How Intel Lost the Mobile Market, Part 2: The Rise and Neglect of Atom

Update (6/1/2020): The post below may have been created in 2016, but it nevertheless stands up as a postmortem of what went completely wrong with Intel’s cellular efforts — with a person really critical omission. Again in 2016, we did not know Qualcomm had been ruthlessly imposing licensing and getting terms that made it proficiently impossible for suppliers to offer Intel-based cellular equipment. I don’t forget thinking why Intel couldn’t uncover a one US firm to deliver a telephone all-around its hardware system for appreciate or funds when the authentic Xolo X900 when compared properly more than enough against a then-recent Apple iphone.

Intel nevertheless made a amount of faults with Atom, as this post discusses, but the actuality that Qualcomm had a stranglehold on the market driving the scenes definitely had an impression on what form of success Intel was at any time likely to reach.

I genuinely favored the Xolo X900 machine I analyzed all those people decades back, and the Bay Trail tablets I had circa 2013 were being fantastic equipment. Atom’s cellular efforts will usually remain an enticing may possibly-have-been.

Initial story below.

In Part 1 of this two-section collection, we mentioned the variation between Intel and TSMC’s foundry designs and how these variations made it extremely tough for Intel to compete in cellular. In Portion 2 we’ll check out the precise conclusions Intel made, the rise and neglect of Atom, and why the company’s top-quality foundry technology was not more than enough to conquer the market.

The typical rationalization for why Intel shed the cellular market is that its x86 cellular processors either drew also significantly energy or weren’t effective more than enough when compared with their ARM counterparts. Intel’s decision to promote its ARM division and XScale processor line in 2006 has been greatly derided as a crucial mistake. It is a basic, typical-feeling rationalization with just a person flaw: It faults signs for lead to.

All of This Has Occurred In advance of

Intel’s struggles in the cellular market did not start with Medfield, Moorestown, or even the decision to promote its ARM business and XScale chip division 10 decades back. As EETimes reported in 2006:

Intel (Santa Clara, Calif.) invested extra than $10 billion to enter the communications business above the decades, but the microprocessor large shed its shirt — if not millions of dollars in the arena. The reported communications-chip sale is mentioned to be section of Intel’s system to overhaul the firm. Intel is also established to contain the layoff or redeployment of 16,000 workers, according to speculation from a person Internet web page.

Transform “communications” to cellular, change the amount of fired workers, and that paragraph could’ve been created nowadays. Intel’s complications in cellular aren’t new Santa Clara has been having difficulties to enter new markets for virtually 20 decades. Other content articles from 2006 emphasize that XScale sales had been quite low, as had revenue from Intel’s networking and communications division.

From Intel’s viewpoint, advertising XScale made feeling. Setting up a cellular processor business all-around ARM cores would have minimal Intel’s potential to leverage its very own IP and experience in x86 manufacturing, even though simultaneously slicing into its gains (Intel would have owed sizeable royalties to ARM if this kind of a style at any time became well-liked). Atom was previously properly into improvement in 2006 and Intel decided to bet on its very own hardware experience and program improvement competencies.

Atom and the Rise of x86 Everywhere you go

Contrary to well-liked belief, Intel was not caught fully off-guard by the rise of smartphones or the popularity of compact, World-wide-web-linked equipment. Atom improvement commenced in 2004 the Silverthorne main that Intel debuted in 2008 had a TDP of just 2-3W at a time when most cellular Main 2 Duo processors were being caught in 35W territory.

Intel-Penny

Intel’s authentic Atom.

Most people today don’t forget Atom as the chip that introduced a thousand netbooks, but that was not Intel’s authentic system. The firm believed nettops and netbooks would be a area of interest market for Atom, not the chip’s most important system. Atom and its successors were being meant to launch an armada of Cellular World-wide-web Gadgets, regarded as MIDs.

Gigabyte-MID

Gadgets like the Gigabyte MID M528, proven earlier mentioned, glance hopelessly quaint nowadays, but Intel was clearly thinking about the potential of cellular computing. The firm envisioned an ecosystem of netbooks and MIDs pushed by its very own customized x86 architecture, a intention the push dubbed “x86 just about everywhere.”

Conflicting Priorities

There were being clearly executives at Intel who recognized how crucial cellular would be to the company’s lengthy-expression potential and pushed for aggressive positioning and merchandise ramps. Regretably, those people efforts were being stymied by many others who were being anxious about the impression Atom and the low-cost equipment it was meant to allow would have on Intel’s most important business. MIDs and later on netbooks were being meant to be bare-bones, low-cost equipment, valuable as secondary devices and for basic jobs, but no extra.

IntelSoC

Intel was doing work on Atom SoCs in 2008, but its to start with totally unified chip would not ship right up until 2012.

Intel’s put up-launch attitude towards Atom is finest summarized as benign neglect. While the chip went by means of numerous revisions to integrate factors and cut down expenses, Intel refused to dedicate the assets that would have made Atom a finest-in-class player in the cellular market. From 2008 to 2013, Intel introduced a cost-reduced model of its Nehalem architecture, the Westmere 32nm die shrink, a new architecture with built-in graphics (Sandy Bridge), a significant-conclusion enthusiast system (Sandy Bridge-E), a new 22nm CPU with FinFET technology (Ivy Bridge), a further architectural refresh (Haswell), and a second-technology enthusiast system (Ivy Bridge-E). That is two comprehensive tick-tock cadences for Intel’s large-main business, even though Atom did not even make the soar to 32nm right up until 2012. Its one architectural refresh to date arrived in 2013, just soon after the launch of Ivy Bridge-E.

Despite currently being at first starved for assets, 32nm Atom chips were being competitive in the midrange cellular market. With Medfield, Intel seemed to have turned a corner, but the company’s models commonly failed to uncover significantly traction in the market. Only Intel’s contra-revenue method gained the firm sizeable pill market share, and those people gains were being only sustained by means of heavy economic losses.

intel-mobile-chart

Intel’s cellular and communications revenue and losses from Q1 2013 by means of Q2 2014.

Atom was not the challenge — Atom was the resolution Intel did not have the guts to chase.

The Rough Selections Intel Didn’t Make

Intel failed to obtain traction in cellular for the reason that it was not eager to chance upsetting the financial design that had transformed it into a titan of computing. The company’s fabs, manufacturing strategies, and assets were being geared towards massive, high priced processors, not churning out substantial figures of low-cost cellular cores. Prioritizing Atom above Main would’ve essential the firm to retool at least some of its fabs to emphasize throughput and decreased expenses in purchase to compete with the ARM processors crafted at Samsung and TSMC. It would’ve meant decreased gross margins and much less gain per device offered.

Intel's Atom Roadmap

Intel attempted to pace matters up finally, but each its 14nm system and the up coming-technology Goldmont CPU main were being delayed.

Intel did choose techniques to improve its competitive standing vis-a-vis ARM and ARM’s foundry partners, but it rarely took them rapidly and usually failed to abide by by means of. Intel purchased Infineon Wireless in 2011 for $1.4 billion, but to this day all of its publicly introduced wireless merchandise, including the XMM 7480 modem, are nevertheless crafted on 28nm at TSMC. Smartphones and tablets have usually made use of SoCs, but Intel did not launch its to start with Atom-based SoC right up until 2012 — 5 decades soon after the Apple iphone introduced and 4 decades soon after Atom’s very own debut.

SoFIA integrated Atom SoC, details

The SoFIA partnership with TSMC raised eyebrows, but not revenue.

Just one point we want to pressure here is that Intel’s decision to guard its main (Main) business and merchandise margins may have been completely wrong, but it was not ridiculous. Refitting fabs, setting up experience in SoC style, and porting modems from TSMC would have essential massive money infusions and choose sizeable quantities of time. If Intel had introduced Atom with an aggressive system to put the chip in smartphones by 2010, matters may possibly have performed out really in a different way. By the time the firm woke up to the threat it faced from ARM and service provider foundries, it was also late to make up the hole.

Why Intel’s Foundry Tech Couldn’t Conserve Its Cellular Small business

Intel’s system technology management couldn’t help save the company’s cellular division for the reason that it was not built to do so. Smartphone and pill OEMs wished equipment with built-in LTE radios Intel did not have them. Even the SoFIA partnership with TSMC under no circumstances came to market, apparently for the reason that Intel couldn’t safe more than enough volume to kickstart creation.

Intel’s 14nm complications delayed its up coming-technology pill processors from 2014 to 2015. Its 10nm node, when envisioned to safe enormous economies of scale above TSMC, has been pushed to 2017 as properly. I really don’t feel these delays performed a substantial position in Intel’s decision to go away the cellular market, but they may have motivated it. In 2012, Intel nevertheless envisioned to be on 10nm by 2016 with EUV (serious ultraviolet lithography) ramping towards comprehensive creation. The now-canceled 450mm wafers weren’t envisioned in-market really this quickly, but Chipzilla envisioned to understand sizeable cost discounts from transferring to the larger wafers in the 2018 to 2020 timeframe — cost discounts that could’ve even more improved its standing against Samsung, TSMC, and GlobalFoundries.

If Intel had started reorienting towards Atom when it introduced the chip in 2008, it might’ve weathered these delays and cancellations without having significantly hassle. Failing to do so remaining its beleaguered cellular business experiencing better-than-envisioned expenses and negligible revenue.

Intel did not reduce the cellular market for the reason that Atom’s general performance and energy use did not compete with ARM research and analysis showed that Atom was able of matching ARM general performance in a number of market segments. It shed the cellular market for the reason that it did not make the changes that would have permitted it to compete on cost with merchandise manufactured at TSMC and Samsung. The exacting policies and exceptional layouts that drove Intel to the leading of a person market could not be conveniently adapted to many others, and Intel was unwilling to chance its posture at the leading of the regular x86 market for a risky payoff in cellular. There is no evidence that trying to keep XScale or establishing ARM merchandise would have transformed that — if something, the ARM division would’ve been beneath even extra pressure to assure it under no circumstances became a threat to the x86 business.

Now examine:

Leave a Reply

Your email address will not be published. Required fields are marked *