Chips Ahoy! – Moore’s (Specialized) Law

Moore’s Law is one of the most important drivers of everything from Tech stock valuations to the billions of dollars that continue to flow into venture capital. That semiconductor speed/dollar will double about every 2 years means smartphones will continue to get both better and cheaper, cars will become autonomous, and AI will grow smarter and more productive. Without future logarithmic growth in chip efficiency, most of these expected developments would grind to a halt or be tossed altogether.

However, even Gordon Moore (who co-founded Intel in 1968), might be surprised that his old firm is not doing all – or even most – of the heavy lifting to keep his namesake “Law” rolling along. Intel and competitors like NVIDIA are seeing their clients design more of their own chips in a bid to create the next wave of tech-enabled disruption. This is a remarkable shift and highlights the importance of physical hardware, long thought to be a commodity, to the next phase of growth across a wide array of products and services.

We gathered up some recent examples across a range of applications to give you a sense of this shift:

Tesla. On their recent earnings call the company discussed its development of custom-designed chipsets to enable truly autonomous driving. Press accounts called this a “tacit admission” that their current architecture (an NVIDIA product) wasn’t powerful enough for next-generation applications.

Read more here: https://arstechnica.com/cars/2018/08/tesla-says-its-dumping-nvidia-chips-for-a-homebrew-alternative/

Apple. The recently minted $1 trillion company is planning to shift away from Intel processors as early as 2020. The idea here is to allow Apple to control its new product cycle for computers, rather than being tied to Intel’s upgrade calendar.

Read more here: https://www.bloomberg.com/news/articles/2018-04-02/apple-is-said-to-plan-move-from-intel-to-own-mac-chips-from-2020

Facebook: Last month the company hired one of Google’s lead chip developers. Their goals are to develop both proprietary server and consumer technologies. Right now the company uses 3rd party modified NVIDIA chipsets to run the AI algorithms that police its platform.

Read more here: https://www.theverge.com/2018/7/13/17570066/facebook-custom-chip-design-shahriar-rabii-hire-google-ai

Google, Amazon, Alibaba: These three companies really kicked off the trend towards proprietary chips late last year. In all three cases, the primary focus is on AI applications to power everything from online search to voice powered devices (Alexa, Google Home) to shopping.

Read more here: https://www.cnbc.com/2018/04/21/alibaba-joins-google-others-in-making-custom-ai-chips.html

Bottom line: Moore’s Law is alive and well, thanks to specialization. As chip design becomes more application-specific, it can continue to improve even as more generic designs bump up against natural physical limits. It wasn’t too long ago that tech watchers were concerned that only quantum computing could keep Moore’s Law in place. Now, it seems to have a clearer runway.