Intel, Apple: High Hurdles to Any Foundry Relationship, Says Bernstein

Bernstein Research’s chip analyst Stacy Rasgon and Samsung Electronics (005930KS) analyst Mark Newman and Apple (AAPL) analyst Toni Sacconaghi this morning issue a combined report on the prospects that Intel (INTC) might win foundry business from Apple, concluding that “such a relationship would be fraught with business and strategic issues (on both sides) that would require extremely careful management.”

The authors note that debate has “increased recently” over whether Intel may win some business fabbing Apple’s “A” series processors for its iPhone, iPad and iPod Touch, given the chip maker’s “significant capacity additions” and the constraints at other chip foundries, including Samsung’s.

Although Intel is substantially increasing capital spending on its factories, with $12.95 billion projected for next year, up from $12.55 billion this year and $10.76 spent last year, that doesn’t necessarily mean Intel will make chips for Apple, despite speculation along those lines. “We disagree with this simplistic line of reasoning � any decision to enter foundry will be a strategic choice, independent and distinct from Intel’s current capex strategy.”

Intel has already said that if demand falls short of what they can produce, the company will repurpose chip making assets to the next node, they write.

On the one hand, it’s unlikely Apple would do what Intel wants, which is to shift from using a processor core based on designs from Intel competitor ARM Holdings (ARMH):

We believe a move by Apple to make broad use of x86 in their iOS line is unlikely. We believe that Apple gains significant advantages from their vertically-integrated stack of semiconductor design, software, and hardware – note, for example, that their iPhone 4S, while manufactured at a mature 45nm node, is better on power and performance than virtually all other devices today in a testament to their abilities (Exhibit 5). While access to 22/14/10nm from Intel would provide significant advantages, we believe a move to x86 on iOS would therefore be a non-starter as Apple would lose their ability to optimize the full platform as it currently does with their internal ARM architecture.

However, given the ongoing patent battles with Samsung, Apple has a real need to consider alternatives, the authors write. They might be compelled to consider Intle if Intel could grant a similar “Chinese Wall” between its own processor development and the tabbing of chips for Apple, as long as Intel could gain insights that would benefit its own mobile chip design.

The authors suggest Intel could manage a favorable profit margin on foundry services for Apple that would be comparable to its overall net profit on its own devices, despite having higher costs than the biggest foundry, Taiwan Semiconductor Manufacturing (TSM):

Consequently, our analyses suggests that overall foundry net income as a % of revenue could be in the ballpark of 20% or so (Exhibit 15), not hugely different from our current model for Intel as a whole (~23% in 2014), assuming a foundry R&D opex burden at 1.5x the rate of TSMCs (e.g. typical foundry R&D, plus an additional amount to cover design services etc that Samsung provides to Apple, assuming that process R&D etc can be leveraged from Intel’s core business) and minimal incremental SG&A (we assume ~1%) [�] Our base-case modeling suggests entering into a foundry relationship with Apple could drive in the ballpark of 30-35 cents in EPS upside, relatively material (our current 2014 EPS forecast, for instance, is $2.97).

The authors cite multiple issues that would be raised that Intel would have to get past:

Accelerating the performance and/or value proposition of Apple’s iPad via better process technology could increase the risk of PC cannibalization, certainly an undesirable outcome. A similar dynamic could conceivably play out in the mobile space, where Intel is attempting to make their own efforts. Apple’s iPhone is already viewed as being markedly superior to most other platforms; should Intel’s process technology make it even more compelling it may hamper Intel’s own efforts to penetrate the space with their Atom offerings.�We must also be cognizant of the fact that such an arrangement with Apple would open Intel up to significant customer concentration and, potentially, risk given what would likely be dedicated capacity should Apple decide to (again) change foundry partners. Finally, we struggle to see a scenario where the return on capital will ever be as good as Intel’s current business; given they currently earn two margins � a manufacturing and design margin � on their products, their return is extraordinarily high.

Intel shares today are up 23 cents, or 0.9%, at $26.77, Apple shares are down $1.71, or 0.3%, at $570.45, and Samsung shares fell 0.8% to ?1.26 million in Taipei trading.