Did you hear? Intel (Nasdaq: INTC) has opened shop as a foundry service, selling its bleeding-edge manufacturing processes to third-party customers.
Before you exclaim that Intel has gone insane, let me calm your nerves. The chipmaker doesn't plan to become an all-around foundry giant in the image of Taiwan Semiconductor Manufacturing (NYSE: TSM) or United Microelectronics (NYSE: UMC). Direct competitors Advanced Micro Devices (NYSE: AMD) and NVIDIA (Nasdaq: NVDA) will never be able to close the gap to Intel's advanced manufacturing processes by simply buying production capacity from the chip giant. Nor is the company producing mobile powerhouse processors on command from Marvell Technology (Nasdaq: MRVL) or Qualcomm (Nasdaq: QCOM). Intel's Atom chip wants to compete with these products, and it would be suicidal to let the foxes in the henhouse.
No, the only client allowed to stamp processors out of Intel's specially processed 22-nanometer silicon wafers is a privately held designer of Field Programmable Gate Arrays (FPGA) named Achronix. FPGA chips are known to be slow and power-hungry, but make up for these deficiencies by being reprogrammable after they leave the production line. You'll find these chips in the design stage of more ambitious microprocessors, specialized cryptography machinery, or super-secret military applications. It's a relatively low-volume product category that poses no threat to Intel's speed demons, and Intel expects to use less than 1% of its next-generation manufacturing muscle on Achronix's behalf.
If an audacious plan to convert Intel into the next Taiwan Semiconductor is the World Series, this partnership is minor-league batting practice. Not that cloning TSMC would be such a bad idea -- the leading foundry produces fatter operating cash flow margins than Intel today, but then sinks most of it into factory upgrades and dividends. No, Intel is simply sticking a timid toe into unfamiliar waters to check the temperature. The company gets an opportunity to try out its next manufacturing process in production volumes, without tinkering with its own chip designs in the process. Intel is even asking its devotees for input on the whole foundry idea.
This is an early test of a new idea, immaterial to earnings and virtually risk-free. At worst, Achronix goes elsewhere after a failed experiment. At best, Intel shows off the flexibility of its manufacturing capabilities to an adoring semiconductor industry, and perhaps signs up a very select list of non-competing clients. In short, this is nothing more than Intel showing off.
