At Seeking Alpha, contributor Brian Gilmartin points out that analog-chip maker Texas Instruments (TXN 1.80%) has reduced its capital expenditures over time and wonders whether Intel's (INTC +23.60%) acquisition of FPGA maker Altera (NASDAQ: ALTR) is evidence that Intel might, too, become a less capital-intensive business over time.
"I thought it would happen now, but I was hoping that Intel would be able to gradually reduce its capex requirement very time. Maybe the [Altera] acquisition is the genesis of that change," Gilmartin writes.
I'm going to cut to the chase and answer Gilmartin's question outright: No. Read on for a more complete take.
Buying Altera isn't a step toward being fabless -- quite the opposite
In the semiconductor industry, the majority of design houses today are "fabless." This means that they outsource chip production to third parties, such as Taiwan Semiconductor (TSM +5.08%) and Samsung (NASDAQOTH: SSNLF). Most semiconductor companies are fabless simply because they don't have the unit volume or revenue scale to support a leading-edge factory.
According to Intel, a leading-edge semiconductor factory "now costs more than $5 billion when fully equipped." If it costs a company $5 billion for a leading-edge factory, then it had better have significantly more than $5 billion worth of leading-edge business per year to fill that factory.
Intel, by virtue of its extremely large PC Client Group and Data Center Group businesses, has the kind of scale necessary to invest in building leading-edge factories. A fabless company such as, say, Altera, with less than $2 billion in annual sales, would be crazy to try to build its own leading-edge semiconductor manufacturing plant.
That's why most of its products are built at TSMC, and why it contracted with Intel to build its upcoming high-end Stratix 10 FPGAs. Over time, I expect all of Altera's future products (as well as products that integrate both Altera and Intel technologies) to be built in Intel's factories.
This, contrary to what Gilmartin suggests, will lead to increased capital spending on Intel's part to make sure it has the capacity to support Altera's products. This capital spending won't show up immediately, since it will take time for the combined Intel/Altera products to go from concept to shipping designs, but make no mistake: Intel will very likely need to build additional capacity at some point down the line to support the Altera volumes.
Capital intensity isn't a bad thing
With the proper revenue/unit scale, owning your own factories and manufacturing process recipes is a competitive advantage, not a disadvantage. In fact, former Intel CEO Paul Otellini often made the point that as an integrated device manufacturer (i.e., a semiconductor company that designs and builds its own products), Intel gets "paid twice" for its chips.
By this, Otellini meant that Intel gets paid the manufacturing margin (what a third-party chip manufacturer would collect) as well as the design margin (what a fabless customer would collect from selling the final product to a customer).
This model works out beautifully -- if you have the right products that people want to buy.
Intel's problem is that it hasn't had the right products
Gilmartin points to Intel's recent capital-expenditure reductions as a positive, but it's quite the opposite. Capital for semiconductor equipment is spent in anticipation of demand that's expected to materialize at some point down the line.
For example, according to Intel CEO Brian Krzanich, the capital that Intel is spending here in 2015 is to support capacity buildout for the demand that the company expects in 2016 and 2017.
Now, to be fair, the capital expenditures in 2015 are unusually low because the company is taking 22-nanometer capacity offline and reusing 22-nanometer equipment for 14-nanometer production, but the point still stands: Increasing capex seems to indicate greater expected future demand.
At any rate, Intel's "problem" isn't that it's spending heavily on capex; the problem is that the company's highest-volume and highest-revenue business -- its PC platform business -- is seeing demand that's flat to down. Intel also doesn't seem to be capitalizing on the significant demand for mobile processors, which could bring in significant revenue and unit volumes to Intel, which would drive the need for additional capital spending.
Intel will (probably) always be a capital-intensive business
If you're investing in Intel in the hopes that it will become a less capital-intensive business over time, then perhaps Intel isn't the stock for you. Chip manufacturing requires a lot of equipment, and if Intel does its job right and is able to capture more unit and revenue share in the overall semiconductor market, then it should continue to spend significantly on chip-manufacturing plants.
But make no mistake: While chip-manufacturing plants are expensive, the returns that a company like Intel can generate on that capital if it has the right products to sell can be quite nice.
