IEEE Spectrum April, 2015 - 33

DATA sourcE: vLsI rEsEArch

But this logic was flawed. It turned out that making a chip
with eight transistors yields a fraction of operational chips
similar to what you'd get by making eight stand-alone transistors. That's because the probabilities aren't independent.
Defects take up space, and many types are distributed randomly, like paint splatter. If two transistors are placed close
together, a single transistor-size flaw can take out both devices.
As a result, putting two transistors side by side carries about
the same risk of death by defect as one transistor by itself.
Moore was convinced that integration would ultimately prove
economical. In his 1965 paper, as evidence of the integrated circuit's bright future, he plotted five points over time, beginning
with Fairchild's first planar transistor and followed by a series
of the company's integrated circuit offerings. He used a semilogarithmic plot, in which one axis is logarithmic and the other
linear and an exponential function will appear as a straight
line. The line he drew through the points was indeed more or
less straight, with a slope that corresponded to a doubling of
the number of components on an integrated circuit every year.
From this small trend line, he made a daring extrapolation:
This doubling would continue for 10 years. By 1975, he predicted, we'd see the number of components on an integrated
circuit go from about 64 to 65,000. He got it very nearly right.
By 1975, Intel, the company Moore cofounded after leaving
Fairchild in 1968, was preparing charged-coupled-device
(CCD) memory chips with some 32,000 components-only a
factor of two off from his thousandfold prediction.
Looking back on this remarkable paper, I'll note a few details
that are often overlooked. First, Moore's prediction was about
the number of electronic components-not just transistors but
also devices such as resistors, capacitors, and diodes. Many early
integrated circuits actually had more resistors than transistors.
Later, metal-oxide-semiconductor (MOS) circuitry, which relied
less on nontransistor components, emerged, and the digital age
began. Transistors dominated, and their number became the
more useful measure of integrated circuit complexity.
The paper also reveals Moore's focus on the economics of
integration. He defined the number of components per chip
not as the maximum or the average number of components
but as the number for which the cost per component was at
a minimum. He understood that the number of components
that you can pack on a chip and the number that makes economic sense are not necessarily the same. Instead, there's a
sweet spot for every generation of chip-fabrication technology.
As you add more components, you drive the cost per component down. But past a certain point, attempting to pack even
more transistors into a given space will raise the possibility
of killer defects and lower the yield of useful chips. At that
point, the cost per component will start to rise. The goal of
integrated circuit design and manufacturing was-and still is-
to hit this sweet spot.

10 b

Price Per transistor
(billionths of $1)

10 20

100 m
10 16

10 12

10 k
100

10 8

transistors
made Per Year

Years

1955

2014

Transistors,
by the
Numbers
In 2014, semiconductor production facilities made
some 250 billion billion (250 x 1018) transistors. This
was, literally, production on an astronomical scale.
Every second of that year, on average, 8 trillion
transistors were produced. That figure is about 25
times the number of stars in the Milky Way and some
75 times the number of galaxies in the known universe.
The rate of growth has also been extraordinary.
More transistors were made in 2014 than in all the
years prior to 2011. Even the recent great recession
had little effect. Transistor production in 2009-a year
of deep recession for the semiconductor industry-
was more than the cumulative total prior to 2007.
The collective pursuit of Moore's Law has driven
this growth. For decades, manufacturing innovation
and simple miniaturization have enabled engineers to
pack more capability into the same area of silicon. The
result has been a steady decrease in manufacturing
cost per transistor (transistor price, which is easier to
track, is plotted above).
This steady, predictable decline in prices was a selfreinforcing gift. Because electronics manufacturers
could depend on Moore's Law, they could plan further
ahead and invest more in the development of new and
better-performing products. In ways profound and
surprising, this situation promoted economic growth.
It has been the ever-rising tide that has not only lifted
all boats but also enabled us to make fantastic and
entirely new kinds of boats. -Dan huTCheson
SPECTRUM.IEEE.ORG

nORTh aMERICan

aPR 2015

http://SPECTRUM.IEEE.ORG

Table of Contents for the Digital Edition of IEEE Spectrum April, 2015