MADISON, Wis. – Intel Corp. Chief Executive Officer Brian Krzanich went to Washington Wednesday and stood beside President Donald Trump in the Oval Office to announce the company’s $7 billion investment in a semiconductor fab, known as Fab 42, in Arizona.
Fab 42 had stood vacant and unequipped in Chandler, Ariz. since the shell was completed at the end of 2013.
Intel's Fab 42 in Chandler, Arizona (Source: Intel)
"The completion of Fab 42 in three to four years will directly create approximately 3,000 high-tech, high-wage Intel jobs for process engineers, equipment technicians, and facilities-support engineers and technicians who will work at the site,” Intel said in a statement. The 7nm semiconductor manufacturing process will be targeted for Fab 42.
Not for the capacity
Nathan Brookwood, Principal at Insight-64, told EE Times he was surprised by the news. He noted that “Intel is reactivating Fab 42 — probably not for the capacity, but for bringing EUV equipment necessary for 7nm manufacturing.”
Rob Lineback, senior market research analyst at IC Insights, also agrees. He believes Fab 42 is where EUV lithography tools will go. “Availability of EUV lithography tools and processes will help make 7nm successful, but this exposure technology represents a major change.” Noting that Intel is still putting the target three to four years out for Fab 42 to be “completed,” Lineback said, “Intel is rightly so being cautious.”
However, when asked about Intel’s adoption of EUV equipment, an Intel spokesman was less than forthcoming. He told us, “We are still evaluating. We haven’t said at which node we may turn to EUV.”
Driving this announcement, Lineback speculated, is that “Intel needs to tell the industry that it's not falling behind TSMC and Samsung in advanced technology.” He added, “We're still talking 2020-2021, based on the timeframe in the announcement. Other Intel fabs will produce 7nm processors and ICs ahead of 2020, but Fab 42 will be a volume production plant.”
Further, Lineback suspects that Intel will probably use Fab 42 for FPGAs now that it owns the Altera business. "Eventually, IoT ICs (embedded processors and microcontrollers) will move to 7nm technology, but that most likely won't occur for a few more years."
What caused Intel to put Fab 42 on hold?
In January 2013 Intel put on hold Fab 42, a massive fab that President Barack Obama once touted as a symbol of the future of U.S. manufacturing. Intel made that decision then because “they simply didn’t have adequate demand on the market,” Brookwood said.
Fab 42 was originally planned for 14nm. What Intel failed to see then was that Intel's Tick-Tock MPU technology roadmap started to struggle in keeping to a two-year cycle for process generations and maintaining the aggressive pace for Moore's Law, Lineback observed.
Also at that time, Intel did not see the success it had hoped for in smartphone application processors, he added. The company couldn't dislodge the ARM architecture. In addition, the PC market has turned south, just about the same time when 14nm was having its rocky ramp, because of yield issues, Lineback noted.
Even though Intel is doing well in the server market, Brookwood said, “Servers don’t need a lot of silicon.”
Next page: How long had Intel been planning on this?