Intel (INTC) is up 2%+ today on its presentations at its annual analyst day. While any investors or potential investors should certainly review the slides and listen to the webcast on their own, this article will give you a quick overview of the important business items discussed. Still to come is financial commentary from CFO Stacy Smith, who's basically my favorite executive ever and definitely worth the listen. Nonetheless, what's been discussed so far is very important.
While I'd love to provide a very detailed analysis of each of the key items, my day job precludes me from having the time to do so. Fortunately for all interested, SA's superstar tech writer Ashraf Eassa is planning a series of deep-dives into topics ranging from process tech to Intel's mobile efforts. Keep an eye out for those. For now, the top-level summary: Intel has its act together.
The speakers took no time at all to throw former CEO Paul Otellini under the bus, admitting within the first few minutes that Intel had lost its way and that change was needed. While I don't fully believe that (Otellini did a pretty good job and oversaw a period of strong growth at Intel), it's what the market wanted to hear.
And that's kind of an interesting point about the whole theme of this presentation: as Ashraf and I listened, we noted a decreased focus on technology, and an increased focus on telling the market what it wanted to hear. This is good for the share price. They even had slides and a real-time video encoding presentation demonstrating why Bay Trail SoCs won't cannibalize Core sales.
Normally I'd be concerned about management of a tech company pandering to analysts rather than focusing on their technology and business, but this is Intel we're talking about - they have the stuff to back it up. Without further ado, here are some of the highlights that investors should be cognizant of.
Overarching Theme: If It Computes, It Does It Best With Intel
This was a point that CEO Brian Krzanich hammered home on numerous occasions. Intel is seen as a CPU company, but it doesn't see itself that way. From the Internet of Things to networking to cloud-RAN to HPC, if it crunches numbers, Intel wants a piece of it.
Foundry Gets Reprioritized
Intel Custom Foundry is a thing that a lot of people have talked about, but it hasn't shown up on the bottom line to a material degree. There are a couple reasons for this. First, it takes a while to ink these deals and bring chips into production. Second and more importantly, Intel has a vested interest in not cannibalizing its own business. As they referenced during Q&A, they'd rather get paid twice for something than once. The reason Intel's margins are so high is that they're an IDM; they capture not only the foundry margin, but also the design margin. When they act as a custom foundry, they only capture one of those halves. Consequently, they pick and choose what they'll fabricate, so as not to encroach on their own chips territory.
Intel management, however, is now placing a greater emphasis on ICF in their slides and discussion. Their view as expressed during Q&A: "if we can make good money... we'll do it." And in one of the early presentations, they had a slide that very specifically mentioned smartphone SoCs under the scope of foundry.
I wish I could show you a picture - I tried to grab a screenshot, but didn't succeed - and the slides are not yet uploaded to Intel's website. However, once they are, check them out. I found this note very curious. Why? During Q&A, they didn't sound enthused about fabbing SoCs for fabless competitors in areas that they play in, one because they are confident x86 designs are vastly superior, and two because they don't want to cannibalize their own sales. Who does that leave?
Apple. Ashraf and I both agree that this theory - which has been delineated many times by SA contributor Russ Fischer - has now gained more plausibility. We'll admit that we were both skeptics, and I continue to be unsure that Intel fabbing for Apple will ever happen. However, we do know that if Intel were to serve as a foundry for smartphone SoCs - which their slides say they may - it would likely only be for a customer who would never use Broxton or Silvermont-based designs - and that leaves Apple. Again, it's speculative, but Russ may really have a good point.
Data Center Still Rocking
I've previously discussed why mobile doesn't matter and DCG is the real story at Intel. Thus, I was most interested in Diane Bryant's presentation. She didn't disappoint, reaffirming 15% CAGR for DCG through 2015. This will be driven by about an 8% CAGR in Enterprise IT and a much higher, much faster rate of growth in emerging areas like private cloud, telecommunications, and HPC. She had some interesting comments about the correlation between Enterprise IT growth and GDP growth, demonstrating that the recent slowdown was temporary.
Besides the growth, there were other interesting tidbits: for example, Intel is offering custom silicon to major customers, which is a direct rebuttal to the numerous charges that Intel's model is antiquated and semicustom is where it's at in semi-world. The example Diane Bryant gave was some customization Intel did for Facebook's cold-storage requirements - these are photos that people post to the tune of millions per day, that are rarely looked at after that but have to be kept around anyway. Doesn't need a lot of compute power; needs to be low power. Intel built a customized solution.
Beyond that: Intel's #1 in datacenter SSDs, and is serious about ethernet and silicon photonics. They're expanding their share in networking.
And finally, Intel threw a lot of Antarctic water on the notion of ARM-in-servers. Diane Bryant implied that the TAM of microservers is 10% of the total server market, and noted that the TAM of ARM-type microservers is more like 3%. (Note that there are Xeon microservers. The latter category refers to Atom.) Considering that Intel has credible, advanced parts shipping while some of the more-hyped "competitors" (who I will leave unnamed) are still in the vaporware stage, the future looks very bleak for the ARMy's much-vaunted entree into microservers.
Intel's slides had some comments from EETimes discussing why ARM-in-servers is a lot of hype and not a lot of reality. As Diane Bryant explained, one of the major problems is that heterogeneous architectures in the data center necessitate different middleware stacks, dev tools, etc. Having one architecture for all workloads - whether light or heavy - minimizes TCO. Fundamentally, this is a major negative for ARM-in-servers, even in a scenario where they were able to offer significant perf/watt superiority. Since that will likely not happen anytime in the near future, it's a moot point.
Reports Of The PC's Demise Have Been Greatly Exaggerated
The PC market is stabilizing. While the overall market will likely continue to decline by low single digits, Intel is looking to take share in the "value" segment, introducing devices that will sell for very low prices (sub-$300). As they explained, they'd basically been ignoring that space since the netbook, choosing to focus on higher-end systems. Well, their shiny new Atom chips change that all up. They're planning to "gain market share against traditional x86 competition" or in other words, AMD. They demonstrated that their share of Chromebooks has accelerated rapidly.
There were some pretty neat PC designs shown off, including some 2-in-1s that will go as low as $100. Take a look at the pictures/video when they become available.
Mobile: 4x The Fun
Intel explained it this way: in '13, they were building a footprint and signing long-term agreements with OEMs. In '14, they want growth, in a very big way. Their plan is to quadruple tablet shipments next year - to the tune of 40M. Let's see if they do it. I think they have a good shot at doing so.
In other news, they demonstrated a working 64-bit Android device, and are making progress on LTE, including carrier aggregation. This is a topic that requires a lot of technical discussion that I don't have time to lay out, so look for Ashraf's articles on these topics later.
A Quick Note On 14nm And Beyond
14nm yields are rapidly improving and we're still looking at the same timeframe. More interesting was the fact that 14nm and 10nm, instead of showing a "slowing" Moore's Law, actually improve the cost per transistor relative to the straight line that Moore's Law would predict. Intel has 10nm test chips running. Finally, they're also focusing more on density, a topic Ashraf has explored previously. This should continue to build their lead over the entire rest of the semi industry.
Intel is one of my largest positions, and I'm even more bullish after hearing what management had to say this morning. It's hard not to see analysts reevaluating their stances and target prices after this. Intel reaffirmed that PCs are stabilizing, DCG is growing, and mobile is exploding. Their valuation is not demanding by any stretch. It adds up to a very compelling opportunity.
Investors should look out for Stacy Smith's financial commentary later today and Ashraf's deep-dive analysis over the next week or so.
Disclaimer: This is solely my opinion, not an investment recommendation or solicitation, and may not represent the views of my employer(s), associates, or other related parties. No guarantees made to accuracy or completeness. I am long the companies mentioned in the disclosure and may change my position at any time without notification. Please see the full disclaimer in my profile, and do your own due diligence before making any investment.