Published in PC Hardware

Intel fabs to churn out ARM parts

by on30 October 2013

Altera blasphemy

The first ARM chips build in Intel fabs will roll out next year. Altera has officially confirmed that it will fabricate its upcoming 64-bit ARMv8 chips using Intel's FinFET 14nm process, starting sometime next year.

The announcement was made at the ARM developers ‘s conference in San Francisco and it has many industry watchers scratching their heads. Intel still doesn’t have mobile parts that can take on ARM designs in terms of price or power efficiency, although performance is no longer an issue. It also lacks integrated LTE, rendering its otherwise competitive x86 SoCs unappealing in the smartphone space.

Bay Trail-T is out and it is basically the best low-end x86 tablet part money can buy right now. However, although it can beat quad-core ARM A15 chips, it should be noted that it’s a 22nm chip, while the competition is still at 28nm. Even then it leverages its superior manufacturing, Intel is barely able to compete.

By the time the ARM alliance gets to 20nm next year, Intel will roll out its first 14nm mobile chips. The prospect of ARM SoCs fabricated on Intel 14nm silicon is interesting and could be very disruptive. However, it is highly unlikely that Intel would undermine its own mobile push by allowing direct competitors to build consumer chips on its 14nm FinFET process.

Altera’s chip is not a consumer part, it’s an A53-based quad core for embedded applications. In other words, it’s not much of a threat for Intel’s own offerings. It is highly unlikely that Intel would allow the likes Qualcomm or Samsung to build 14nm FinFET consumer parts in its own fabs.

Like we said, the only thing keeping Intel’s mobile parts competitive at this point is the 22nm process and this will be especially true of upcoming Airmont parts. By the time these 14nm products roll out, high-end ARM-based SoCs will still be stuck at 20nm, allowing Intel to maintain a technological edge and keep its mobile parts competitive.

Giving up such a strategic advantage for a few quarters of fab revenue from sweetheart deals with Apple or Samsung is simply not an option. It could happen only if Intel decides to pull the plug on its own x86 mobile business and there’s no indication that this is even on the table. 

Rate this item
(0 votes)

Read more about: