Seeking Alpha
Long/short equity, deep value, special situations, growth at reasonable price
Profile| Send Message| ()  

Intel (INTC) reports first quarter 2013 earnings today (April 16, 2013) after the market close.

This former high flyer has declined in stock price by 22% over the past year and has spent the past eight years in a high teens to mid 20s trading range. The P/E of 10 is that of a no growth equity. The stock price is underpinned by a 4.2% dividend. The PEG ratio is .93 caused by an indicated low "G" in the P/E/G of the PEG formula. Intel has been on a CapEx spending tear that would indicate preparation for a multi-year growth rate well above that indicated by today's multiple. My valuation formula, given even a 15% growth rate yields a price target of $40-50 per share.

With the recent reports of a higher than expected decline in PC shipments, most followers of the company are in agreement that revenue and earnings are likely to be less than wonderful. While the earnings are expected to be weak, if they were a total breakdown the company would have warned a few weeks ago.

Anything above the low range of guidance is probably a home run. It's going to be a long conference call unless management throws shareholders a few bones on what the Intel future will bring. First is the capacity issue.

IC Insights estimates total Intel wafer capacity at 5.3 million wafers for 2013. I think that is a little light, but we can use it for now. Compared to the next best manufacturer at the 28/32nm those Intel wafers equal two of the next best wafers, but we will forget that for now too.

The Haswell plus the associated Voltage Regulator takes about 225 sq. mm of silicon, so, on average, they should get 200 good sets per wafer. Figuring 57,000 sq. mm of net silicon per 300mm wafer/225 = 253 candidates, 80% yield =~200/wafer.

Intel has to be prepared to produce 300 million Haswell processor/voltage regulator sets, so 300 million/200 = 1.2 million wafers.

They need capacity for 10 mill big chip servers at about 40 per wafer = 250,000 wafers

Total wafers to run the existing business = 1.45 million wafers at 22nm. In 2010-2011 that would have been equal to about 2.90 million 32nm wafers, which is about what they had back then.

Intel today has nearly 4 million wafers more capacity than can be explained by the present level of x86 business. That's 3.65 times what they need. They could add 2.65 MORE Intels before they were at 100% capacity.

The Chandler, AZ fab (Fab 42) isn't on stream so it can't be included in the 2013 wafer start numbers. Fab42 is targeted for 14nm and 450mm wafers. If you forget about EUV (Extreme Ultra Violet), since it is a technology that might never get here, and figure on that that 450mm plant running today's lithography, FAB 42 will be equivalent to four 300mm 22nm fabs of today. 450mm wafers produce twice as much output as 300mm wafers and 14nm produces twice as much output as 22nm. If that plant were configured for memory, it would again produce twice as many wafers as for logic. Building memory, Fab 42 would be equivalent to eight 300mm, 22nm, logic fabs.

From here it is speculation.

The Ultrabooks and North Cape class PCs are going to need full SSDs, not the cheater hybrid drives with 20GB SSD and 500GB of spinning memory now found in most Ultrabooks. That hybrid drive defeats the Ultrabook concept on performance, weight and power.

The industry is going to have to thumb up the equivalent of about 300 million MORE 128GB SSDs than are available today. Where is the NAND memory to make those drives going to come from?

300 million more SSDs will take 8.5 million MORE 300m wafers of 20nm NAND. Even the excess capacity of Intel doesn't fix that problem.

Now if that 450mm fab were used for memory, it could produce the equivalent of 5 million of the 8.5 million wafers needed. That's a pretty good start.

So the wild card is Fab42. They don't need it for CPU chips. They don't really need EUV to make Fab 42 pay off, they just need the 450mm handling equipment which has been developed, but has no customer for, until now.

Under the above scenario, Intel could build 128Gb chips for about $4 each. Eight for a 128GB drive is $32 plus another $13 to make a drive out of them. $45 cost, $128 resale = 65% gross margin on $38 billion of new business. That seems like a good deal to me.

And they would still have three+ 22nm fabs left over for mobile chips and foundry.

Foundry:

Altera (ALTR) and Cisco (CSCO) are drops in the bucket. Apple (AAPL) has to be hand in order to start to fill the three fabs left over after building all the x86 products. The memory issue that could be in the way of Apple business is all but solved. Again we run into memory as an impediment to Intel growth.

Mobile:

There is more to the mobile issue than, "Intel missed the mobile business". The prime competitors in mobile are Samsung and Qualcomm (QCOM). Intel has been about nine months behind QCOM on LTE communications technology. That should be solved in the next few months as Intel comes out with a multi-mode 4G LTE solution. Once there is parity on LTE, there is nowhere for QCOM to go, Intel will be on a level playing field.

With Samsung as a competitor in mobile, memory (DRAM) raises its ugly head again. Every application processor come with associated mobile DRAM actually mounted on top of the AP chip. If Intel doesn't have a reliable source for mobile DRAM, making application processors loses urgency. That problem should go away with the closing of the Micron (MU)/Elpida acquisition since Intel and Micron have a joint venture on memory.

Intel TV+

It's about time Intel opens up a little about what they bring to the party in media delivery to the home.

Since there is still a lot of "dark" fiber crisscrossing the country, maybe Intel intends to light this up to the home and flood the home with wireless, data rich RF signals. The most attractive thing to the consumer would be totally wireless entertainment services. I could see standard Intel dongles hanging from every electronic device in the home. Those dongles could be high performance low power computing devices that would decode the "cloud" of RF signals for each appliance. Whatever it is, it is time to share the vision with shareholders.

Many of the impediments to Intel growth get solved by summer, so maybe that is why management seems to have optimism for the second half.

The above is Fischer speculation, but I haven't heard anything better up to this point. I hope the analysts serve up something more than the puff ball questions we have all come to expect.

Source: Intel: Earnings Call, Intel Tell Us About The Plan