For 18 months Intel has dabbled in making 22nm chips for small companies such as Achronix, Tablua, and Netronome. Intel said its goals were to get some outside feedback on its process technologies and experience offering foundry services, but under former CEO Paul Otellini foundry was a limited offering.
Late last year, rumors emerged Intel struck a billion-dollar deal to make 22nm ASICs for Cisco -- neither side confirmed them. Others said Intel was trying to win away from Samsung the business of making Apple's A-series SoCs. More silence.
The first clear sign of a shift came in February when Altera announced it struck a deal to have high-end FPGAs made in Intel's 14nm process. In October Altera said those parts would be its Stratix 10 chips using four ARM Cortex A-53 cores. Not exactly competing, high-volume parts but a big step in that direction.
There's a crossroads ahead for Intel. Krzanich clearly wants the company to find a killer smartphone/tablet design it can make at a cost advantage at 14 and 10nm. If it cannot create such a killer design he may have to figure out whether he can pivot to becoming a killer foundry for an Apple or Qualcomm.
That would indeed be what former boss Andy Grove would call an inflection point.
— Rick Merritt, Silicon Valley Bureau Chief, EE Times