Can Intel's Foundry Services Move the Needle?
Intel (INTC) is the world's largest semiconductor company with annual revenues of over $50 billion. Unlike other large semiconductor spenders like Samsung (SSNLF.PK), Global Foundries and TSMC (TSM), Intel does not manufacture semiconductor components for other companies on a large scale. Intel has been facing low factory utilization because of a sharp slowdown in global PC shipments. This has made Intel look for ways to utilize its manufacturing capacity. The company will give more emphasis to foundry manufacturing (according to the previous CEO Otellini). Intel has already managed to get six customers for its foundry services and it will keep getting more customers, as no other company can offer 22 nm edge manufacturing capacity.
The main question is whether Intel's foundry services can become a meaningful contributor to Intel's topline. I think that they will become a huge contributor as soon as Intel gets a big customer like Apple or Texas Instruments. These companies can buy billions of dollars in chips every year, if Intel can give them a compelling offer in terms of price and capacity. Intel has said that it will never do foundry services for competitors like Nvidia, AMD etc. However, there are many semiconductor companies in the memory and logic semiconductor industry that do not compete directly with Intel. These can become large Intel customers as the company focuses on increasing its manufacturing lead. Intel is already making big strides in mobile chip manufacturing. Foundry services could become another major growth driver for Intel. I remain highly positive about Intel's stock for many other reasons and think that it offers one of the best ways to invest in the technology space.
What is Foundry and the "Fabless" Model
Intel is a vertically integrated company that designs and manufactures chips itself. Other big semiconductor companies like Qualcomm (QCOM), Texas Instruments (TXN), Nvidia (NVDA) and Advanced Micro Devices (AMD) follow the "fabless" business model. In the fabless business model, companies design the chips while the manufacturing is done by foundries such as TSMC and Global Foundries. Intel is the last remaining major non-memory company that does not use the fabless model. The reason behind the fabless trend is the huge expense in building a new "fab" or semiconductor factory. It can cost ~$3-$5 billion to build a new fab, which is beyond the financial capabilities of most companies. While the fabless model has the advantage of reducing capex, the disadvantage is that you cannot get sufficient capacity at the leading edge. Even big foundry customers such as Qualcomm and Apple (AAPL) have faced chip shortages, which has forced them to push out their product launches.
Intel has a big manufacturing lead over other companies
Intel doubled its capex spending in 2011, as the company focused its efforts on getting into the mobile processor space. This jump in expenditure allowed the company to build new 300 mm fabs focused on the most advanced technology node of 22 nm. Intel's servers and PC processors have decimated the competition and now the company is looking to capture mobile processor space. Intel spends as much money on building processor capacity as the rest of the industry. Readers should note that almost all of Intel's expenditure goes into building "Logic" capacity, while that of other big spenders like Samsung goes into both "Logic" and "Memory." As semiconductor processes have become more advanced and complex, most semiconductor companies like Texas Instruments have left manufacturing and outsource production to foundries like TSMC and UMC. There are only five-six companies left that can build a new 300 mm fab because of the prohibitive costs. This chart from IC insights shows that there are only five companies that will spend more than $3 billion in semi capex in 2013.
The Lead will only get bigger
Intel has bought a substantial equity stake in the largest semiconductor equipment maker ASML. Note ASML is only one out of the three companies that can make advanced lithography machines for the semiconductor fabs. It has a sizable technology lead over competitors like Nikon and Canon in making this very expensive equipment. Intel has invested money in ASML to accelerate technology advancement and further its lead over competitors. Intel is also spending $2 billion to build the first 450 mm fab in the world. These new fabs will use 450 mm wafers instead of 300 mm wafers. The 450 mm wafers will dramatically decrease overall cost as the new wafers will produce almost 2x more chips at the same cost.
Why Intel is going into the Foundry business
One of the criticisms that Intel is facing is that its fabs are running at low utilization. The reason is that PCs are declining at a faster pace than expected and Intel wants to keep a low inventory level. Intel's capacity is set to grow bigger as it completes the new 14 nm fab, which will grow its capacity tremendously. Even if Intel wins a huge market share in mobile chips, the company will still have a large leading-edge capacity lying idle. Some commentators are also speculating that Intel might use its advanced fabs to produce leading-edge NAND memory chips. While this may be true, Intel has already got some customers like Altera (ALTR), Microsemi (MSCC) as well as networking giant Cisco (CSCO).
Microsemi, the mixed signal military IC supplier which bought Actel, has become Intel's fifth announced foundry customer after Achronix, Tabula, Netronome and Altera.Microsemi will be getting its first parts made by Intel either late next year or early in 2015. Microsemi will only use Intel for its digital products.
Will Foundry Services become a meaningful contributor to Intel's revenues?
I don't think small vendors like Altera can substantially contribute to Intel's revenues given that these companies are quite small. Intel's foundry business can become a big contributor only if it manages to win a big customer like Apple . Intel has said that it will not provide foundry services to competitors. Apple currently makes chips using ARM (ARMH) chips. However, Apple is a monster in the technology industry and can easily become a $5-10 billion customer for Intel. This will mean compromising on its strategy of not making chips for competitors (ARM), but Intel could easily make ~$5 billion in extra profits by having Apple as a customer. As noted, there are also other companies in the semiconductor industry, which do not directly compete with Intel. However, these are too small to provide enough revenues to a giant of Intel's size. Samsung has managed to successfully compete with and become a supplier to Apple. Intel is not directly a competitor to Apple as it does not make electronics products like mobile phones and tablets.
We are very interested in being a selective foundry manufacturer for certain customers. We don't see ourselves as a general purpose foundry or competing with the general purpose foundries. The kinds of things we would do, we would not take business that would enable a competitor. We would certainly consider business that would enable and strengthen relationships with strategic partners.
The kinds of things that we've announced so far have been in the programmable logic area, which is an area that Intel is not in today, so that makes perfect sense and those kinds of companies need leading edge technology. So, that's the pattern for which we would do this, so yes, we are in a position to be able to handle those kinds of customers. It's we've been building up that capability for several years now and we are now going into production. I described this I think in May as a crawl, walk, run strategy and we are still in the crawl part of it.
And to the second part of the question around our expectations around return, I mean it's pretty simple. To the extent that we engage with these foundry customers, we want to make money at it. We want to get paid in terms of margin and we want to get a return on our invested capital commensurate with our technology leadership.
Stock Performance and Valuation
Intel's stock has been going up in the last few months and has broken its trading range of $20-22. Analysts had feared the worse due to the PC shipment slowdown and the stock went up when the fears turned out to be false. The company is trading at~$24, which gives it a forward P/E of ~11.5x. The biggest attraction of Intel's stock is the ~3.75% dividend yield, which is the highest for a big technology stock. The company also buys back a massive amount of stock each year using its monstrous operating cash flows.
I have repeatedly written about the advantages of buying Intel stock, as the market is not factoring in Intel's new catalysts. The world's largest semiconductor company possesses massive advantages in manufacturing, which is not well understood by the market. The company's newest 22 nm fabs are running at ~50% utilization and the company still makes gross margins in excess of 55%. The company recently appointed a new CEO who has a strong manufacturing background. This means that Intel will continue to invest strongly in building and upgrading its fabs. The company has already started winning over small customers for its foundry services. Finding more customers will not be hard as it is the only company in the world offering services at the 22 nm node, which has a big advantage in terms of performance over other foundries. The market is not currently factoring in the possibility that Intel might win a huge foundry customer like Apple. The stock could easily go up by 50% if this materializes. Intel is also set to capture a large share of the mobile chip market with the introduction of "Baytrail" and "Merrifield" processors later in the year. I think that Intel has a terrific risk-reward profile and should be bought at the current price.