Jerry Sanders' famous quote from the early days of the foundry industry is rightly mocked today, but a slightly modified (and more politically correct) version may still be applicable.
Another idea has grown up alongside the trend of
crating up fabs for the auctioneers. Brand recognition is important in
any business, but at some point, marketing experts convinced us that a
brand can take on a life of its own. That's undoubtedly true in consumer
markets. Regular folks pick either the blue brand or the red brand even
if the soap inside is pretty much the same.
For an IC industry moving toward the building block model for making
chips, fabless vendors may look to the automobile sector. The car
manufacturers seem to do little besides assemble third party components
and advertise themselves more as a lifestyle brand. A few brake system,
engine and transmission companies supply everyone in the business.
What's under the hood depends little on the badge on the other side.
Unfortunately, chips are a different story, despite the reuse of
consumer advertising approaches in the semiconductor industry. Engineers
are a lot more sophisticated than average consumers.
This was highlighted by Rick Merritt, covering the fabless story back in
2008, with an interview of Infineon's CEO.
According to Wolfgang Ziebart, "You don't approach your customers with a
marketing department that does studies and finds clusters of demand and
establishes product lines to address them. That's old thinking. Today
you have to engage with your customers in a much more intimate way." Not
only do you need to engage to understand product requirements to create
the chip they need, you also need to deliver more value than the
collection of third party IP determined by the product specification.
You can engage with customers, help them implement your design, and keep
smiling through the whole transaction, but if your chip carries a price
premium, you won't win the socket. Soon enough, a small group of design
engineers at the OEM will be able to create exactly the chip they want.
The barrier to entry becomes small for competitors as well as your
customer. If you need an SoC for your cell phone platform, why not
approach it the same way you approach the board-level design? Fabless
chip suppliers could end up as little more than middle men.
The fabless ecosystem levels the playing field. But finding a way to
convince customers to pay a premium is the name of the game. Still, some
semiconductor companies have found a way. Consider Intel. How do they
do it? Investing in their own factories and getting to each new logic
process node a year or more ahead of their competitors probably has
something to do with it. It doesn't sound like process technology is
commoditized to me.
The sexism inherent in Jerry Sanders' comments about real men and fabs
made his quote good fodder whether or not you thought the idea of
in-house chip manufacturing was just as anachronistic. But some wisdom
can still be extracted from Sanders' comment if we simply update the
language. It takes real leadership to create a long term strategy and
then execute it for the good of all involved—employees, customers and
share holders. Staying in the IDM game and building your own fabs are
gutsy gambles these days. Comments about real men aside, let's hope that
at least Jerry Sander's maxim, "People first, products and profit will
follow," is not considered as outdated as his comment on fab investment.
Maybe the two Sanders lines could be shortened to one:
"Don't take the easy way out."