Purging Inefficiency

The less obvious cost and expertise benefits of not doing everything yourself.


By Jack Harding

The Value Chain Producer (VCP) segment of the semiconductor industry was recently formalized by the GSA. This is particularly good news for the fabless world, which can benefit from a business model designed to increase quality, efficiencies and access to technology.

Quality is addressed by a VCP through a variety of means, but it all stems from the aggregation of skill sets and the opportunity to apply those same capabilities over and over again. Said differently, the small or midsized FSC simply cannot afford to have the breadth and depth of expertise in-house on the assumption that something may go wrong in a certain area.

To be sure, something will go wrong, but what? Will it be ESD, thermal packaging issues or a signal integrity problem? No one knows in advance. A well-staffed VCP should have all the in-house disciplines required to spot and repair the unpredictable. The result is a better chip in a shorter timeframe.

The efficiency issue is critical. There is too much waste in the fabless supply chain to continue to be successful from either the remaining IDMs or the largest FSCs. The competent VCP purges that waste.

This isn’t margin stacking, either. The VCP can’t sell a chip for an above-market price. It’s irrational. But it can buy the supply chain elements at a lower price. It comes from aggregating purchases and the inherent strategic value to the supply chain when the VCP is calling on the prospect that is not on the monster supplier radar.

More importantly, VCPs save money by doing every day what many firms are now doing once in a while. They’re doing it even less frequently these days than in the past, given there are fewer design starts. If a VCP is good it will lay out a smaller die; yield more die per wafer; enjoy a shorter test time (and/or lower DPPM); diagnose and fix problems faster and, otherwise, develop a better, lower cost chip. It is from these savings the VCP earns its margin–not by selling above market price. In that case everyone fails, including the VCP.

Access to technology boils down to continually learning what really works and when. IP is an obvious piece of this puzzle. It also requires the ability to access the inherent technology built into a given (often advanced node) wafer and, generally, knowing what raw elements make for a world-class part. Of course, this occurs by having the full team, continuous and repetitive opportunities to perfect methodologies and meaningful supplier relationships.

Scott McNealy from Sun once told me, “Strategic partnerships are built upon stacks of purchase orders.” He was right. This fantasy that a company buying a few thousand wafers a year can have a deep, rich relationship with a multi billion dollar supplier is hubris at best, delusion at worst.

–Jack Harding is the CEO of eSilicon