Over on Seeking Alpha, in a thread discussing Intel's (NASDAQ: INTC) acquisition of Altera (NASDAQ: ALTR), one commenter said the following: "It is undeniably a good thing that [Intel] gets to use the excess capacity for a different kind of chip and put more money in their pockets."
This is a sentiment that is surprisingly common, but I believe it's fundamentally wrong. Here's why.
Chip design takes years
It takes in the neighborhood of four years to bring a complex semiconductor design, such as an FPGA, to market. As a relevant example, Intel and Altera first announced that they would be working together to build the latter's Stratix 10 FPGAs in early 2013, but Alter's press release today says that engineering samples won't be available to customers until the second half of 2015.
Now, given that just about every chip in Altera's product portfolio is built at non-Intel factories, and given that Intel CEO Brian Krzanich said that Intel has no real "need or desire to move existing [Altera products] into the Intel silicon," most of Altera's products will be built externally for years to come.
So, given that it takes years for chip designs to go from design start to mass production, and given that the majority of Altera's chips are going to be built externally until new Intel/Altera FPGA designs hit the market years from now, Intel won't be able to use Altera to "fill up its factories" anytime soon.
Speaking of "filling up capacity"...
Another problem with the idea that Intel bought Altera to "fill up excess capacity" is that semiconductor capacity additions are planned years in advance. On the Intel conference call discussing the Altera acquisition, analyst Timothy Arcuri asked if the Altera deal affected Intel's planned 2015 capital spending.
Krzanich said that 2015 capital spending is "laying out the capacity for 2016 and 2017," and that the company "already had Altera built in as a foundry customer." The integrated FPGA and Intel CPU products that Intel hopes to build with this acquisition are, according to Krzanich, "outside of this capital cycle."
So, if the Intel/Altera products are outside of this capital cycle, then how could these products possibly be used to "fill up existing capacity"? Indeed, if anything, the move of Altera's future FPGAs to Intel silicon, as well as the added silicon footprint that integrated FPGAs with CPUs will require, will likely increase Intel's capital expenses for capacity that's built to satisfy demand in 2018 and beyond.
If Intel didn't buy Altera then it would simply spend less on building out capacity for those years.
"Excess capacity" is a misused catchphrase
There seems to be this widespread view among investors that Intel has tons of excess capacity that it desperately needs to fill. However, if you take a look at this chart that Intel CFO Stacy Smith showed at the late 2014 investor meeting, you'll see that this notion is -- to put it mildly -- misguided:
The green line is factory loadings, and the orange line is total equipped capacity. The closer those two lines are, the more utilized Intel's factories are. As you can see, by the end of 2014, Intel's capacity and factory loadings were quite close to each other, implying very little unused capacity.
Intel did signal on its last earnings call that it was lowering the utilization of its 22-nanometer factories as it plans to repurpose that capacity for 14-nanometer as a result of weaker-than-expected demand for 22-nanometer desktop chips in the first half of the year, so utilization is not as good as it was at the end of 2014.
However, Intel still isn't taking excess capacity charges (Intel executives said that the lower 22-nanometer utilization will show up as an increase in cost of goods sold in 22-nanometer chips), and I'd imagine that 14-nanometer utilization will be quite healthy as Intel continues to ramp 14-nanometer products.
At any rate, the idea that Intel just dropped $16.7 billion on Altera in order to fill capacity that's outside of the current capital cycle just doesn't make sense to me.