Quad-core processors are conspicuously absent from the newest networking-processor families. Risk aversion and rising costs have led vendors to neglect value-tier communications processors, forcing customers to choose between newer chips with a few CPUs disabled and older chips.

Based on a large die and housed in large packages, big processors with cores turned off are expensive. On the other hand, “mature” processors will lack the most-recent CPU cores as well as I/O and memory interfaces. Not only does this situation deny OEMs access to faster technology, older processors are unlikely to have the supply longevity a new system design requires.

Behind this trend lies changes in vendors’ risk tolerance and in semiconductor economics. Companies now emphasize returning cash to shareholders, causing management to look for big design wins while containing development costs and risk. As process technology has progressed, tapeout costs have increased. At the same time, the era of each new node significantly reducing the cost per transistor has passed.

If vendors review their processor portfolios, they will likely find some of their best sellers showing their age. Refreshing a successful product should be low risk, even if the vendor can’t identify a single major customer. A basic four-CPU processor with PCI interfaces and Ethernet will find broad appeal in industrial and other systems, not just networking.