An historic technological evolution is underway, with powerful themes that are transforming business and society while offering unique opportunities for investors, according to Ark Investment Management LLC.
The New York-based firm manages US$11.1-billion in client assets and aims to identify investment opportunities by focusing on companies that enable and benefit from disruptive innovation.
“We look at where technology is going, not where it is,” says James Wang, an internet analyst with the firm. “Our job is to find the companies we believe are going to be trillion-dollar companies tomorrow.”
Ark recently published its annual list of “big ideas” that fit this bill and there is no shortage of radical change among the 11 it identifies. It includes aerial drones, 3D-printing, driverless cars, advances in DNA and genome sequencing and the rise of bitcoin.
Yet, Ark believes the biggest idea is deep learning, a term referring to software that learns as it goes without human help.
The software can see, hear and understand natural language at near human levels of accuracy. That includes software that serves up ads when we go to a website, robots that help doctors create surgical plans based on thousands of similar operations, or Google Home, which gets better at answering questions.
Ark manages five actively-managed exchange-traded funds (ETFs) as a subadvisor for Emerge Canada Inc., which launched the ETFs on the NEO Exchange in Toronto last year.
Mr. Wang, who has a degree in computer engineering, explored the implications in a recent interview. Here’s an edited and condensed transcript of that conversation:
Why is deep learning having such an impact?
We believe it is the most important foundational technology since the invention of the internet. It’s a completely new way of thinking about software.
Traditional software is written by programmers. It can’t do things like see what’s in a picture, understand what’s said by a voice, comprehend a sentence, that kind of stuff.
Deep learning is software that learns from the data it collects. This makes it able to learn human-like functions, such as vision, caring, language understanding, strategy and so on.
Do you have an example that comes to mind?
Sure. Our most powerful cognitive function is probably vision. We can see the world and understand 3D relationships. We know where we can walk and where we can’t walk. We know what’s a cat, what’s a dog and the different breeds.
Until 2012, we couldn’t write software that could differentiate between a cat and a dog with any level of accuracy. It wasn’t possible. You were using math to write equations to describe these differences.
In 2012, deep learning emerged. Now, you can show a computer thousands of pictures of cats and dogs and the software has achieved human-level accuracy for telling which is which. It’s used across billions of devices, in everything from your iPhone camera, to surveillance videos, to tagging friends on Facebook.
So what’s next?
A conversational computer, which is a computer without a screen. You talk to it rather than look at it and click. This wasn’t possible five years ago.
What else is coming?
Self-driving cars. Alphabet Inc. (GOOGL-Q) and Tesla Inc. (TSLA-Q) have been working on this technology for years. Previous efforts didn’t work because the sensors couldn’t understand the world with enough accuracy.
Now, the sensors can use video to understand what’s happening from a 360-degree point of view. Humans are not able to do this.
How far off are driverless cars?
We expect they will be on the road as of next year, in 2021. Waymo vehicles have already travelled more than 20-million miles autonomously. (Waymo LLC is Alphabet’s autonomous driving subsidiary.)
Where are the investible opportunities?
Tesla. Waymo is also doing great work. Other companies include DiDi Chuxing Technology Co., a private company based in China, and Nvidia Corp. (NVDA-Q), which supplies the most powerful microchips for deep learning.
What about electric vehicles?
We believe electric vehicles (EVs) will be cheaper than gas-powered cars within four years. The biggest cost of an EV is its lithium-ion battery – and they are being produced at ever-increasing scale. As production increases, costs decline, which should drive price declines in the cars themselves.
We believe that the average price of an EV will fall to about US$33,000 this year, which is the price of a mid high-end sedan. By 2022, we think there’s a crossover when an EV costs just a little less.
There’s going to be a seismic shift when that happens.
What horizon should investors have?
It takes time for these technologies to mature, so five years or more. The best way is a portfolio-based approach. Our portfolios typically have 35 to 50 companies. And while we think each of them could do well, in the end, a few should do spectacularly well.
Adam Mayers is a contributing editor to the Internet Wealth Builder newsletter.