Skylake in 2015, and its Kaby Lake, Kaby Lake R, and Coffee Lake successors this year. In 2018, it will launch Cannonlake, a 10nm Skylake chip.
This next-gen CPU should widen Intel's performance gap against AMD (NASDAQ:AMD), which started offering performance comparable to that of Intel's Kaby Lake CPUs with its cheaper Ryzen chips earlier this year.
Pivoting toward graphics
Intel is the largest GPU maker in the world thanks to its massive reach in lower-quality integrated graphics solutions, but NVIDIA (NASDAQ:NVDA) and AMD dominate the higher-end discrete GPU market with their add-in boards.
However, Intel recently partnered with AMD to integrate the latter's Radeon graphics into its new x86 chipsets. It also formed a new Core and Visual Computing Group -- led by Raja Koduri, AMD's former Radeon chief -- to produce its own discrete GPUs.
This could be bad news for NVIDIA and AMD, especially if Intel leverages its scale to produce cheaper GPUs and bundles them with CPUs or other products.
Countering NVIDIA in the data center market
Intel controls about 99% of the data center chip market with its flagship Xeon processors. However, NVIDIA also established a foothold in that market with its high-end data center GPUs, which are paired with Intel's CPUs for machine learning purposes.
That trend is worrisome for Intel, since NVIDIA claims that its Tesla GPUs can outperform Intel's Xeon Phi in certain high-performance computing (HPC) applications. As a result, enterprise customers might postpone their CPU upgrades in favor of buying new GPUs. This would hurt the growth of Intel's data center business, which already slowed from double-digit sales growth to single-digit growth over the past year.
Intel's answer to NVIDIA is a combination of newer Xeon Phi chips (Knights Mill) optimized for deep learning tasks and the programmable chips it gained through its acquisition of Altera. It's unclear if Intel can halt NVIDIA's advance into the HPC market, but it will likely make it a major priority next year.
Expanding its higher-growth businesses
Intel generated 86% of its revenue from sales of Client Computing (PC and mobile) and Data Center chips last quarter. Client Computing sales were flat year-over-year, while Data Center sales rose just 7%.
But the rest came from its higher-growth Internet of Things (IoT), memory, and programmable chips businesses, which all posted double-digit growth. Therefore, Intel's priority is to expand those higher-growth businesses to offset the slower growth of its older businesses.
It's accomplishing this with the introduction of custom chipsets of wearables, drones, and other IoT devices, and through major acquisitions in the IoT, automotive, and computer vision markets.
Specifically, investors should keep an eye on its Movidius computer vision chips, which already power various drones and cameras, as well as the development of a driverless platform with its subsidiary Mobileye, which recently partnered with BMW and Fiat Chrysler. Investors should also follow Intel's development of newer 3D NAND and 3D Xpoint memory technologies with Micron, which should strengthen its non-volatile memory business.
Keeping an eye on Apple
One of Intel's worst mistakes over the past decade was losing the mobile chip market to Qualcomm (NASDAQ:QCOM) and other ARM licensees. However, it regained some lost ground when Apple (NASDAQ:AAPL) split the production of its baseband modems for the iPhone between Intel and Qualcomm last year.
The revenue from those modems supported its Client Computing group, but the partnership might not last much longer, since various reports indicate that Apple is ready to produce its own modems to reduce its dependence on Intel and Qualcomm. If that happens, Intel's push into mobile chips could hit another brick wall.
The bottom line
Intel, like many aging tech giants, is trying to evolve with a changing market while defending its core businesses. Therefore, 2018 probably won't be Intel's "best" year, and I'm uncertain that it will rally as much as it did in 2017.
Instead, it should be a transformative year for Intel. Nonetheless, Intel remains fairly cheap at 14 times next year's earnings, and its forward dividend yield of 2.5% gives investors a good reason to sit tight and wait for its investments to pay off.