Moore’s Law is Dead, Long Live Moore’s Law

When someone tells you that Moore’s Law is dead, or dying soon, be skeptical. Physics is not the constraint, it’s the economics, and there is a lot of money at stake.

At its core, Moore’s Law is about transistor density. Transistor length divides by two every 18 months. Memory area divides by two every 12. This has been going for 50 years. Things are getting small.

Image: Stéphanie Kilgast, CC

Processing power used to grow exponentially thanks to the processing speed gains from shorter transistors, until power density became an issue. Now, processing power grows by packing more parallel cores into the same area.

[ref: kurzweilai.net]

Money drives Moore’s Law. The $300B semiconductor industry relies on it. The industry has seen time and again: if they keep shrinking chips, new opportunities will arise, birthing new industries and new revenue opportunities. “Build it and they will come”. And we see it: right now Moore’s Law is key to meeting the performance/weight/cost specs we want for the next generations of VR (virtual reality); on the heels of that, AR (augmented reality); on the heels of that, a marriage with smartphones (and a trillion dollar market). Moore’s Law is also key to meeting the power/size/cost specs for the next generations of IoT devices, especially wireless, battery-less sensors in the field.

Conversely, if Moore’s Law stops, the semiconductor industry doesn’t know what to do. So, it does everything in its power to keep it continuing. The industry relies so much on Moore’s Law continuing, hundreds of experts meet regularly to identify challenges, marshal resources to solve them, and update its schedule accordingly. This schedule is called the ITRS (International Technology Roadmap for Semiconductors) — it is quite literally a roadmap. To me, it’s remarkable that Moore’s Law — the backbone of technology for humanity — has a roadmap.

And, the industry has been effective: (silicon-based) Moore’s Law has now run for 50 years. There will always be naysayers saying it will stop. I’ve seen that worry for decades. Engineers always solved it, because of the massive money at stake. We’re currently at 14nm/10nm. Carbon nanotube research points to at least 0.04nm. That’s a lot of breathing space.

There are always new reports that Moore’s Law is dead. Even from the professionals. I saw about four waves of this in the last three decades, while studying Electrical Engineering then being a practicing professional in the semiconductor industry. All reports were all wrong so far. There’s breathing room in the physics, and extreme financial motivation to keep it going.

So, maybe Moore’s Law will die sometime soon. But don’t hold your breath — money talks.