Intel (NASDAQ:INTC) is known as a company with a heavy reliance on the health of the PC market. However, the company's Data Center Group -- which sells server processors and related products -- has been a phenomenal business for the chip giant. While some worry that ARM (NASDAQ:ARMH) based vendors will start to gnaw at Intel's sales here, this business is quite robust and has a large moat around it.
Top to bottom, Intel's got you covered
The "data center" doesn't consist of one single use -- these processors and related peripherals and interconnects are used in everything from high-performance computing (think hard-core number crunching such as physics simulations) to serving up webpages. These demands require different types of chips with various performance levels and integration, so having a top-to-bottom stack of products is key for a solutions vendor.
While the product stack appears unmatched by any other company in the industry, there have been rumblings of large data-center operators such as Amazon.com (NASDAQ: AMZN), Microsoft (NASDAQ:MSFT), and Google (NASDAQ: GOOG) tinkering with the idea of designing their own processors for their specific data-center needs.
Warning: Competing with Intel is really tough
The design and validation of a world-class server processor that could even have a shot of going above and beyond what Intel offers is quite expensive. Google and Amazon, both of which have no problems investing billions in research and development, could conceivably and over the long haul develop their own solutions. However, there are a couple of issues with that:
- Developing world-class chips is expensive. To develop a complex processor on the upcoming 16-FinFET process node from Taiwan Semiconductor (NYSE: TSM) costs, according to semiconductor expert Handel Jones, about $450 million-$500 million . For custom chips to be viable, the vendor would need a pipeline of at least two to three generations of designs going on at any given time.
- Beating Intel's manufacturing lead will be tough. Intel has a clear manufacturing lead. This means Intel can pack more chips economically in a given area, and that that the actual transistors -- the building blocks of the chip -- are higher performing.
On top of these economic concerns, the bigger question is just why these companies would want to reinvent the wheel when there's a better option.
Semicustom designs make more economic and technological sense
Intel's processors are already well suited for the majority of data-center applications, but there will be specialized uses that a more customized chip may suit better. For example, Microsoft's new server technology, known as Catapult, , uses FPGAs alongside Intel Xeon processors (think of FPGAs as chip that can be "reprogrammed" to perform specific functions on the fly) in order to see an up-to 40 times' speed increase in delivering Bing search results.
In fact, probably not coincidentally, Intel just announced at the GigaOM Structure conference that it would offer a line of Xeon processors with an FPGA (from an undisclosed vendor) built together on the same package . According to Intel, for the right workload, such a solution can increase speeds by 10 times over traditional chips.
However, this is just one part of a larger scheme. As Intel transitions to a system-on-chip company, it will be able to more swiftly integrate third-party IP blocks into its Xeon processors. So, for example, if Google designed a dedicated search engine acceleration block, instead of designing the rest of the processor or system-on-chip around it, it could simply work with Intel to integrate that block onto a Xeon.
While the impending threat of the various ARM vendors encroaching on Intel's business makes for fun headlines, and while the idea that every company with deep pockets will try to build its own chips, the economics of semiconductor manufacturing and the attendant design costs simply don't make sense, even if you're a fairly large customer like Google, Microsoft, or Facebook.
Intel's position is strong, and though no business is invulnerable, the company seems to be making all of the right moves to position itself for continued market share leadership in the data center.
Ashraf Eassa owns shares of ARM Holdings and Intel. The Motley Fool recommends Amazon.com, Apple, Google (C shares), and Intel. The Motley Fool owns shares of Amazon.com, Apple, Google (C shares), Intel, and Microsoft. Try any of our Foolish newsletter services free for 30 days. We Fools may not all hold the same opinions, but we all believe that considering a diverse range of insights makes us better investors. The Motley Fool has a disclosure policy.