Over the last six months or so, I have been eagerly awaiting to see how Intel (INTC) would answer a very critical question regarding its strategy. Early leaks were actually quite discouraging, especially in light of the quickly approaching competition from ARM (ARMH) licensees such as Nvidia (NVDA) and Qualcomm (QCOM) in this critical market. However, at the company's recent developer forum, Intel confirmed that it is indeed not ignorant of how important having a competitive solution in this area is and will actually have competitive solutions available this year. Intel just made a huge strategic decision, and I believe that it will have significant repercussions going forward.
It's Not All About Haswell
Most people following the Intel story are well aware of an upcoming wonder chip known as "Haswell". While raw performance per clock on the CPU side goes up, and while the graphics subsystem gets a huge overhaul, the real magic of this chip is that there has been such a heavy focus on reducing idle power characteristics of the chips - particularly the Ultrabook focused ones - to levels unheard of from high performance processors. While "Sandy Bridge" and "Ivy Bridge" sported idle power characteristics that drained battery life, "Haswell" is supposed to fundamentally change that by helping to bring platform power down to sub-100mW (slide courtesy of IDF presentation "Form Factor And Average Power Innovations For Ultrabooks"):
"Haswell" will be a big deal for power-constrained, battery-powered devices, but it is a decidedly high-end processor and will be priced accordingly. That's why while I am excited for "Haswell" for premium notebooks, the real excitement will actually come from the 22nm "Atom" based notebooks based on the "Bay Trail" platform. Intel confirmed at IDF 2013 that not only would "Bay Trail" for tablets be shipping in Holiday of 2013, but that notebooks/convertibles as well as even desktops would get this chip in 2013:
This is a huge deal, and I believe will be even more important than "Haswell" will be during 2013 and 2014.
Why Is This A Big Deal?
I don't doubt for a second that "Haswell" will be an amazing, killer product. The engineers are Intel are pretty relentless when they target a particular performance/watt metric, and I fully expect that mid to high end PCs will be sleeker, faster, and have longer battery life than ever before. But, as Intel's own business over time has shown, people want cheap. Why do you think Apple's (AAPL) 10-inch iPad is falling flat on its rear end, while the 7" iPad Mini, as well as older, discounted versions of the iPad are selling like hotcakes? While most consumers typically have a "main" computer (a home desktop, gaming laptop, a work laptop, etc.) that comes oozing with performance, they also buy incremental secondary machines that are more about low cost, low weight, but still high quality.
Before the slew of Android/iOS tablets powered by cheap ARM based SoCs showed up, the only choice anybody had here was a crummy netbook with a neutered processor, crappy screen, and probably an ugly, cheap plastic body. With "Bay Trail" for laptops/convertibles, Windows PC OEMs will finally be able to build sleek, low cost, but high quality solutions based on a relatively inexpensive, but still quite powerful, 22nm "Atom" processor for the sub-$599 notebook/convertible space:
Intel will have the bases covered at all price points, and this time it seems it isn't going to neuter its "Atom". You can thank ARM and its licensees for pushing Intel to move the low cost PC industry forward.
The major problem here, as I have articulated in previous posts, isn't margins. Intel's "Atom" chips command margins roughly in line with the corporate average. This makes sense given that Nvidia recently disclosed that its "Tegra" mobile SoC business carried roughly 50% gross margins. Given that Intel owns its own fabs (and doesn't pay royalties to ARM), gross margins in the 60%+ range are completely plausible. The problem is that raw ASPs for the chips are much lower than that of the traditional notebook and desktop chips.
Selling a $25 - $30 processor isn't going to give you the raw margin dollars that a $100 processor will, even if the gross margin percentage is the same. If we start seeing a trend where people are simply going with the Atom based solutions rather than the Core solutions, then this will of course be a problem for Intel at the top and bottom lines. But if we see the "Core" solutions staying mostly flat with the rejuvenated Atom helping to gain back market share from the ARM vendors, then this is pure upside for Intel.
My guess is that the "truth" is going to be somewhere in the middle. The people who need performance, will always need performance, and the people who generally bought low cost, would have bought the cheaper "Celeron" and "Pentium" products (these aren't too much more expensive than an Atom/ARM SoC) anyway. I expect that the difference is that while today's "Celeron" and "Pentium" products generally end up in crappy systems with bad screens, slow hard disks, and lousy battery life, the "Atom" products will end up in much more compelling systems, as the PC OEMs/Intel can't really afford to keep the good stuff confined to expensive systems that people may not be buying anyway.
Intel made the right move to unleash Atom and to grin and bear the potential blended ASP erosion that is sure to happen. The key, then, is to focus not on blended ASPs, but to keep an eye on total revenue and gross margin dollars. If these grow as a result of Atom, then great - Intel gets rewarded with a higher multiple as it will have proven its viability going forward, and increased revenues/earnings will only further serve to amplify the share price. If revenues stagnate, then Intel still made the right decision (because it is likely that without Atom being competitive, ARM based chips would have caused continued negative growth), but will need to really focus on increasing the total # of devices that it serves.
In no way is making Atom more competitive a "mistake", and Intel would rather cannibalize itself than let the other chip vendors do it. The big question mark is how total sales are going to be, and whether a competitive Atom at the low end PC + tablet spaces is going to be enough. My bet is "yes", but nothing is ever sure when it comes to business.
Additional disclosure: I am short ARMH