After 17 years in the semiconductor (silicon) industry, I am switching to software. Why? Many reasons, but one is worth blogging about. I believe the long-term trend is economic contraction in hardware (silicon), and significant growth in software.
The trends I see are secular trends — a fancy way of saying very long-term. In fact the trends are just a continuation of the trends of the last few decades. What ever you call it — hardware, silicon, or electronics — continues to be commoditized:
- DRAM becomes a commodity — 1980s.
- Storage (hard drives) become commodities — late 80s, early 90s.
- Chip-sets — late 90s.
- Low-end graphics: early 2000
- Other sub-systems: Ethernet, audio, USB, PCIe, cable-modems, etc. Early 2000s
- 64-bit computing — 2004
- Routers, Cable Modems — late 2000’s
- Mid-range graphics — late 2000’s, early ’10s
- SSDs — 2014
The transition from premium product to commodity is a continuum. A premium product does not become a commodity overnight. The premium just gradually decreases for a given class of products. Another was of saying this is that profit margins gradually erode as competing products accelerate the “race to the bottom.”
Today some of the last hold-outs — high-performance, high-reliability computing, and high-end graphics — are showing early signs of diminishing premiums. Some analog and mix-signal silicon commands premium prices too. However, the overarching trend is towards lower profit margins.
This tectonic shift in silicon margins will create winners and losers. Consumers, technology users, and software vendors will tend to be favored. Hardware suppliers will tend to face headwinds. Similarly, those who work in software-related fields will tend to benefit, while those working in hardware-related fields will tend to become stuck in a low-growth environment.
Commoditization is not the end of the road. After all, oil is a commodity that makes billions of dollars per year for companies like XOM. It simply means that gross profit margins for silicon are likely to fall from 60% to perhaps 30% in the next 5-10 years.
Overall, I expect silicon volume (units) to keep increasing, silicon revenue to modestly increase, while silicon profit and profit margins decrease. The mantra of “silicon everywhere” is misleading, while the model of “cheap silicon everywhere” is quite apt.
Conversely, I see a brighter future in software, app, and web development. Online retail revenue was about 6.5% in 2013, but the upward trend is strong. Hosting E-business in the cloud will become cheaper as hardware performance increases while hardware cost decreases (and hardware performance/Watt improves). In this environment of healthy growth, software will differentiate; content will differentiate; and hardware will simply serve.