At the close of the 2/25 trading session, Intel (INTC) and FPGA giant Altera (ALTR) announced that the latter would be leveraging Intel's upcoming 14nm tri-gate process for its future high performance programmable logic parts. While Intel had announced a couple of smaller foundry deals with a few FPGA startups, this is the first evidence that we have that Intel can actually score tier-1 customers with its foundry services.
So, just why is this Altera win a big deal?
Xilinx Could Be In Trouble
The most obvious implication is that Xilinx (XLNX), Altera's long-time rival, is likely to face some serious trouble at the high end of the FPGA spectrum. While the middle to lower end will apparently continue to be serviced by Taiwan Semiconductor (TSM) for both firms, it seems that at the bleeding edge, Altera just got an indisputable advantage.
Altera and Xilinx essentially run a duopoly in the programmable logic space. Sure, there are some smaller players like Lattice Semiconductor (LSCC), but these folks play in the very low cost, low power space in which bleeding edge process technology isn't all that important. At the very high end, these two giants, much like Nvidia (NVDA) and AMD (AMD) in graphics, are pretty much on even footing. They're about the same size and are always pretty competitive with one another because they are both fundamentally restricted by the same process technology.
Altera, however, has significantly disrupted this by signing on with Intel for foundry services. It will now have a 2-4 year lead over its nearest rival, which means that in a given area and thermal envelope, it can pack more transistors, which ultimately means a higher performance, higher ASP product.
Unfortunately for Xilinx, it will not be able to also sign on to Intel's foundry services as the agreement between Altera and Intel stipulates that Altera be the only major FPGA vendor to be allowed access to Intel's 14nm node technology. However, this makes perfect sense for what Intel is aiming for: strategic customers.
Intel Is Getting Paid For Its Transistors
Intel's management has been crystal clear that it does not want to fab for low margins; the company's transistor technology is far ahead of anything that the foundries have (no, "roadmaps" from the foundries claiming that they'll simply magically "catch up" don't count), so it wants to get paid a premium for the privilege.
The kinds of deals that Intel is making are obvious: find companies that don't directly compete with Intel that need an edge and, for the right price, give them the edge that they need. While I suspect that Intel's foundry margin is healthy and above what TSMC charges, I also believe that in the case of a duopoly like Altera/Xilinx, having the better transistor tech will allow the one paying Intel's premium to have significant pricing power over its competitor. If a company is locked out of Intel's fabs, there is literally nowhere else on the planet that can build similarly advanced chips -- talk about a competitive advantage!
Is There More In The Pipeline?
Intel has made it clear that there are more such deals in the pipeline, but that they would be revealed when the customers were ready to do so. If Intel really pushes this (and it looks like it's doing just that), then Intel Custom Foundry could be the secular growth engine that Intel really needs in order to shield it from the ups and downs of the PC market. Yes, Intel is expanding into smartphones and tablets, and yes the Data Center Group is growing nicely, but this could be a high growth, highly profitable division for Intel going forward, especially since it leverages existing infrastructure. We'll have to see how the numbers play out over the next few years, but it's clear to me that this is exciting for the long-term investor.
Why Intel Won't Fab For Competitors
Get it out of your head that there's any chance that Intel will fab for Apple (AAPL), Qualcomm (QCOM), or Nvidia (at least for mobile SoCs). While foundry margin is juicy, the margin obtained from design and foundry is juicier. It seems widely recognized that Intel's process advantage lead ultimately enables a significant competitive advantage in every area of semiconductors that require performance. PC processors, FPGAs, and GPUs, and so on are all clear beneficiaries of a manufacturing advantage. So, why not mobile SoCs?
Doesn't Wall Street understand that once Intel is able to leverage its process technology in these spaces that it will likely have a better product from a performance/watt standpoint than any of its competitors? This means that Intel will start stealing more and more design wins as it will soon be able to offer a better product (measured by performance/watt -- the key mobile metric) than any of its competitors, all of whom are stuck using the same exact fabs with the same abilities/limitations.
Why give that away for mere foundry margin? If Intel can really get its SoC processes materially ahead of its competitors then it will build the better mousetrap (I know this assumes that Intel doesn't botch its designs, but Intel is a world class CPU design house) at a better cost structure than its fabless peers, and ultimately can take significant share in the tablet/phone space and make more money than any of its peers while it's at it.
Intel wants to drive everyone else out of the market, and get into as many Apple and Samsung phones/tablets that it possibly can. The goal with regard to Apple isn't to build Apple's chips, but to convince Apple that Intel's chips are better and that spending all that money designing the A-series of processors just makes no sense. If Intel's SoCs turn out to be materially better than Apple's own homebrew, then Apple will have no choice but to ditch its own and hop on board the "Intel Inside" bandwagon. While I view this as pretty unlikely in the near term, that's the only scenario in which Intel builds phone/tablet chips for Apple.
I'm excited that this could be an opportunity for Intel to really grow its top and bottom lines over the longer term. A steady stream of income from the foundry business could completely supercharge net income/EPS, and ultimately keep the dividend-growth engine alive and well for years to come. That, my friends, is a really attractive prospect at today's 4.40% yield, although capital appreciation on the stock is still likely to be subdued in the near-to-medium term as these deals really won't start to pay off for a while. With the market looking toppy, and with smartphone hype in full force, you'll still have plenty of time to get in (or in my case, add).