Can you really value SoCs in dollars per square centimeter? Part Deux

Back before the holiday break, I wrote a blog entry titled “Can you really value SoCs in dollars per square centimeter?” and posted it to LinkedIn. The blog post drew a couple of thoughtful comments from members of the LinkedIn Silicon Valley Semiconductor Executives Group that I want to share with you.

First, John Brewer, Jr. who is VP of Corporate and Business Development at SiGe Semiconductor wrote:

“…probably the single largest issue in the semiconductor industry is the treatment of wafer real estate like it is a commodity. This ridiculous notion drives component pricing across the electronics industry. By definition, smaller die = lower price to customer. The electronics industry constantly turns to the semiconductor sector for innovation – functional integration, performance, features, and cost. The electronics industry relies upon the semiconductor sector like a meth addict relies upon his dealer. Yet, semiconductor companies don’t get paid like dealers…

Books are a great analogy for semiconductors. Do we sell books by the pound? Do Stephen King’s books sell at the same price as an unknown author’s book with the same number of pages? Content defines price because the customer is forced to recognize value by the seller.

So, why do people attempt to value SoCs in dollars per square centimeter? Because we in the semiconductor industry have taught our customers that the primary value of our technology innovation is realized by the customer in lower prices. That is, our content is only valuable in defeating competition for the socket, not in creating value that defines price.

As a marketing exec in semiconductors, I can say the semiconductor industry has some of the worst marketing talent on the planet. We do a horrible job of defining value propositions for our products. And we do a horrible job of competitive analysis, often choosing to rationalize an optimistic competitive position instead of realizing an objective one. We’re good at comparing data sheets with competitors, but we’re weak at realizing the customer’s needs, wants and pain points (besides price, of course…), defining products that address them, and standing behind our strategy.

But the real driver on this topic is fear in senior management. Sales management is afraid to compete on parameters other than price because the marketing support is weak. Captive fabs and foundries live in fear of excess capacity; we will use the rationalization of declining marginal cost to justify lowering prices in an attempt to increase fab utilization. In essence, fear of underutilized fab capacity means we fall into the trap of selling solely on price. And we teach our customers that our only true value-added is pricing.

One of the value points of the fabless sector was supposed to be the notion that not having a fab means companies would focus on value in pricing. Instead, many fabless companies continue to obsess on achieving wafer cost reductions through volume increases instead of increasing gross margin through value pricing. The fear of losing a socket that brings additional volume which will incrementally decrease wafer cost overcomes the notion of getting paid for value.

How does this all get turned around? Well, in a different reality, the knowledge gained through effective marketing overcomes the fear of losing the next socket. However, in the universe where we all live, I suspect the “forever cost down” model for pricing in semiconductors ends when all captive and foundry fabs in a market sector are running at full capacity. Then the fear of spending capex on an empty shell and the tooling to fill it overcomes the fear of losing the next socket.

In the RF sector where I live, 2011 promises to put the GaAs industry in that position. All of the captives except Anadigics are full, foundries WIN and AWSC are full, and the wireless connectivity RF front end market continues at 30% CAGR. One would think we’d start seeing pricing increases in 2011 and 2012. But with silicon-based RF front end technologies matching GaAs performance these days, that probably won’t happen. RF-capable CMOS and BiCMOS have plenty of available foundry capacity…”

Volker Politz, Chief Marketing Officer at Renesas Electronics Europe, wrote:

“The simple answer is – yes and no: yes, because that is what is happening in the business process. To some extend understandable as SoCs are not easy to compare, even so they do the same or similar job. And they come with software, middleware and even platform support. The answer should be ‘no’ when taking into account the design effort and embedded IP – it remains a debatable issue, probably alongside with a time constant – yes, for mature designs in high volumes, no, for new , innovative designs in ramp up phase.

Maybe we need to find ways to present the SoC in components – silicon, IP, software, platform, support – and value those seperately.

I am sure this will be an ongoing debate !”

And so the debate continues.

About sleibson2

EDA360 Evangelist and Marketing Director at Cadence Design Systems (blog at
This entry was posted in EDA360, SoC Realization. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s