Intel believes its strength in semiconductor manufacturing is a competitive advantage that will help it stay on top of the processor market, its CFO said this week at the company's primary US manufacturing facility in Hillsboro, Oregon.
Industries of scale
The past few years have been rough for the world's largest semiconductor maker as it overhauls its design and marketing strategy away from fast single-core processors to slower chips with more than one processing engine. But its manufacturing operation continues to move forward in shrinking the average size of its transistors every two years, or scaling, giving Intel a cost advantage that its rivals can't match, according to Andy Bryant, Intel's chief financial officer, speaking before press and analysts.
"To be perfectly candid, a lot of it I don't understand," Bryant joked. But what Bryant does understand as Intel's financial chief is that the cost of making a single transistor continues to decrease even as the cost of building new chip fabrication facilities, known as fabs, soars past $5bn (about £2.9bn).
Chips are etched onto circular silicon wafers. Intel and the rest of the semiconductor industry tend to reduce the average size of the transistors on those chips every two years, and increase the size of the silicon wafer every ten years.
Decreasing the size of the transistors allows more of them to be crammed into the same size chip, generating a more powerful chip within the same space. And when the size of the wafer itself is increased, more chips can be cut from that wafer, reducing the cost to manufacture an individual chip.
At present one transistor costs one-billionth of a dollar to manufacture, Bryant said. By comparison, when Intel was founded in the late 1960s, the cost of manufacturing a transistor was more than $5, he said.
That dramatic reduction in costs is the result of billions upon billions of dollars in research and development expenses, Bryant said. This year alone, Intel will spend around $11.1bn (£6.4bn) on capital expenditures and research and development combined, it said when it announced third-quarter financial results.
And that underscores Bryant's main message for the day: that chip companies need to ship huge volumes of processors to maintain that level of investment and keep the economics of chip manufacturing in line.
Intel is facing its strongest competition in years from AMD in the PC and server processor markets. AMD has managed to gain market share with a more efficient processor design and by making the most of its more limited manufacturing resources.
Intel has changed its design strategy in recent months, and will start to ship processors that analysts believe will close the performance gap with AMD's chips in the second half of next year. But Intel is well ahead of AMD when it comes to introducing new manufacturing technologies that reduce operational costs. By the end of next year, Intel will have three high-volume fabs building 300mm silicon wafers using a 65-nanometre processing technology while AMD will have just one.
This advantage will allow Intel to plough dollars into researching new technologies that will help it continue to shrink the size of its transistors, and the company is touting this as something that will help tip the competitive balance in its favour. By the next decade, features within transistors will approach the atomic level, and no one is exactly sure what to do to make things smaller when that point is reached.
While it might be hard for end-users to grasp the minutiae of chip manufacturing strategies, it's not hard to understand that Intel's manufacturing prowess will allow it to plunge billions of dollars into future research and development spending that could produce game-changing products, said Cody Acree, managing director of equity research firm Legg Mason Wood Walker.
Intel believes its cost model will give it the financial flexibility to take on the challenge of keeping Moore's Law alive into the next decade, Bryant said. Coined by Intel co-founder Gordon Moore, this law states that the number of transistors on a chip will double every 18 months, and has been the driving force behind the decreasing costs of electronic gear.
"If the economics look like this, people will find a way to [extend Moore's Law]," Bryant said.