February 2015
30
requires an investment of $1.3 billion for
complex devices but only $356 million for
derivatives [source: International Business
Strategies, Inc. (IBS) (2013/2014)]. The IBS
study shows that companies must spend 650
engineering years to design a complex ASIC
at 28nm. In comparison, a derivative 28-nm
ASIC design requires only 169 engineering
years to develop, a 3.8x reduction.
Assuming ASIC teams are developing new
designs in step with Moore’s Law and are
working on a two-year development cycle, it
would take 325 engineers to complete a com-
plex 28-nm ASIC in those two years. However,
it would take only 85 engineers to complete a
derivative 28-nm ASIC in two years. Or if a
company were to use all 325 of engineers to
develop the derivative as well, they could com-
plete the job in six months (figure 2). Further,
as illustrated in table 1, if we assume that the
initial complex design achieved its 10x revenue
payback of $1.3 billion using 325 engineers, a
derivative design with a smaller addressable
market that is only 80 percent ($1.04 billion)
the revenue size of the initial market of the
ASIC would require just 85 engineers over two
years to develop a product that would garner
a net present value (NPV) that is much better
than the NPV of the initial ASIC design. (NPV
is defined as the difference between the present
value of cash inflows and cash outflows.
The concept is used in capital budgeting to ana-
lyze the profitability of an investment or proj-
ect). What’s more, the derivative would have a
much more favorable profitability index or PI
(NPV divided by R&D money spent) than the
initial ASIC. Even if that derivative addressed
a market half the size ($650 million) of the ini-
tial design, it would have an NPV better than
the initial ASIC, with essentially the same PI.
Increasingly semiconductor companies, as
well as electronics system companies, are
turning to platform strategies as a way to
quickly create derivative products and max-
imize profitability in the face of rising R&D
costs, increased competition and customer
demand for better everything. Platform strate-
gies further reduce product development time,
time-to-market and engineering-hour costs
while simultaneously increasing the profit-
ability of each derivative or next-gen product.
As the IBS study shows, developing derivative
designs is a way for companies to “optimize
revenues and profits.” And developing multiple
derivatives on the same node (in other words,
derivatives of derivatives) using a platform
approach allows companies to further optimize
revenue and profit, as each subsequent design
can benefit from lessons learned in the prior
design, reuse and a more precise understand-
ing of customer requirements.
Two of the biggest business decisions a com-
pany can make when deploying a platform
strategy are actually vital technical decisions:
which one of the many processing systems
will be at the heart of your product platform?
And which silicon implementation of that
processing system is the best for improving
profitability? In a platform strategy, a pro-
cessing system must meet or exceed appli-
cation software and system requirements. It
must be scalable and easily extendable; must
have a large, established and growing eco-
system; and must allow architects and engi-
neers to leverage prior design work. Finally,
it must come from an established, stable sup-
plier with a road map and a track record of
not deviating from that road map or of issu-
ing endless errata. While there are candidates
that fit some of these qualifications, the one
that meets or exceeds all of them is the ARM
microprocessor architecture.
ARM has become the de-facto-standard
embedded processing architecture for just
about anything that isn’t a PC. A vast majority
of electronics systems today that use advanced
embedded processing, from mobile phones to
cars to medical equipment, employ ARM pro-
cessor cores. In particular, ARM Cortex-A9
processor architecture is at the heart of many
types of systems-on-chip (SoCs). It can be
found in ASIC designs typically created for
highest-volume value-added products like
bleeding-edge smartphones and tablets, as
well as in many ASSP designs for companies
wishing to enter established low- to moder-
ate-volume markets that typically compete on
pricing for lack of feature differentiation.
To add differentiation to their products, many
companies create product platforms that pair
an FPGA with an off-the-shelf ASSP based on
an ARM processing system. In this configu-
ration, they can differentiate in hardware as
well as in software, creating a broader feature
set or a higher-performing end product that’s
flexible and upgradable - one that helps them
outshine competitors offering me-too soft-
M
icrocontrollers
& S
o
C
s
Figure 1. The initial cost of developing an IC rises with the introduction of each new silicon
process technology.
Figure 2. Derivative designs reduce time-to-market, development time and costs, making
profitability goals easier to achieve.
