Business, Enterprise, Hardware

Earnings Call Analysis: Intel all but Confirms Foundry Plans


Few days ago, we’ve posted the news on the Intel’s earnings call for the first quarter of 2012. We focused on the aspects that will be important for the performance (or lack thereof) of the company in the future. During the call Intel executives repeatedly said that some information will be held back until their Analyst Meeting in May.

Regarding their strategy around manufacturing, Intel executives were adamant to point out their competitive advantage in performance, power and cost versus other semiconductor companies. As an example they gave the accelerated silicon roadmap for their Atom offerings, which is moving at twice the rate of Moore’s Law until 2014. This is a spin based on the fact that Atom started behind the curve and so it had a lot of catch up to do. Intel’s manufacturing technology is indeed ahead of the competition, with the company unveiling its 22nm products today on a press event in San Francisco, CA. All 22nm silicon incorporates so-called "3D transistors", while all the competitors use 32nm/28nm process based on conventional planar transistors. Competing foundries will not introduce FinFET i.e. 3D transistor before 14nm process in 2014/15, as we previously published.

Intel 22nm wafer with FinFET transistors on Computex Taipei 2011The company plans to get three additional 22nm factories online during second quarter 2012, spearheading a more aggressive ramp of Ivy Bridge. While this ramp initially leads to an increase in unit cost, it should quickly be compensated as the ramp progresses but gross margin won’t improve until the second half of the year. Even though Intel has more production capacity available for 22nm compared to 32nm at the same point in the ramp up curve, company executives claim they will be able to sell every single Ivy Bridge chip they can make, so fab utilization will be high throughout the year.

Ivy Bridge is poised to be the fastest ramping product in Intel’s history. In the 2nd quarter alone, Ivy Bridge will make up 25% of their CPU shipments and move beyond the 50% mark sometime in fall. Due to the smaller die size the platform cost will be lower at the end of the year compared to Sandy Bridge.

Intel is also heavily investing in building new factories. Right now they spend a considerable amount of money in new buildings. Compared to fab equipment, these don’t depreciate as quickly. Of course at some point they will have to be equipped, driving CapEx (Capital Expenditure) up at some point in the future.

That leads to the question what to do with all that production facilities, considering the PC market doesn’t grow that quickly anymore and Intel’s positioning in tablets and smartphones exists more on paper than anything else. While this could change – Intel is working on that – it seems unlikely in the near term. As repeated rumors alleged, Intel might enter the foundry business at some point. Technically Intel already is in the foundry business, but for now it only supplies a handful of FPGA suppliers which don’t compete with Intel on any business field.

For now Paul Otellini, the Intel CEO reinforced that they are very comfortable with their current strategy of keeping their manufacturing technology largely to themselves. He also pointed out that this allows them to rake in the margins of both the chip designer and the manufacturer, compared to foundries and fabless chip designers. As long as they are able to utilize their fabs to a high degree with their own products this is unlikely to change.

But the CEO also remarked that the possibility of offering foundry services in the future is clearly there. They use the contemporary fab contracts as learning opportunities, to build up development libraries and tools to enable rapid chip design and modularity which will be required for their SoC business in the future. But don’t be fooled, Intel is also making healthy profit on the way. When asked directly whether he’d prefer Apple or Qualcomm as a foundry client, Otellini remarked he’d find Apple a lot more attractive.

The CFO chimed in on the foundry business subject, explaining that their integrated approach allows them to do quick ramps as we shall see with Ivy Bridge. This is something that their competitors are not capable of. While not naming any specific company, he also pointed to other fabless companies that struggle with ramp and yield rates, alluding to problems faced by AMD, NVIDIA, Qualcomm and the likes at the 28nm node.

The CEO didn’t stop to tout the advantages of Ultrabooks. To this date, 21 different designs are shipping with 100 more coming around later this year, most of which will be based on Ivy Bridge. Intel invested heavily in various aspects of Ultrabooks by means of their Ultrabook fund. Some of that money will be spent on making sure that enough capacitive touch screens can be produced to meet the demand as well as come down in price.

Otellini believes that there will be an incredible amount of differentiation. Towards the end of the year Intel also expects a few designs incorporating touch interfaces and hybrid designs that combine the advantages of the tablet and notebook form factor. Think of it as the Intel-powered version of the ASUS Transformer Prime. These devices should become particularly disruptive with Windows 8, which got a touch-optimized UI. Intel believes that this combination of innovative products will deliver growth.

Price-wise the CEO pegs Ultrabooks at $699 to $1299. However, he doesn’t think it needs mainstream prices (i.e. $499 to $699) to achieve high volume numbers. At the end of 2012 he expects 40% of consumer notebooks to be Ultrabooks.

The Intel executives again reiterated that they see tablets as complementary devices to PCs/notebooks. With Ultrabooks they now try to offer some of the features that made tablets so attractive in the PC space, like touch interfaces, instant-on capability and the sleek form factor. In the future Otellini sees devices that are somewhere in between todays tablets and Ultrabooks.

As a funny side note, Intel is trying to apply the "eat your own dogfood" approach to Ultrabooks, even though their CIO seems to be a bit reserved in that regard according to the CFO. Later this year there should be vPro enabled Ultrabooks too, which should help improve enterprise adoption.

Server business remains healthy with a growth target of 15% over the course of the year. This is driven by the growth of Internet use and cloud services around the world. In the first quarter Intel finally launched the Romley platform which brought Sandy Bridge CPU architecture to the server space. Compared to Boxboro, Romley shipped about twice the volume in the same timeframe. Some of that might be due to the delays the platform
s faced at the end of 2011.

In the "not too distant future" we should expect the MP version of Sandy Bridge-EP/Romley platform which will replace the current Westmere-EX/Boxboro based offerings in this space with up to 10 cores. A bit after that a lower end version of Romley should emerge as well. At this point no specifics about these products are given. There was also no word about a possible 22nm refresh of the server chips, which could mean leaked roadmaps suggesting Sandy Bridge-EP/EX should be good into 2013 might come true.

For the microserver segment, Intel plans to release Atom-based SoC codenamed Centerton featuring 2 cores, 64-bit, ECC memory support and virtualization – all within a 6W thermal envelope. The chip was first mentioned at IDF Beijing last week. It will be really interesting to see how microservers evolve once AMD gets their first products based on the SeaMicro fabric out. Expectedly, SeaMicro is expected to be the biggest customer of Centerton. We believe this will provide invaluable hands-on experience for AMD engineers.

Intel Xolo 900 is a mass-produced version of the Medfield Reference PlatformIn the smartphone segment, Intel still sees an opportunity for business with its own chip designs. A few days ago the first smartphone featuring the Intel Medfield SoC launched in India. However, at a price of $423, it is not hitting the previously announced product strategy of going after Qualcomm, NVIDIA, TI and others from the bottom, but rather joining the crowded space of high-end smartphones.

Second smartphone product comes from Orange, but both phones are actually mass-produced versions of its own reference design, rather than a unique take. Apart from a few niche designs, Intel currently doesn’t have a lot to bring to the table in this segment. Paul Otellini explained that their manufacturing technology is one of the assets they want to leverage to further this business. He also pointed out that there are good relationships with carriers since Intel helped them build data centers and network infrastructure.

PC market and HDD shortage
As the biggest semiconductor maker Intel is seen as an industry barometer regarding the HDD shortage following the events in Thailand last fall. In particular Intel executives explained that it lead to a decline in inventories, that is, OEMs and the channel reduced their inventories down to a minimum to reduce the risk of not being able to sell them.

Due to a replenishing supply chain and the launch of Ivy Bridge in the second quarter, Intel expects to be above seasonal. Going forward into the 3rd quarter, HDD availability shouldn’t be a problem anymore and the supply chain should build up a bit bigger inventories again leading to higher sales for Intel.

The greater trends in the PC market remain unchanged. The developed markets, i.e. North America and Western Europe, continue to show stagnant or even slightly declining sales numbers. Most of the growth comes from emerging economies, which is a bit more desktop-centric according to the CEO. Overall notebooks are outgrowing the desktop though. In short the environment didn’t change.

What might change things a bit is the launch of the next operating system from Microsoft – Windows 8. There circulated some rumors that Windows 8 might be delayed into 2013, but if Intel’s CEO is to be believed this is nonsense. Either way, it won’t affect the bottom line of Intel before the holiday shopping season at the end of the year.