During Intel's (INTC) fourth-quarter 2010 conference call CEO Paul Otellini first told us about the Intel 22nm process. He also told us about "Intel's three fabrication manufacturing model." The surprising thing was that he next said that the company felt the 22nm process would be in such high demand that it planned to build a fourth fab center for 22nm. The combination of the process shrink from 32nm to 22nm and the additional fab center would nearly double the potential chip output.
Back then Intel was placing a wager, betting the company could build these revolutionary 22nm Trigate products in very high volume and with a yield that would continue high profitability for Intel. No one knew what the yield would be, and I think that uncertainty is the real reason Intel built the fourth fab. The extra capacity would provide insurance that if yields on the new 22nm process were low, the company could brute force extra wafers through the factories to insure there would be no shortages of Ivy Bridge processors to end customers.
Now we are ending 2012 and the four 22nm fabs are online. According to Mark Bohr's comments during multiple interviews at the recent Intel Developers Forum, "22nm yields are excellent," "22nm yields are better than we expected," and "22nm yields are the same as 32nm." Bohr is an Intel Technical Fellow and Chief Process Technologist. Bohr, like many technical professionals, is genetically incapable of lying or even fibbing. Now, if Bohr were a marketing guy, his comments might be suspect.
It appears that we have a major new fact: 22nm yields turned out to be much better than expected. That means that the fourth fab was unnecessary and will be empty or under-utilized. Ottelini's next step could be to go to Apple (AAPL) and offer to build the A6 chip in 22nm Trigate at the cost (yes, the cost) of the A6 chip now being built on Samsung's 32nm process in Austin.
The A6 chip is about a 97 sq mm chip now; on Intel's 22nm it would drop down to about 55 sq mm, or about a 44% size reduction. The A6 chip built on a 22nm wafer will produce approximately 78% more chips than on the Samsung 32nm wafer. 22nm Trigate wafers cost more to manufacture than 32nm wafers. Nevertheless, Intel's cost to make an A6 chip would be at least 30% lower than Samsung's cost. This means that Intel could sell its version of the A6 chip to Apple at or near Samsung's cost and still make 30% gross margin and fill that fourth fab center in the process.
Samsung, which has about 50% of the smartphone SoC business (Apple's A6 and Samsung's own requirements), would suddenly lose 50% of its SoC business. Intel would go from "zero to hero" in one stroke.
When the industry recovered from the shock of this single action, the pundits will understand very clearly that one node, let alone two nodes (14nm next year), is such a smothering advantage for Intel that the fabless semiconductor model (at least for these very high volume, leading edge chips) can't compete against an aroused Intel.
During 2013-14 Intel would have such an overpowering advantage in new SoC design wins over Qualcomm (QCOM), Nvidia (NVDA), and even Samsung that it would go on to ultimately dominate the mobile SoC business. Other than the Apple chip, these SoCs will be Intel Atom-based, because in the real world customers don't care anymore about processor architecture as long as the chip does what they want at very low power and at a low price. That means ARM (ARMH) gets nicked in the long run as well.
While all this is going on, and since the cost of Ivy Bridge processors is lower than expected, Intel will likely cut prices on selected devices in order to put more pressure on AMD (AMD). That will allow Intel to pick up the 20% of the PC CPU business still at AMD.
A great deal of the current Nvidia discrete GPU business will just quietly migrate into the Ivy Bridge chip with "good enough" graphics, leaving Nvidia in a very precarious position. The next Intel CPU architecture, due to ship in late 2013, with even better graphics could well kill off AMD and Nvidia.
All this happens because Intel's 22 nm yields exceed expectations. It's a crazy business!
Action: Buy Intel or Intel LEAPs and watch the story unfold. Sell or short AMD, Nvidia, ARMH, and QCOM, since there is no reason to own them.
Disclosure: I am long INTC.