Intel Corp. is in a quandary. The PC sector is flagging, its efforts in mobile communications haven’t paid up quite as expected and the fledging Internet of Things market is still too young to show robust sales. Can the world’s No. 1 semiconductor company find salvation in the foundry business, making custom chips for fabless chipmakers and leveraging its advanced technology process and manufacturing expertise to dislodge current leaders in the sector?
Executives at Intel aren’t planning to turn their business into a pure-play foundry; the market isn’t large enough and becoming a near-monopoly wafer contractor, which the company is in the PC microprocessor industry, is highly unlikely considering the dominance already established by Taiwan Semiconductor Manufacturing Co. Ltd. Also, Intel (NASDAQ: INTC) is not considering spinning off its process technology and manufacturing resources into a pure-play foundry similar to the action taken by Advanced Micro Devices Inc., which spun off its manufacturing arm as GlobalFoundries in 2009.
So, what’s Intel’s goal in the foundry business? Is the foundry focus simply about diversifying the company’s revenue stream slightly or is it part of a major strategic shift similar to others Intel has made in the past? It’s not certain that Intel’s executives know with certainty what exactly they want out of the foundry market. What’s obvious is that the foundry market is growing at a stronger pace than Intel’s core PC business and the company has the capacity internally to meet the surging demand for contract wafer production from the industry.
“The uptake [from potential customers] has been strong,” said Brian Krzanich, Intel CEO during a conference call with analysts last week. “There’s deep interaction between the technical teams on both sides, understanding what the process of silicon technology incorporates, what the design tools are, what IP we have to offer and where they can for third-party IP. Those have been going on with many customers.”
The foundry market is certainly growing at a double digit clip and seems likely to keep on expanding at the same rate as more semiconductor firms outsource wafer production. Demand is surging from companies exiting chip production due to sky-high fab costs – and Intel can quickly become a Top 5 market leader within only a couple of years if it aggressively courts chipmakers. IC Insights estimated the foundry market at $43 billion in 2013, up 15 percent from $37.6 billion in 2012 when it grew 18 percent from $31 billion in 2011.
But while the year-over-year percentage sales increase is impressive, the sector pales beside Intel’s annual revenue of approximately $52 billion, a large chunk of which comes from the PC market. What this means is that while Intel is eager to diversify its revenue source, it can – for now – count on only incremental sales from the foundry market.
This reality has been reflected in the cautious approach of Intel’s management. Intel executives are reputed for being aggressive and direct when making strategic operational shifts. In the 1980s, the company exited the commodity DRAM business to focus on personal computer and server processors with resounding success. Also, over the last decade it has tried to shift operations into the mobile communications market with acquisitions and investments that now run into billions of dollars. More recently, it has begun investing in what executives describe as “new devices and Internet of Things division.”
That same level of passion has been lacking in its foray into the foundry sector. Here, the world’s biggest semiconductor supplier by revenue has so far appeared tentative, almost as if executives were being pushed against their wills to enter the market. After years of internal discussions and one year after signing Altera Corp., its first major foundry customer, Intel recently declared it would expand its custom chip production to other companies. There are limitations even here, though. Certain companies who compete against Intel in other segments of the semiconductor market may not become its foundry customers, company executives noted.
“A company like Altera that doesn't compete with us, we would put them as close to that leading edge noses as they want to be in as they are willing to pay for it,” said Stacy Smith, CFO of Intel during the conference call with analysts last week. “Then, you get into much more interesting strategic considerations. Take the extreme example of Qualcomm. If you give Qualcomm leading edge process technology at a foundry margin, the answer is probably not, because it would make their product so much better. And so now you're taking a foundry margin on one side, but you're creating a much more significant competitor on the other. So those become much more nuance kinds of considerations and conversation.”
Rivals Not Welcome?
The implication is that Intel may not be able to serve as a foundry to certain large semiconductor suppliers against which it competes in its other businesses. In response, these companies may also be unwilling to partner with Intel out of concern their intellectual properties could be compromised at the hands of a competitor. Executives at TSMC must be delighted at this conundrum Intel faces as it tries to enter a market they have dominated for decades.
Foundry rivals shouldn’t be complacent, though. Intel is a formidable player armed with sufficient financial resources to make a dent in the market. Although its PC business is slowing, it remains a large money spinner; in the first quarter Intel posted revenue of $12.8 billion and reported cash and short-term investments of $19 billion, down slightly from $20 billion in the December 2013 quarter.
With this kind of resources, it is possible for Intel to even challenge TSMC for dominance of the foundry market in the next decade, if it wants the crown. Does it? That’s the question investors, analysts, potential customers and rivals would like Intel to answer definitively soon. So far, all the industry has received from Intel are half answers that point towards some interest in becoming a foundry but only as a side operation.
Whatever Intel decides and how far it goes in pushing into the foundry business will have major implications for the entire electronics industry. The company’s capital expenditure on chip fabs and research and development expenses in the last five years alone run into tens of billions of dollars, securing for it tremendous advantages over all other players in the industry. Furthermore, its process technology is considered highly advanced and one of the best in the market; matched only by giants like Samsung Electronics and to some extent TSMC.
Today, the semiconductor industry is awash with fabless chip suppliers that depend solely on foundries like TSMC, GlobalFoundries and other smaller players for wafers. The challenges facing Intel in the foundry market are numerous, but most of these could be easily resolved should the company decide to become fully focused on the sector. The problem is that Intel cannot afford to exit its bread and butter microprocessor business and is unlikely to become a pure-play foundry, that is, a company like GlobalFoundries and TSMC that make wafers solely for other semiconductor suppliers.
Intel won’t be the first integrated device manufacturer (IDM) company to make some of its manufacturing resources available to competitors. Samsung Electronics, the No. 4 global foundry, according to IC Insights, supplies wafers to other chip makers and even offers its services to consumer electronics rival Apple Inc.
“In 2013, Samsung had a 15 percent increase in its foundry sales and was less than $10 million behind the third-largest IC foundry in the world—UMC,” IC Insight said in its research report referenced above. “Samsung has the ability (i.e., leading-edge capacity and a huge capital spending budget) and desire to become a major force in the IC foundry business. It is estimated that the company’s dedicated IC foundry capacity reached 150K 300mm wafers per month in 4Q13. Using an average-revenue-per-wafer figure of $3,000, it is estimated that Samsung’s IC foundry business segment has the potential to produce annual sales of about $5.4 billion.”
Intel, too, could generate several billions of dollars annually from the foundry business but does it have the resolve and can it convince competitors to tap its offerings? Company executives themselves continue to waver. CEO Krzanich said “the interest since we opened up the foundry has been high,” but noted “just getting people comfortable with what the technology is, takes several months typically.” Additionally, “we still have a lot to learn about how to be a good foundry,” Krzanich said.
Apparently, getting Intel itself comfortable with being a foundry appears to an even bigger challenge than securing customers for this business.