It's Mental

Sustained growth depends on how broadly you define your business—and how carefully you gauge your customers’ needs.

— Theodore Levitt, “Marketing Myopia”, 1960 Harvard Business Review

The general mindset of mainstream FPGA producers goes like this:

• The economics of larger die at smaller feature sizes limits the number of economically justifiable complex digital chip development projects to a small select, number of tape-outs
• As a result, systems companies have a decreasing number of ASSP and even ASIC choices from fewer large semiconductor suppliers… Hallelujah! We get more designs!!
• This gap can be filled by our bigger and bigger FPGAs at smaller and smaller geometries, and these tape-outs can be justified by being spread over many customer designs
• As a result, the “programmable imperative” puts the design burden on the systems customer who uses FPGAs with its own designs and acquired IP

There is a major flaw in this logic: the hardware development teams of almost every systems company are dwindling, with the notable exception of larger Chinese companies. But Chinese companies often lack fresh architectural capabilities. The development teams that remain often become replicators of reference designs that are provided by complex ASSP product companies – sometimes supplemented by an FPGA design, with help from the FPGA company and its IP network. Quite simply, systems companies are decreasingly capable of devising new hardware architectures and implementing them into FPGAs. While it’s feasible and even desirable to do so, the resources to invent and realize fundamentally new designs are no longer available. Incremental new systems and ASSPs get designed, but the breakthrough new stuff does not.

These conditions create two fundamental (and structural) problems in the semiconductor and systems industries:

1. The lack of innovation by larger ASSP producers severely limits the variety and new features in the hardware of new systems
2. Software ends up becoming the primary differentiator of systems producers. As a result, they are increasingly marginalized by Open Source initiatives and ODM hardware producers (see what’s happening in network switches as a primary example).

Where does this leave each of the players in the food chain? 

Grim, huh?

Next installment: how a mental shift can bring it all back home

Copyright @ Synthesis Consulting LLC. All rights reserved.