What's with Broxton?.....this guy says not till mid 2016......if so.....just more terrible execution.........where's mas???
Intel Broxton: Time To Set Realistic Expectations
Mar. 5, 2015 12:50 PM ET | About: Intel Corporation (INTC), Includes: QCOM
Summary
Intel's Broxton could be delayed to H2 2016.
The delay will allow Qualcomm and MediaTek to close in on FinFET drive.
Intel may have to subsidize or provide deep discounts to make its Broxton competitive.
Intel (NASDAQ: INTC) has been widely criticized for its contra revenue program in investment and technology forums over recent months. The chipmaker's mobile business group posted a gigantic loss of $4.2 billion in FY14 primarily due to the subsidies it provided to tablet and smartphone vendors in order to capture a meaningful market share in the cutthroat mobile industry.
The chipmaker, however, claims that its upcoming SoFIA/Broxton chips will solve its bill of material-related problems, which should in turn, negate the need for its contra revenues. These chips have already been termed as disrupters in a few technology forums. While this may portray a rosy and optimistic picture, I'm of the opinion that Broxton won't be an outright game changer for Intel.
Delay on cards?
First and foremost, I keep a close eye on Zauba.com - an online database of India's daily import/export data - to gauge the progress of technologies that are under-development. Intel has its R&D facilities set up in India, so the units that it ships in and out of India for sampling and testing purposes are bound to reflect on the website.
For instance, Intel's Cherry Trail samples started showing up on the site almost a full year before their mass-rollout. However, the cause of concern here is that Intel's Broxton chips haven't shown up on Zauba yet. They're scheduled for launch in early 2016, and the absence of their import entries on Zauba lead me to believe that they would end up getting delayed to H2 2016.
(click to enlarge)
(Source: Zauba.com)
To put things in perspective, Intel had originally scheduled its Broxton chips for launch in mid-2015, which the chipmaker later revised to early-2016. A delay to H2 2016 would simply spell trouble for the chipmaker. The cutthroat semiconductor industry is dominated by companies that are ahead in manufacturing designs and processes. The delay would now allow Intel's competitors to close-up on the 14nm/16nm FinFET drive.
Stifling competition
I compiled a chart of Geek bench scores to put Intel's competitive landscape in perspective. Intel's latest 14nm-based Cherry Trail chips barely manage to perform in-line with existing high-performance 28nm and 20nm-based Qualcomm (NASDAQ: QCOM) offerings. Even downmarket offerings from MediaTek, manufactured using a relatively older and cheaper 28nm process, compete fiercely with Intel's latest Cherry Trail. The rollout of 14nm/16nm respective Qualcomm and MediaTek chips will only make Intel's Cherry Trail look dated in terms of computing performance.
(click to enlarge)
(Source: Geekbench)
Where Samsung has already started the mass production of its 14nm FinFET chips, Qualcomm and MediaTek will join the FinFET bandwagon with 14nm and 16nm chips by 2015-end, respectively. The only thing separating Intel and its competitor's chips at this point in time would be their architectural differences, transistor efficiency, production yield and price points.
With Intel's node advantage narrowed by then, the chipmaker will have to compete with Qualcomm's high-performance designs or downmarket chips from MediaTek. Intel may choose to market its Broxton as affordable high-performance chips, but there's still no guarantee that it would be able to beat 16nm MediaTek chips in terms of performance or price.
Suggesting that Intel's next-gen Goldmont architecture can deliver impressive gains over existing platforms would be pure speculation. Intel is yet to prove its prowess in mobile designs - its Airmont chips lag Qualcomm and MediaTek offerings by a significant margin on an adjusted-node basis. Qualcomm's upcoming Snapdragon 820, based on a new 16nm platform called Zeroth and scheduled for a year-end launch, may further extend a lead over Intel offerings in terms of compute performance.
Going by Intel's tick-tock cycle, Broxton's successor will most likely be built on the Goldmont architecture as well. This essentially means Intel will be locked-in with the architecture at least until 2017, while its competitors continue to up the ante with new designs and die shrinks each year. This, in turn, suggests that the Goldmont architecture would have to ride out two FinFET generations from Qualcomm and MediaTek - something that could allow Intel's competitors to further extend their lead.
Final thoughts
Intel's Broxton line-up may not have bill of material problems, but that doesn't necessarily make them competitive offerings. I'm of the opinion that the Broxton line-up will lag offerings from Qualcomm and MediaTek in both price and performance metrics, keeping the aforementioned points in mind, which would make Intel's Broxton chips tough sells.
The contra revenue program lowered its Bay Trail's ASPs and essentially made chips competitive on a price basis - something I believe is the main reason Intel why was able to sell 46 million tablet chips in the first place. The chipmaker might have to again resort to its contra revenue program, or provide deep discounts, to make its already-delayed Goldmont-based Broxton chips competitive on a price basis.
Source: Intel Broxton: Time To Set Realistic Expectations
Share this article with a colleague
Tweet
