Intel (INTC) makes its money from x86 chips but the PC market for those is dropping. Last quarter, according to both IDC and Gartner, the PC market was down nearly 8% year-on-year and according to NPD, down 21% at US retail in the 4 weeks after the release of Windows 8. The expectation from the analysts was that the new version of Windows would boost PC sales. Manufacturers didn't see it that way and haven't rushed to put out new Windows 8 PCs for the holiday season. Without an obvious productivity gain, businesses too are likely to continue on Windows 7 or XP, where their existing software already works and which requires no new training and allows them to delay buying new PCs. When this is combined with the slow economic recovery in the US and the much worse situation in Europe, PC sales are likely to be off this quarter at least 5-10% year-over-year.
Tablet sales, though, will be up. They should top 30 million for the first time this quarter, with over 20 million Apple (AAPL) iPads and iPad minis and Amazon's (AMZN) Kindle Fire and Google's (GOOG) Nexus increasing sales. Many consumers are realizing they don't need so much PC power at home for email, the web, photos and basic entertainment. That it's good to be able to carry around a big enough screen with instant on. However all these big selling tablets are ARM based and Intel doesn't make ARM chips yet.
Smartphones too will have a record quarter but the iPhone, the best selling Android phones, the Blackberries and even the Windows phones don't use x86 chips either. So far, Intel has managed to miss out on mobile in a big way.
This leaves next quarter's x86 tablets as the main new desktop hope for Intel but will these even fill the gap from the drop in PC sales? Microsoft's Surface with Windows RT has had lackluster sales so far, despite it being displayed and demonstrated in 65 Microsoft stores and pop-up stores across the USA, and more recently in Staples and Best Buy. It sells at the same price points as the iPad but lacks the app catalog and having Office available hasn't made up for that. As x86 tablets like Acer's W510 don't yet offer great Windows 8 performance at iPad prices and Microsoft's x86 tablets will be even more expensive than the Surface RT, a consumer stampede seems unlikely.
So, if it makes economic sense for businesses to largely stay with Windows 7 and XP and continue to use their old PCs for longer, who will buy the x86 tablets? After a Touch cover or Type cover is included with a Surface, they cost MacBook Air-money. So why wouldn't any dedicated Office user, who needs portability, not buy an Ultrabook instead? They have a better keyboard, better battery life, larger screen, don't weigh much more and generally are cheaper.
Paul Otellini, Intel's CEO, will retire in May. He launched Ultrabooks to compete with MacBook Airs, subsidizing their promotion with the idea they would take over 20% of the PC market. This hasn't happened yet, but the Ultrabook campaign naturally caused friction with Apple.
Intel's new CEO will be able to look at Intel's mobile strategy afresh, and sit down with Tim Cook to see how Intel can profitably capture Apple's ARM chip business. Between the iPhone, iPad and iPod Touch, this quarter Apple will use about 80 million ARM chips, redesigned in-house from the architecture supplied by ARM Holdings (ARMH). Apple's ARM chip needs should exceed 300 million in 2013, which is about the same size as the PC market and so, after taking out the AMD share, that's more than the x86s Intel will sell for PCs.
Intel fabs are running at much less than capacity and are over a generation ahead of all the other possible foundries for Apple. Intel is already running at 22nm and moving to 14nm, while TSMC, the largest alternative, is now at 28nm and moving to 20nm. Apple could therefore skip a generation from the 32nm used in the iPhone 5 and iPad 4, going straight to 14nm, and use the process shrink for better performance and better battery life and be a generation ahead of competitors. Even if Apple takes a more conservative approach, it can move all its ARM chip production to Intel's already proven 22nm process while rivals deal with any teething troubles in the ramp up of equivalents.
Apple also wants to move all the business it can from Samsung (SSNLF.PK), its greatest rival in the high end smartphone market. It would certainly prefer that Samsung didn't see Apple's optimized ARM chip designs before they are part of a shipping iPhone or iPad. In addition Samsung has no great incentive to even move as fast as TSMC, as while it produces Apple's ARM chips, Samsung and Apple are both on the same fab process. This can create the impression that Apple is falling behind other rivals and therefore the iPhone and iPad are not as premium as they used to be.
Apple's main advantage now in ARM chips is the integrated graphics licensed from Imagination. In benchmarks, Apple's A6X is competitive in CPU tests but surges ahead in areas like games, which can use the better graphics. However Samsung too can integrate better graphics, and given the size of the Galaxy screen, has every incentive to do so.
So it makes sense for Apple to move to Intel and once again increase the performance gap between iPhone/ iPad and the Android phones and tablets. If the usual 2 years for a chip generation holds, iPhones made now with chips from Intel's 14nm line will still be competitive in 2015 when Samsung moves to that process. So then Apple will be able to market its cheaper 2 year old iPhones to the mid range as being as up-to-date as the best Samsung phones.
In summary, if it produces Apple's ARM chips, Intel:
- gains new production of 300+ million chips / year
- $5.5-6bn (+11-12%) more revenue at $18 to $20 per chip
- 50% gross and 40% operating profit should be possible (TSMC makes mid to upper 40s gross with an operating profit of low to mid 30s)
- possible exclusivity premium from Apple and / or payments for equipment for foundry upgrades
- strong further use of the current 22nm lines where depreciation has largely been paid for with x86 production
- good yields, as many problems will have been sorted out with x86
- better fab utilisation
- diversifies revenue stream away from x86 and Wintel
- will reinforce Intel's position as the no. 1 chip producer
- exclusivity from a trusted partner, possibly in exchange for a premium, who won't produce chips for Samsung.
- strong partner it has no need to subsidize, able to produce all the ARM chips Apple needs
- worldwide chip production from up to 8 fabs reducing possibilities of disruption in the supply chain
- US produced chips - important given the commitment to US manufacturing
- faster access to future chip processes which give longer battery life and/ or better performance
- helps with marketing iPhone as always ahead of other phones
- removes the Samsung drag on improvements and increases the time it takes for Samsung to catch up
- revenue down by $3.5-4bn (current ARM chip production for Apple)
- chip costs for its own phones and tablets go up
- no more foundry subsidies from Apple paying for plant and new equipment
- lower throughput to take care of depreciation.
- weakens position as no.2 chip producer by revenue
- reduces likelihood of holding on to a large part of Apple's NAND memory chip needs (Apple is the world's largest buyer)
Samsung clearly hopes that it can replace the profits from chip production by selling more phones and tablets. It may well do, if being a generation behind Apple doesn't make Samsung's top end ranges look like also-runs.
Disclosure: I am long AAPL.