Snapshots of the Exponential Age

Many of us are familiar with the declining cost curves in computing, so well popularised by Moore’s Law. The relationship describes how we get more from less with every given cycle in the chip industry. A single dollar spent today would buy 10,000,000,000 times more computing than a similar dollar spent fifty years ago.

In my book, The Exponential Age / Exponential, I argue that this incredibly creative and world-changing experience of the computer industry is merely a harbinger for similar trends in other industries.

The book is published next week and I wanted to reprise some of my favourite analyses of general technologies with exponential characteristics. That exponential improvements lend themselves to technologies other than computing is a key element of my thesis. 

Cell programming

Between 2017 and 2025, the cost of cell programming, the ability to tweak the outputs of a cell, will have dropped around 100 fold. The potential here is to enlist nature’s power to help meet many of our needs, from new materials to novel therapeutics.

Today cell programming using computers is quite a bit cheaper than doing it by hand. As the price drops toward $10 and below, it’s a technology that could become commonplace. (The data comes via Gingko Bioworks

Thar’ she blows

During the Exponential Age, even the mighty gigantic wind turbine is not immune to the cadence of exponentially. In the US, the levelized cost of electricity generated by land-based wind declined by 40% between 2016 and 2020 (or the equivalent of 12% per annum). Via NREL

More Moore

While Moore’s Law has been a valuable tempo for the past fifty years, its worth noting that there are still chips that will buck the trend. Take Cerebras’s Wafer Scale Engine: it is the first trillion-transistor chip. (The first chip on this graph, the 4004, released in 1971 had 2,300 transistors.) Roughly speaking the more transistors on a single chip, the more powerful it will be. Via Predict

We’ll need more

And we’ll need more performant chips to meet the demands of artificial intelligence. The largest AI models are scaling at 1000x every couple of years. Expect multi-trillion parameter models soon: these enormous models will need millions (perhaps billions) of times more computing than those of just a few years ago. Via Cerebras/Anandtech

It all adds up

Fast-improving technology is creating enormous economic opportunities upon which investors are latching. Median Series C valuations, those typically given to tech companies less than five or six years old, for startups have recently leapt to $1bn up from $89m just five years ago. I roughly date the pivot point for the Exponential Age, the moment of inflection, to between 2013-16. It’s being reflected in the upward shift in these private company valuations. (Caveat: there may be some exuberance that isn’t entirely rational in those prices.) Via Wilson Sonsini via Harry Briggs

Where does this take us

These trends highlight a fundamental shift in human affairs which will have widespread ramifications on the way we order our companies, our economies, our geopolitics and our societies. I explore the technology and its impacts in my new book, coming out at the beginning of September.

Please do take a click and order a copy today.

Cheers,

Azeem