After Intel's (NASDAQ:INTC) Q4 2013 conference call in January, I surmised that Intel was cost-disadvantaged in mobile processors relative to competing ARM (NASDAQ:ARMH) architecture processors. Much of this was based on deduction and intuition, as I lacked key facts that would have allowed me to establish the magnitude of Intel's cost disadvantage.
Contra Revenue, Contra Profit
As a result of organizational changes, as well as more information in Intel's Q1 2014 report, analysts and investors can now estimate with reasonable confidence the magnitude of Intel's cost disadvantage. It isn't pretty.
Where before mobile chip sales were buried in the Other Intel Architecture Group, starting this quarter, Intel now breaks these out in the Mobile and Communications Group, as well as providing comparison data for previous quarters. M&C revenue plummeted from $404 million in Q1 2013 to $156 million in Q1 2014. Here, we're clearly seeing the impact of Intel's "contra revenue" pricing policy designed to make Intel chipset costs for tablet makers cost-competitive.
Intel also disclosed that M&C had shipped 5 million tablet processors, giving us for the first time, an ability to estimate Intel's cost per device. Although M&C also sold phone SOCs as well, this is probably a small number (or else, Intel would have announced it), so let's approximate the tablet processor number as the total number of processors sold by the group. This yields an approximate ASP of $31.2, which might be lower in actuality, but represents an upper bound. This is still higher than the IHS estimate for most ARM tablet processors. For instance, the Samsung (OTC:SSNLF) Galaxy Note 10.1 Exynos processor cost $18.8, according to IHS, and this is very close to the $18 IHS estimates for the A7 in the iPad Air.
In the conference call, Intel CEO Brian Krzanich boasted that mobile processor unit shipments were up for the first time since Q2 2012, but this merely demonstrates that if Intel is willing to cut prices sufficiently, it can sell mobile chips. It doesn't demonstrate that Intel can do so profitably.
Intel also reported that M&C had an operating loss for the quarter of $929 million, increasing from the operating loss of $703 in Q1 2013. This means that including cost of sales and operating costs for the group, each tablet processor cost about $217. Probably that cost per processor would come down with higher sales volume, but without a specific breakdown of sales vs. operating cost for the group, it's impossible to say how much higher sales volume would help. Certainly, Intel is making very heavy investments in mobile.
Process Node Hopes
When pressed by analysts on the Q4 2013 conference call, Intel offered that by 2015, mobile processors would be profitable. Presumably, Intel pins its hopes on its next-generation 14 nm FinFET processing node. On the Q1 conference call, Intel once again reiterated that it was on track to begin 14 nm production in the second half.
There's already considerable discussion about whether Moore's Law is dead, and whether process nodes smaller than 20 nm really will offer a cost advantage. Broadwell is Intel's 14 nm version of Haswell, so it will first be rolled out in Core processors, with the 14 nm successor to Bay Trail, Cherry Trail due in Q4. Will these processors be significantly lower-cost than today's 22 nm versions? I really doubt it.
It's starting to look like the contra revenue approach is not so much about temporarily adjusting price for Intel's device partners until Intel's mobile offerings become magically profitable, as it is about predatory pricing designed to drive ARM device makers out of business, whereupon Intel can raise prices to the point of profitability.
Intel investors and supporters may cheer this as a workable competitive strategy, but it's not likely to work. Here's why:
1) Apple's (NASDAQ:AAPL) A7 processor has demonstrated that Intel's process advantage is much less of an advantage from a performance standpoint than many had thought. The A7 is fabricated on Samsung's 32 nm high K metal gate process, so theoretically, it should be inferior to Bay Trail (Intel's 22 nm FinFET), but it's actually very competitive. This provides a rough calibration of cost benefit: ARM processors can lag a process generation behind Intel, while being comparable in performance for lower cost.
2) The ARM ecosystem may not lag by even a process generation if Global Foundries makes good on its claim that it will be in 16 nm FinFET production this year. If this is achieved, and it's a big if, Intel's process advantage will be wiped out. Although I'm skeptical that GF can pull off 16 nm this year, I consider it inevitable that either GF, Samsung or TSMC will bring a 14-16 nm FinFET process into production some time this year or the next. Probably not enough time for Intel to capture the tablet market, no matter how predatory the pricing.
3) The paradigm shift to integrated device makers, such as Samsung and Apple gives Intel much less clout. Between them, Apple and Samsung controlled half of the tablet market in 2013. Both design their own custom ARM SOCs, and Samsung fabs the SOCs of both companies. So both have a clear vested interest in making sure that Intel doesn't dominate the tablet SOC market. The cost savings that Intel can offer matter less to them than the overall device cost. Designing and engineering these devices builds upon their prior experience and intimate understanding of their custom processors, which serves to lower the overall cost of engineering the device. This advantage is something Apple, especially, is unlikely to relinquish.
Intel's Bigger Financial Picture
To be sure, Intel's overall financial performance wasn't that bad. Total revenue grew by 1.5% y/y to $12.764 billion, mostly due to growth in the server segment, while the PC Client Group revenue was flat. Total operating income was also flat at $2.533 billion.
Intel's guidance for Q2 is for flat revenue of $13 billion, but effective guidance for operating income could be optimistic. With gross margin of 63% and operating costs of $4.9 billion, including restructuring, this yields $3.29 billion in operating profit, a y/y increase of 21%.
To achieve Intel's goal of selling 40 million mobile processors in 2014, mobile sales will have to double at least in Q2. Given the contra revenue scheme, M&C losses could increase by about 30% to $1.2 billion, on revenue of roughly $300 million. Here, I've assumed that roughly half of M&C's Q1 costs were operating costs, an assumption that is open to challenge. But I think that there's likely to be some increase in the operating loss of M&C going forward, and this calls into question Intel's optimistic guidance.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.