It continues to be old vs. new.
That’s according to Seeking Alpha contributor and tech expert Mark Hibben, at least when it comes to the semiconductor sector. His take: As investors assess their holdings, especially among semiconductor stocks, they should steer clear of companies stuck in the old paradigm.
As part of Seeking Alpha’s 2018 Outlook series, editor Michael Hopkins caught up with Hibben to get his take on what’s next within the semiconductor space, what investors can expect from Apple (NASDAQ:AAPL), and how he assesses investment opportunities.
Michael Hopkins (NYSEMKT:SA): Let’s start with Apple (AAPL). There are the new iPhones, focus on Services, and ongoing pushes with iPad, Macs, etc. Could there be something else with Apple in 2018 that will surprise investors?
Mark Hibben (MH): 2018 will probably not see any completely new products from Apple. Apple is reportedly working on “smart glasses,” which would feature AR capability and possibly substitute for a smartphone or tablet. However, those are thought to be about two years away.
That leaves new versions of existing products which may or may not surprise. The most prominent product makeover known to be in the works is the next gen Mac Pro. As I discuss in a recent article, the iMac Pro is a more powerful iMac, but doesn’t meet the need for PCIE expandability that many professionals need. The next Mac Pro has been advertised as a “modular” system, but it’s not clear that it will include PCIE slots.
The Apple Watch Series 3 with LTE may surprise with greatly expanded unit sales. A report from Digitimes claims that Watch sales could hit 23-35 million units next year, which would be about double calendar 2017 sales.
Although Digitimes often passes along specious rumors, I also expect significant growth in Watch sales. I have written about the potential for LTE connected smartwatches to take over voice and text communications functions, and the Watch Series 3 is blazing that trail for Apple.
The Series 3 with LTE actually arrived earlier than I expected, but now that it’s here, Apple will continue to develop its capabilities, especially in the areas of increased talk time.
I also expect Apple to expand its use of 3D sensing to include iPad and other iPhone models. Apple’s recent $390 million award to Finisar (NASDAQ:FNSR) should pave the way for expanded functionality for 3D sensing beyond FaceID. We may see 3D sensors mounted on the backs of iPhones and iPads to support advanced augmented reality features.
Finally, Apple’s autonomous vehicle efforts may see their first public trials in a driverless shuttle service known as Palo Alto Infinite Loop or PAIL. It’s not known when this could get started but 2018 seems likely.
(MH): Since both companies are in the Rethink Technology Portfolio, I try to update members regularly (the most recent being on November 29) and so I’ve been following the legal developments fairly closely. Despite the fact that Qualcomm has been on the defensive regarding its business practices, I consider Qualcomm to have the edge in its legal battle with Apple.
Let me first give a conceptual overview of the battle. One way to think about it is that this is a battle between the new paradigm in semiconductors, represented by Apple, and the old paradigm, represented by Qualcomm. In the new paradigm, mobile device makers design their own processors rather than purchase them from a commodity processor supplier like Intel (NASDAQ:INTC) or Qualcomm. The old paradigm is the PC model where computer makers bought commodity processors and had very little influence on their design.
In mobile devices, the new paradigm has pretty much conquered the middle and high price tiers. Apple, Samsung (OTC:SSNLF), Huawei and soon LG design and market their own processors. In lower priced smartphones, the old paradigm predominates, represented by Qualcomm and other chip vendors such as MediaTek.
But Qualcomm has been able to hang onto a piece of the higher-end smartphone market through its cellular modems. Apple’s going after Qualcomm can be thought of as Apple trying to free itself of this last vestige of the old paradigm.
Technically, Apple’s need to do this was becoming increasingly pressing, since most non-Apple smartphones have the cellular modem incorporated into the system on chip (SoC). Apple needed to incorporate LTE modems into its A series processors, which meant it needed to license patents, especially standard essential patents, (SEPs) from Qualcomm.
Leading up to the outbreak of hostilities, there had been ongoing licensing discussions between Apple and Qualcomm. It was the failure to reach a licensing agreement that set things off. Apple apparently reached the conclusion that Qualcomm was behaving as a “monopolist” and that the withholding of licenses on its terms was a wrongful act on Qualcomm’s part.
To some degree, Apple probably did what Qualcomm alleges, specifically, going around to various competition regulatory agencies and complaining about Qualcomm. Qualcomm however miscalculated. Qualcomm retaliated against Apple by suspending partial reimbursement of royalties paid by Apple’s contract manufacturers (CMs). This led to the Apple suit filed earlier this year to recover the balance of the payments under the reimbursement agreement, as well as all the rest of the legal actions between the two companies.
Apple has subsequently expanded its suit to raise fundamental issues of patent licensing and what it claims are Qualcomm’s wrongful monopolistic business practices. Apple has among other things sought “disgorgement” of all royalties ever paid by its CMs on its behalf and encouraged cessation of royalty payments by the CMs.
Here, Apple committed its own miscalculation. Apple had been focused exclusively on SEPs, for which it felt it was paying too much money, since the royalty was assessed as a percentage of the CMs’ – revenue per device. But the licensing contracts that the CMs had signed with Qualcomm covered a broad portfolio of patents both SEP and not. In encouraging the breaching of the licensing contracts, Apple left itself open to patent infringement lawsuits by Qualcomm for any and all applicable patents in the bundle covered by the CM agreements.
This has provided Qualcomm with its principal vehicle for legal retaliation against Apple. Qualcomm has sued Apple for patent infringement of non-SEPs. In the case of non-SEPs, Qualcomm is entitled to charge whatever the market will bear in licensing fees, so this negated Apple’s complaint that it was being gouged on SEPs. Qualcomm has also used the same patents as the basis for a US International Trade Commission (ITC) complaint that seeks to block the importation of certain iPhone models.
I don’t think there’s any question based on prior actions by the Chinese National Development and Reform Commission (NDRC), the Korean Fair Trade Commission (KFTC), and the Taiwan Fair Trade Commission (TFTC) that some of Qualcomm’s business practices have been abusive and anticompetitive. Qualcomm was able to settle with the NDRC and pursue its business in China with little fundamental change to its business model, however.
Specifically, although there have been complaints about Qualcomm’s bundling of patents, no jurisdiction has been willing to ban the practice outright. Broad portfolio licensing is not uncommon, and there would be really no basis to outlaw the practice. Therefore, it’s likely that the licensing contracts with Apple’s CMs will be found to be valid and enforceable. This leaves Apple severely vulnerable to the legal actions that Qualcomm launched against both Apple and the CMs.
On the broader issue of Qualcomm’s business practices, there seems to be no end to regulatory scrutiny, with actions by the US FTC and the EU yet to be resolved. Most likely, Qualcomm will face further fines and “corrective actions.” However, I don’t expect that the above agencies will be able to negate Qualcomm’s right to license its patent portfolio in “bundles” of SEPs and non-SEPs.
Apple will probably be required to repay royalties withheld by its CMs. This leaves the matter of direct patent licensing still unresolved. Ultimately, Apple needs to reach an agreement with Qualcomm for direct licensing. When the dust from all the litigation finally settles, I expect that a patent licensing agreement between the two companies will emerge, with terms favorable to Qualcomm.
(SA): Digging deeper into the semiconductor space, there’s Nvidia’s meteoric rise. Despite recent tumbles in stock price, is there any stopping Nvidia (stock was up 84% year-to- date as of Nov. 30)?
(MH): This one’s pretty easy. I don’t think there is any stopping Nvidia, as I wrote recently.
Nvidia has managed to do what few other commodity semiconductor companies have done, which is adapt to the new paradigm. Nvidia made a conscious decision to deemphasize commodity OEM sales in favor of providing integrated hardware/software platforms.
This has led to one of its largest sources of growth, machine learning. It was Nvidia that pioneered deep learning on the GPU, and invested billions to develop machine learning APIs under its CUDA development platform. CUDA allows programmers, scientists and engineers to use the Nvidia GPU as a computational engine via standard programming languages such as C.
The development of CUDA has made Nvidia dominant in GPU-based machine learning and in GPU-based supercomputing. Nvidia has become a force in the broader cloud computing market as a result. Every major cloud services provider now offers Nvidia GPUs for AI and accelerated computation.
Nvidia also embraces certain elements of the new paradigm in its ARM-based mobile processors. In addition to creating custom ARM cpu cores, Nvidia can endow its processors with GPU sections based on its latest generation GPU architecture.
Nvidia then goes further by offering these processors bundled with its software to support machine learning and autonomous vehicles. Nvidia has found the perfect market for its ARM processors in the autonomous vehicle market, and numerous manufacturers are using Nvidia hardware in autonomous vehicle systems. These include Toyota (NYSE:TM), Daimler, and of course, Tesla (NASDAQ:TSLA).
The potential markets for machine learning in robotics, cloud, and automotive are rather staggering, and despite Nvidia’s rise this year, I don’t consider it overvalued. Nvidia does face competitive pressure, but I regard Nvidia’s competition as weak.
Despite expectations that ASICs would come to dominate machine learning, that simply hasn’t been the case. Nvidia’s Volta architecture anticipated this challenge and built in special hardware called Tensor Cores to enable ASIC-like performance. Combined with the advantage of being fully programmable, I believe that Nvidia’s GPUs represent the best balance of cost and performance.
(MH): Generally, I avoid companies that I regard as “old paradigm,” which includes both Micron and Intel. Micron’s business is perhaps the most commoditized of any semiconductor manufacturer.
I looked at Micron earlier this year for inclusion into the Rethink Technology Portfolio and ultimately decided against it. Micron failed to make the cut based on a scoring system that I use that looks at five key criteria: Business model, financial strength, price target, innovation, and relative competitive strength.
I realize that Micron’s stock has done well this year, but I don’t regret my decision. The Portfolio is intended for stocks that have high growth potential over a long 5-10 year timeline. The cyclical character of the memory business argued against that long-term growth potential, along with stiff competition from Samsung and SK Hynix.
Intel is kind of my worst old paradigm nightmare. Intel has adapted poorly to the new paradigm, and despite efforts to muscle its way into mobile devices, has fundamentally failed in mobile. Intel inside didn’t turn out to work at all for mobile.
I have almost no faith in Intel’s management. I get the impression they still don’t understand why they failed in mobile. ARM processors seem to have a fundamental advantage in that a physically smaller ARM processor can do the work of a larger Intel processor.
Assuming equivalent manufacturing processes (which is never quite true), it appears that this areal advantage translates into a cost advantage as well as a power efficiency advantage. It was these advantages that allowed ARM to defeat Intel in mobile.
And it is these advantages that will ultimately allow ARM architecture to make inroads in Intel’s other bastions of the PC and the datacenter. Qualcomm is leading the charge in both fronts. It is supplying its ARM processors for Microsoft (NASDAQ:MSFT) Windows 10 “always connected” PCs. Qualcomm has also started selling the Centriq ARM server processor, which also seems to have cost and energy efficiency advantages.
I wouldn’t touch Intel no matter how attractive the valuation.
(SA): You’ve been less enthusiastic about AMD. Your thoughts on this battleground stock as we head into 2018?
(MH): I’ve always been down on AMD, fundamentally because it’s an old paradigm processor maker that has, like Intel, shown very little inclination or ability to adapt to the new paradigm. Unlike Intel, AMD has far less in the way of resources and profitability.
AMD has been rather resourceful in making use of its limited R&D budget, but its limits still show. Ryzen, Threadripper and EPYC are all based on the same 8 core silicon slice. AMD managed a significant improvement by introducing multi-threading with Ryzen, but even then, couldn’t catch up to Intel’s single core performance.
Next year, I expect Intel to introduce processors on its new 10 nm node, and when it does that, AMD is once again going to be non-competitive. I expect Nvidia to introduce new GPUs based on TSMC’s (NYSE:TSM) 10 nm process, with the same result.
AMD is down about 9% YTD, and I consider it all down hill from here.
(SA): What will drive these companies in the new year and beyond? Opportunities in PCs, data center and/or GPU?
(MH): The x86 PC market has been trending down for years, due to the impact of ARM-based mobile devices. I expect that trend to continue and accelerate, as ARM devices become ever more powerful. Apple is leading this trend with its A11 Bionic processor in the latest iPhones, with Nvidia and Qualcomm hot on its heels.
Ultimately, I see ARM architecture taking over for Intel architecture, in the datacenter and in personal computing, which is why I avoid Intel and AMD. Nvidia’s GPUs will continue to perform more of the “heavy lifting” in the datacenter, in supercomputing, and in AI in general.
(SA): Overall, what are the broad trends, opportunities and challenges tech investors should look for in 2018?
(MH): I think the broad trend that I identified several years ago (new paradigm vs. old paradigm) continues to be relevant. I look for companies that are a better fit for the new paradigm. I try to avoid commodity semiconductor companies, unless they have unique differentiating technology, as in the case of Qualcomm.
I consider photonics semiconductors to be an area of opportunity. I’ve branched out this year into photonics companies such as IPG Photonics (NASDAQ:IPGP) and Finisar (FNSR). Generally, I still look for tie-ins to the new paradigm, and Finisar, as an Apple supplier, is a good example of this.
The semiconductor industry is going to continue to consolidate as more and more functions are pulled into the SoC, leaving less market for discrete semiconductors. The new paradigm has driven consolidation and will continue to do so. This is the major challenge for semiconductor company investors.
(SA): Tech has seen a recent decline in sentiment among investors. Nonetheless, tech is up big this year on Wall Street. The SOXX semiconductor ETF is up nearly 42% year-to- date, the XLK tech ETF is up 33%, and the broader Nasdaq QQQ ETF has jumped nearly 32%. Can the momentum continue in 2018?
(MH): Probably not, except for selected companies. I don’t expect the broad semiconductor industry to perform as well as the Rethink Technology Portfolio companies in 2018.
Disclosure: I am/we are long AAPL, NVDA, QCOM, FNSR, IPGP.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.