As an Intel (INTC) long that has put a significant amount of effort into understanding every last detail of the investment thesis, I'm aware of where Intel is strong and where the company needs some serious work. As far as CPUs go, Intel is first class. I expect that Intel will likely offer the best performance/watt from Bay Trail onward. As far as GPUs go, it remains to be seen how competitive Bay Trail will be, although I expect that it will be pretty competitive, at least from recent demonstrations at Computex.
That being said, Intel can do the compute/SoC side of things pretty well -- there's very little doubt of that following the release of Haswell, which brought tablet-like battery life to ultra thin and light notebooks. I expect that in tablets, Intel will be just fine and will take a solid amount of market segment share over the next few generations. But while I'm not too worried about tablets, what investors need to keep an eye on going forward is progress in the modem/RF space.
Smartphones Aren't Just About Compute
Smartphones are quickly evolving into fully-fledged computing devices in their own right. These devices certainly require high performance in low power envelopes (which Intel's manufacturing lead + architectural expertise is certainly well suited to), but there's more to the story than that. In addition to having an entirely different business model (smartphones are sold through and are subsidized by carriers), the incremental capabilities required are quite different.
In particular, smartphones require cellular connectivity, which in turn requires some quite sophisticated RF and cellular baseband technology. This is an area in which Qualcomm (QCOM) is currently the unequivocal market leader. While the rest of the world, including Intel, is struggling to ship its first multimode LTE voice/data modem, Qualcomm is now shipping its first LTE-Advanced design bolted right onto its Snapdragon 800 applications processor. It's definitely going to take a lot more than what we've seen so far on the comms space if Intel is going to have a real shot at gaining non-trivial smartphone market segment share.
Intel's LTE Roadmap Problem
At Intel's 2012 Investor Meeting, Mike Bell and Hermann Eul showed the following slide to investors and analysts, which claimed that the LTE-capable XMM 7160 modem would come paired with the company's Clover Trail+ SoC in 2013, and then Merrifield (new SoC based on the "Silvermont" architecture) would come with the XMM 7260 just a year later:
Well, that unfortunately didn't happen -- the XMM 7160 will actually ship with the Merrifield platform in Q4, for device launch sometime near Mobile World Congress 2014. Clover Trail+ shipped with a slightly updated version of Intel's XMM 6260 3G modem, and as a result failed to win any US-based designs (these require LTE). This suggests that something went very seriously wrong with the company's cellular modem roadmap. Now, what's interesting is that the company's "highly integrated" platform -- known as "6331" -- is likely to be its first integrated baseband + apps processor chip on the market. The company has largely gone silent on this particular part, although Intel VP Dadi Perlmutter claims that an integrated apps processor + modem will be coming during 2014:
Intel in 2014 will introduce a mobile chip that integrates LTE on the same piece of silicon as the application processor, Perlmutter added. The company has been gaining some traction in smartphones, but volumes remain low. Perlmutter expects higher volumes in 2014.
While the Infineon acquisition brought much needed 2G and 3G expertise, Intel has continued to expand its modem/RF teams as leadership in this space could not simply be bought for $1.4B (the price Intel paid for Infineon Wireless). Intel recently set up a modem development group in San Diego, and has been aggressively recruiting talent. Further, Intel recently confirmed that it acquired Fujitsu Wireless, a team well known for developing best-in-class LTE and LTE-Advanced RF front ends. This doesn't buy Intel a ready made LTE-Advanced cellular baseband, but it helps to bolster Intel's RF talent going forward.
It is my view that while Merrifield with XMM 7160 will allow the company to gain significantly more traction in phones than it currently sees today, I would hope that Intel will be able to roll out an LTE-Advanced solution fairly quickly following the LTE solution, although color will not likely be available until the Investor Meeting.
Is There Hope?
Anandtech recently ran a video interview series with Aicha Evans, GM of Wireless Platform Research and Development Group. This interview series is a treasure trove for anybody wanting to gain deeper insight into Intel's wireless group and its plans/strategy. Evans offered this interesting tidbit with respect to Infineon's LTE development:
So, they [Infineon Wireless Group] finally on 3G/2G did extremely well, but when it came to that LTE transition, you know, they had their challenges. So, this was a marriage made in heaven because of that. So, when they came in, we had to accelerate LTE and we had to get it done.
So, where have we been for the last two to three years? The reason we can show off at Computex with a demo, the reason we could go into some benchmarking in static(?) mode, was because we were furiously working on LTE, and also on keeping our customer base. We do serve a lot of customers currently in the 2G/3G space. And, also, preparing the next generation because the thing with wireless is winning one generation, one battle is one thing. But being in the game, battle after battle after battle is important. So, we're working on the current generation to ship it, working on the next generation, and the one after that. So we're working on three pipeline generations at the same time.
Now, Mr. Shimpi dug a little bit deeper:
Anand Shimpi: You left, effectively, the high end to other folks. Which, in my mind is very atypical of Intel, because you don't see that on the CPU side.
Aicha Evans:[...] so, yeah, they are going through a transition from fast follower to leader.
Now, while the first part of my article probably had you feeling significantly less optimistic about Intel's chances as a modem vendor, I found this interview (and again, I encourage everybody to go watch it) to validate -- at least in my mind -- that Intel is doing its best to really become a leader in this space. It certainly is no coincidence that following the closure of the Infineon Wireless deal in early 2011 that Intel's R&D spend has been going through the roof:
I find it difficult to believe that this incremental spending has been simply due to beefing up the system-on-chip and Atom processor core teams. In fact, a quick look at the CFO presentation of the 2012 investor meeting gives away that a good $2B of Intel's R&D is funneled into the "Other IA" segment, which includes the modem teams:
(click to enlarge)
The bottom line is that Intel will eventually get the hang of the modem/RF side of things. Things are improving, for sure, and the XMM 7160 + Merrifield platform will definitely allow Intel -- for the first time -- to gain meaningful share in the smartphone space. On the SoC integration, low power CPU, low power GPU, image signal processing, etc., the company is most certainly on the right track and I have no doubt that Bay Trail, which is a tablet SoC slated for launch later this year, will finally show Intel's capabilities in these areas.
The Investment Case: It's A Long Game
The truth of the matter is that major new products in this space take time, and things don't change overnight. While there is significant doom and gloom with respect to Intel's financial position and bottom line, the reason that net income has taken a hit is that it costs money to enter these markets. This means hiring more engineers, training more sales reps, taping out more chips, developing the low power version of the semiconductor process faster, and so on. It takes investment.
Intel could certainly, in light of stagnant revenues, simply choose to "downsize." Would this help the bottom line significantly? You bet. Would Intel be fundamentally doomed? Of course. In technologically oriented industries, R&D is key. When the going gets rough and your core business starts to erode, you -- as quickly as possible -- use every last dime that this business generates that you need to in order to right the ship. Think about it, folks -- Intel's "bad" 2012 involved generating ~$11B in net income. Now, what would a long-term investor prefer that Intel have done? Lay off half of the workers so that in 2013 net income would look better? NO! Invest, invest, invest!
The dividend is nowhere close to being in danger (4% yield) and the company is on the cusp of finally having competitive products to fight with in these new markets. The substantial increases in R&D will, very soon, finally begin to pay off and then both the top and bottom lines will start to be on the upswing once again.
But the question is, are you a longer-term investor, or are you a day trader? Neither one is inherently "good" or "bad," but the type of investor that you classify yourself as will determine whether Intel is worth buying here or not. If you buy the long-term story, then you are at least buying a company that is investing every last penny that it needs to in order to be a long-term "winner" in this new space. It's not going to be easy, but if Intel weren't spending this kind of money to do it, then Intel's days would truly be numbered.
If you're a day trader? Stay away on the long side. Sentiment right now is ugly, and it's not going to get better until the business starts "wow'ing" the Street again. This means revenue beats, EPS beats, gross margin momentum, and every talking head on CNBC claiming that the stock is the best mobile play since sliced bread. You won't catch the bottom once the momentum does reverse, but you're just here to make some money trading it, right?
While I am really disappointed that the Infineon group seems to be having much more trouble than expected, this kind of thing tends to happen. The only reason Intel was able to buy Infineon Wireless for pocket change of $1.4B was that it was truly in deep trouble. Intel is pumping it full of capital and giving it the resources that it needs to succeed, but these things take time. I'm pretty confident that if Intel gets its mobile ducks in a row, we'll see ~$50/share by 2016. If it can't, then I suppose TSMC (TSM) will find itself with a new pure-play foundry competitor.