There have been many good articles written post Intel's Analyst Meeting covering the possibilities of the company with its new products and its leading edge process technology. I have followed Intel (NASDAQ:INTC) for many years and remain of the opinion that when it can synchronize the market dynamics with its Fab process technology, it ends up winning big. When things are out of synch, it struggles and is doing so now with mobile because the very valuable high-end piece is out of its control, driven by Apple (NASDAQ:AAPL), who happens to want to own the apps processor. As a result, I believe there is a much more profitable way Intel can win in mobile by opening up its Fabs to every fabless vendor and thereby absorb the profitable part of the industry without investing heavily in chip design R&D and having to time the market perfectly with spot on features. I will use the term "Richer Mix" Foundry model to describe this segment of business that is available to the company if it can execute.
The selection of Brian Krzanich as Intel's new CEO was an interesting one in that he rose suddenly and after Andy Bryant was appointed Chairman. Bryant was CFO for many years and now is the man with the most executive seniority at the company. For many years, he oversaw the Fab footprint which meant making decisions on what to invest, how much and most importantly when. He knows what Wall St. is looking for in order for the stock to move and that means expanding the number of leading edge chips that run on the company's process technology in a larger number of Fabs. Unlike several years ago, mobile as represented by Apple, Samsung (OTC:SSNLF), Qualcomm (NASDAQ:QCOM), Nvidia (NASDAQ:NVDA), Broadcom (NASDAQ:BRCM) etc. see value in reaching for the front end of the semiconductor process curve.
Intel today is very profitable but with flat growth from its traditional x86 business (Server and Clients). With an ongoing annual capex budget of roughly $11B, the company is continually questioned on its size and how it can be leveraged better for growth. The fear is that the equipment will be underutilized like it was in the summer of 2012 when Ivy Bridge only required half the planned capacity (partially this was due to high yields). Contrast that with Qualcomm, in the same summer, unable to meet demand in its foundry and you have a picture of what could be possible.
Intel reached peak profitability (NYSE:ROI) in the 1990s. The PC industry and Server markets then as now moved in lockstep with their product introductions. Dell, HP (NYSE:HPQ), Compaq, IBM had to be at the introduction on day one or lose business to one of its competitors and thus allowing Intel a great deal of flexibility in not just managing fab utilization but also being able to sway the market towards higher performing processors that sold for hundreds of dollars more than lower MHz versions. This was especially true when PCs were playing catch up to the heavily weighted Microsoft (NASDAQ:MSFT) O/S or when allocations hit the industry. This is known as selling a "Richer Mix" and Wall St. cheers when they hear this on an earnings call.
For those of you who bought a car in the early 1980s, you may remember that our government placed an import restriction on Japanese cars in order to save our auto industry. The Japanese complied with the unit count but then began shipping in cars that were loaded with features that drove up the price. They were selling a "richer mix," which is much more profitable. Worried that the US would impose stricter import restrictions, Honda and the others started opening up plants in the US.
Intel's Xeon and Haswell line are strong revenue and profit producers. They are like Boardwalk and Park Place with many repeat visitors. If they were a separate fabless company, without the concern of fab utilization and mobile inroads, then the valuation would likely be much closer to a market P/E. The Data Center business is growing at 15% a year according to Intel and has a 50% operating profit. It is offsetting revenue declines in the Client area, which as a business is more insulated from AMD and Nvidia than at any time since the 1990s. The fact is that the Ultrabook market with its very small form factors, batteries and cost constraints is well suited to having an Intel Haswell chip but not the secondary graphics chip which just adds cost, takes up room and consumes power.
The mobile market led by Apple and Samsung is much more vertical than the PC market and the utilization of outside suppliers rest with whether or not it exists internally or if it provides a value beyond what can be found elsewhere. Many have the assumption that an advanced Atom chip will be so desirable that Apple must adopt it. This overlooks the fact that Apple is customizing its processors to work tightly with its ARM based iOS and has started branding around its chips. However, Apple would, I believe, always be open to building its chips at Intel for the right price/performance and price/power. Moreover, if Intel is 2, 3 or 4 years ahead of the industry in process technology, then it should be able to sell its services at a premium, especially if Tri-Gate offers a wider range of performance and power tradeoffs that let the same processor be used in a wider range of physically sized platforms.
Some may ask why Qualcomm is in Apple's iPhones and iPads. Simply put, the baseband function is the high dollar silicon content enabling LTE and all its protocol flavors. In fact Qualcomm's release of its chips the past several years have been in synch with Apple product launches. As a very heavy power user, Qualcomm could benefit from Intel's Tri-Gate process but a partnership is on hold until it is determined if Intel will remain a mobile player.
Opening up the Foundry to all fabless chip vendors would allow Intel to set up the "Richer Mix" model on a process technology basis as customers would see the benefit of being half the die size or less and running at higher performance and lower in power. The margin stack formula would shift from today's model where someone like Nvidia receives 50-55% GM on top of TSMC's (NYSE:TSM) 45-50% to one with Intel possibly getting well over 50% while Nvidia shrinks below 50%. If Nvidia thought it would get a first mover advantage or if it thought it needed to be in Intel's fab to stay at least equal to its competitors, then it would sign up. The same would be true of others even though Intel would likely limit its partners at first.
It is interesting that after all the investment in outstanding chip development and marketing, Nvidia's operating margin, as of last quarter, is only 13.5%. It varies across the board with other fabless vendors but it is quite possible that a strong Foundry capability open to many chip vendors would allow Intel to take a greater slice of the margin stack to the point the fabless customers would see their margins compress like the PC OEM vendors and then you know Intel has maximized its ROI for its Capex investment.
When Paul Otellini left Intel he gave a long interview with the Atlantic Magazine regarding his tenure at Intel. I happen to think that when he took over as CEO he had two goals. The first was to rebuild the server business after letting AMD beat it to the market with the first 64 bit x86 processor and the second was to push AMD and Nvidia out of the PC market by absorbing the graphics unit and its associated dollars. Based on Intel's market share and profitability, it appears that Intel has both those goals well in hand, as Intel today dominates servers and is in the process of cannibalizing AMD and Nvidia thanks to the market shift to Ultrabooks, which the two have no effective competitive solution.
In Otellini's interview, though, he was asked about Apple. Otellini stated that Intel had a shot at building Apple's chip but the profit was too low. Relative to its existing products, this is no doubt true. It is also likely that an internal group at Intel that was charged with an analysis that showed Intel could build something better and when necessary could launch into the market with something that would repeat the success it has in the PC market with the same business model.
As the new CEO in a company like Intel whose products take years to develop and put in production, Krzanich is compelled to allow the Atom group to continue their work and prove to the company that they were viable, being able to win market share and become profitable. It's a tall order as this year the Atom group is on track to lose $2.4B on $4B in sales. If Krzanich is forward looking, then my guess is he is working on a Foundry transformation that will be unveiled in the next 12 months depending on how Atom performs. Without Atom, the Foundry is available to nearly all and if Atom remains, the Foundry opportunities will be limited.
At the analyst meeting, the most interesting highlight for me was the comparison of Altera's (NASDAQ:ALTR) 14nm FPGA die with Xilinx's (NASDAQ:XLNX) TSMC built versions. It is a very powerful visual representation of what is possible and a show piece for the Foundry business. The question now is: Will the Foundry be unleashed so that Intel can Deliver on the "Rich Mix" business model to the rest of the mobile universe?