The Diversification Point, or what comes after Moore’s Law

28 August 2018
28 Aug 2018
Southside Berkeley
10 mins

For the first few years of the modern smartphone, in the late oughts, there was only one kind of a smartphone: the one that looked like the iPhone, with a multi-touch screen and applications and cameras that were just acceptably good enough for daily photography.

In that era, there was also only one kind of a good smartphone: the one with good battery life, responsive software, a big screen, and good cameras. If you wanted the best smartphones, you had to pay up, because the more expensive, the better. There was only one kind of smartphone, and everyone, from Apple to Motorola to Huawei, was aiming to seize the same market with the same kind of product.

Fast forward a decade, and there’s no longer the one true blueprint for a popular smartphone. What was once a luxury category of $600+ devices has diversified and broadened rapidly into a dizzying array of niches and styles and consumer priorities, from $100 smartphones for kids or the developing markets to the classic, affordable iPhones and Samsung Galaxies, to the still higher-end, status-symbol iPhone X and Vertu. Even beyond that, there are phones that are ruggedized, made for longer battery life above all, made for gaming, made for video production, and the list goes on and on. We’ve done away with the One True Smartphone model, and hundreds of companies are targeting hundreds of niches with thousands of devices with different priorities and focuses.

So what happened here?

At some point in the story of the modern smartphone, the technological building blocks that we need to build the smartphone — cellular modems, low-power processors, high quality displays and cameras, and so on — became cheap enough that it made less sense to make one kind of a component and mass-produce, and instead made more sense to make different kinds of devices for different types of consumers.

At this inflection point, the fervent rush of tech companies to race to the pricing bottom and commodify the components switched, to become an exploration and diversification of what’s possible when the basics of the technology is worked out.

This inflection point, which I’ll call the diversification point, isn’t unique to consumer electronics. The automobile industry, the commercial airline industry, the software services industry have all passed through the inflection point of diversification in the process of evolving from high-tech, low-volume markets to mature industries that reach billions of people.

More importantly, whether or not a technology can hit this inflection point determines whether the technology can become commercialized at scale sustainably.

First, things get cheap

The first trend in commercializing any new tech is to make it affordable. This happens because of two forces, one of push and one of pull.

The first time something becomes possible, it’s a feat performed only once, usually in a research laboratory, with a specific budget outlined just to produce one instance of the piece of technology. But once something becomes possible, the logical next step is to reproduce it — to make it happen once more, then a few more times, than a hundred more times, until it becomes reliable. This push towards reliability and consistency means that any useful technology, even if niche, tends toward scale. Few innovations are useful performed once, and to make things happen many times, we need to figure out how to make it cheaper to do so. This is the push force that makes new technologies more affordable.

But the push force can only get you so far. The more powerful force in democratizing technology is the pull force of business. Once the demand for some piece of tech, like mobile phone processors, reaches a certain level, by virtue of being able to sell more, businesses that build the technologies build more. Almost always, when you build more of something, you spend less per unit. And the price goes down again. This is the pull force of business.

In the 1990s, the United States government set out to fund a watershed project in molecular biology. Though it ended up being financially bloated and ran straight over the planned schedule, the Human Genome Project provided the initial kick in the process of making machinery involved in gene sequencing more affordable for companies and laboratories. Genome sequencing technologies are still very much in the “push” phase, with price declines funded mostly by advancements in technologies and only a little bit by the rising demand for consumer genome sequencing itself. But it may be close to the shift, as companies like 23andMe make consumer genome sequencing more accessible.

One of the most dramatic shows of price declines by consumer demand, the “pull”, is in solar cells and solar panels. In a matter of over a single decade, the price of solar panels have plummeted orders of magnitude while the efficiency and reliability have gone up. Solar panels are solidly in the “pull” phase of price declines, where consumers — still mostly large, electricity producers — are looking more closely at solar power as an alternative to fossils and hydroelectric power.

Lastly, let’s look at an industry long into the pull phase of price declines — personal computing. The switch from push to pull price decline in personal computing occurred early in the 1980’s, as machines that were powerful enough to be useful to the average people grew up from being novelty, early-adopter tools to more accessible home gadgets. As almost everyone had one, or two, or even three computers in their homes, manufacturers churned out tens of millions of chips and screens and circuit boards and hard drives a year, and the scale pushed the price lower and lower.

Then, things get diverse

Once it becomes sufficiently cheap to make a piece of technology like a lithium battery or solar cells or energy-efficient smartphone chips, something very interesting happens. When a technology becomes cheap enough, rather than making one type of product based on the technology, it becomes more reasonable to make multiple variations of it, targeting different kinds of buyers. Because there’s significant scale, even if individual parts are more expensive as a result, the growing market push the industry from a “make one, sell many” mode into a “make one for everyone” mode.

This is when things get diverse.

In the inception of touchscreen smartphones, there was only one kind of smartphone microprocessor. It had a single CPU core, a power-constrained graphics chip, a lowly few million transistors, and worked just barely quickly enough to power the smartphone user interfaces of the time. It was expensive, hard to manufacture, and so companies made very few variations of these chips, and they made up the cost by selling many of the same chips.

But the industry hit an inflection point of scale, where it made more sense to produce chips that were not the absolute peak of possible technology. It became possible to run smartphone software on less than top-of-the-line specs. The screens you could get for half the price of the best ones were still good enough to use, and so on. And so the industry moved away from making a single type of phone and selling it to everyone, and began to diversify, making many different kinds of smartphones and selling it to even more people who had different priorities and preferences.

Another example of technology in the “get diverse” phase is the hybrid car. For the first few years, while the technology behind a gas-electric hybrid drivetrain was being perfected and made reliable, there was only one type of a hybrid car — the slightly awkward, slightly futuristic, Prius-esque compact car. But as hybrid technology matured and the core technology was scaled out, hybrid evolved from the distinguishing feature of a single product in a car maker’s product line to an aspect of almost every modern family car. While the tech was being built out and made cheaper, there was a singular vision of the product; but once the initial technical hurdles were resolved and hybrid became something of a commodity, the category of hybrid cars diversified, to SUVs, trucks, compact cars, and even record-setting supercars.

And then, design innovation takes over

The significance of the inflection point of diversification isn’t merely that the market expands and the technology matures. The most important consequence of an industry that grows diverse is that technical innovation exits center stage, leaving room for the next entrant: design innovation.

My favorite example of this is the relatively new product category of “smart speakers” — Amazon Echo, Google Home, Sonos, and the growing family of internet-connected speakers that listen to voice commands and deliver prompt actions and answers.

The smart speaker isn’t the product of arduous technological innovation. All the technologies we needed to build the first Amazon Echo — Wi-Fi modems, low-cost mobile chips, accurate voice recognition, and online services with a wealth of personal data and intelligence — were present and affordable by the time the first Echo device rolled around. The innovation wasn’t in what was technically possible, but in how the commodity building blocks of the product were pulled together into a novel design. This design innovation was only possible because the rise, commodification, then diversification of mobile phones had left behind a mature network of supply chains, software, and engineers with the knowledge to quickly build products with the existing technologies. Design innovation follows technological innovation, and in between was the scale-up and diversification of the underlying technology.

But smart speakers aren’t the only new products to come out of the proliferation of core technologies behind smartphones. Everything from electric skateborads to smart lightbulbs to wearable fitness trackers — these were all design innovations built on the fact that the basic building blocks of these smaller devices became affordable and plentiful as a result of the smartphone industry’s massive diversification into dozens of price points and use cases.

Once a technology becomes both sufficiently cheap and sufficiently diverse and adaptable, it’s the designers’ turn to innovate; and when designs hit a wall of technical limits, engineers step in again. And the cycle continues.


Where are you in the cycle?

If you’re working on commercializing some piece of technology, it helps to know where you are, so you can realize what kinds of push you need and how close you are to getting there.

The last important thing to know about the inflection point of diversification is that, once a new technology hits this point, it’s very rare that the technology “dies out” in the commercial market and disappears back into the labs. Technologies that are just now in the verge of this inflection point, like virtual and augmented reality and precise gene editing, have been through multiple “this is it!” moments before, and many times, they’ve disappeared from the commercial world back into the labs, until the next best iteration could be put to the test.

If you’re working in emerging technologies, it’s useful to think about your position with respect to this inflection point in the adoption curve. Rather than a simple downward curve of lowing prices, well known from Moore’s Law, in the long tail of the adoption curve, we need to bring in a second trend, that of diversification. In the diversification phase, the price is no longer the key variable, but the diversity of possible designs that could be produced from the basic, commoditized elements of technology.

This trend of diversification is what made the transition from simple, blocky, desktop-mounted personal computers to mobile phones and smart speakers possible. It’s what leads the transition from a technical innovation-focused industry to one that thrives on the diversity and possibility of design innovation, and I think it deserves a much closer inspection that we give it today.


If you enjoyed this piece, you might also enjoy my next post, Hustling efficiently.

Have a comment or response? You can email me.