Benedict's Christmas Present...ation: The End of the Beginning

Every year Benedict Evans of a16z does a "big" presentation on the future of mobile and digital media. I put the word "big" in quotes because the other annual take on the internet is Mary Meeker's Internet Trends. Here is the 2018 version which comes in at almost 300 slides. The video below is Benedict's presentation this year called The End of the Beginning. There doesn't seem to be a copy of the presentation alone online, but the video version is just 24 minutes long, so well worth your time. EDA and semiconductor are not internet businesses, of course, but ultimately what happens in these markets drives electronics, and electronics drives semiconductor, and EDA is the point of that needle.

I will cover some of what he says in this post, but for an investment of 24 minutes, you should see it for yourself. It's at the end of this post.

The End of the Beginning

What Benedict means by the end of the beginning is that "the access story has finished, but the use story is just starting to begin." If you look at the above chart, we've connected most people to the internet (largely through mobile outside the US) and we will connect the rest. Be e-commerce penetration is still low. Here's an amazing statistic I bet you didn't know. Gas-stations are a big market, and they are probably going to disappear over the next 10 to 20 years. They are bigger than US e-commerce. I'm not implying that gas-stations are going to be driven out by e-commerce, it's electric vehicles which will do that, but it shows you just how far there is to go, which is Benedict's theme that this is the end of the beginning. Or "most people are online, but most of the money is not."

The above chart steps back even further. On the left is retail spending, with e-commerce the little orange bar at the bottom. On the right is the rest of consumer spending (all the numbers are in $T by the way). So in the big picture, e-commerce is about 4% of consumer spending.

Next, Benedict took a look at advertising and marketing. We have very arbitrary splits between the two: if you pay for placement on Amazon it's advertising, whereas if you pay for placement at Walmart it's marketing. So in some ways you need to step back and look at "money we spend to reach a customer" and then you need to add in not just marketing, but start throwing in retail rents and shipping costs, since these go away for digital products.

The opportunity is global. Just as manufacturing was invented in the UK, and then the mantle passed "to another big country on the other side of the ocean", and now has moved on mostly to China. Computing is the same. In 1993 about 3/4 of all computing capacity was in US and Europe. Now (counting smartphones as computers because they are) China has more subscribers (so more computers) than US and Europe combined.

If you live in the US, especially in Silicon Valley (or, in this case, maybe Seattle) it is easy to get the impression that e-commerce penetration is really big here. But as you see from the above chart, the US is a bit of a laggard. Note that these are penetration percentages, so the fact that China has over a billion people is not what this story reflects. As a Brit (as is Benedict) I was surprised that e-commerce penetration is the UK is almost twice what it is in the US, and in South Korea (SK on the chart) it is higher still. So e-commerce started as a US story (ironically, led by a company named after a river in the country on the very right-hand end of the chart). But it's really not a US story anymore.

We are switching from thinknig about a few hundred billion of US e-commerce to worldwide total consumer spending of something like $40T. Investment has followed with unicorns ($B valuation companies) outside the US, most of which you've never heard of. Venture capital, which started in silicon valley, spread to the US, and is now spreading worldwide.

The Next 20 Years

In the first 20 years of the internet we did things that worked with low penetration (few users) and small amounts of capital, and low touch, so people would buy them even though they weren't really comfortable buying online. Amazon started in books for a good reason—books are the same in a way that, say, clothes are not. We have high penetration now, so we can use lots of capital to get to everyone. People are used to buying high touch products like clothes and shoes online. 

A good example is Yelp and Doordash. Yelp is a low-capital information arbitrage business. Doordash has to get you a hot meal in half-an-hour. So it is much higher capital, much higher touch, and it is a "full stack" company with a lot more than just a bunch of servers and some engineers.

Benedict splits retail up into retail as logistics (Sears (catalog era), Walmart, Amazon) and then there is retail as taste (boutiques, big cities). "We've never been able to scale retail as taste outside big cities and outside small shops. But now we are moving to tastemaking and have more and more things that are not just a specialized version of FedEx." 

Online groceries is another big market that requires a lot of capital. It also requires a completely different logistics system, but it's "worth it because groceries are so much bigger than e-commerce." Not surprisingly, China is big, and so is US (blue bars). But if you look per-capita, the US has a lot of catching up to do (China too) with the leaders being Japan, UK, France, and Korea.

Entertainment is in transition. Benedict had a great quote from Jeff Bewkes, CEO of Time-Warner in 2010, who was asked if Netflix was a threat. He said it was like worrying about the Albanian army taking over the world. "I bet he's keeping an eye on the Albanian army now."

But TV is changing much more than that, since the definition of TV is changing. Youtube games and Twitch are both larger than Netflix, HBO, or ESPN. The record global opening weekend was not the latest Avengers movie, but Red Dead Redemption 2, a game. Watching people play games is catching up with the size of watching people play traditional games like basketball or baseball. But the revenue has yet to catch up so you can expect to see a lot of changes there. Traditional TV viewing is going down:

The Tech industry has spent 20 years trying to get into the living room, but it didn't do it with TVs, it did it with mobile phones.

Drug discovery and some other aspects of medicine are increasingly using AI, but perhaps more importantly, if you really need to do a trial on 10,000 people then social media offers totally new ways to get them, and wearables (Apple watch, Fitbit etc) a totally different way to monitor the trial. It's also interesting to think beyond drug discovery to healthcare, but another level past is to look at the value of not getting sick, of not dying. We can start to think about addressing this sort of problem now.

Boss Mode

Benedict's analogy is that it's like we've gone through all the easy monsters in the digital world, and now we've reached the boss: "the giant robotic brain with a Gatling gun." These are harder markets to address than we addressed in the past.

We started with blind alleys like AOL and the information superhighway, but then we got permissionless innovation: you didn't need to ask AT&T to put up a website. That led to the explosion, but we needed an organizational layer and so we got Google and Facebook, and we've spent the last 10 years thinking about Google and Facebook. But these platforms are centralized, and highly abstracted. They don't really understand what a "thing" is, they can only look at it several removes away. 

Now we have two new layers, crypto and machine learning.

If you asked in 1993 why the internet matters, once you got over arguing about whether you should say Internet or Internets or World Wide Web, you'd conclude that it was a distributed, decentralized, permissionless, network where anyone could build applications. In 2018, crypto matters for similar reasons: you have a distributed, decentralized, permissionless network on which anyone can build applications, but which also has native trust, value exchange, and payment.

So we've had a pendulum (see the bubbles above) from closed (AOL) to open (internet/web), to closed (Google, FB), to open (ML and crypto).

As we think about the next decade or two, we have some new fundamental building blocks. The internet began as an open, ‘permissionless’, decentralized network, but then we got (and indeed needed) new centralised networks on top, and so we’ve spent a lot of the past decade talking about search and social. Machine learning and crypto give new and often decentralized, permissionless fundamental layers for looking at meaning, intent and preference, and for attaching value to those.

The Video

One more thing before the video. Benedict produces a weekly email newsletter (except when he's in focus mode preparing these presentations). Here are the archives and a box to subscribe. And while you're signing up for that, if you've not already done so, then I produce a weekly email called Sunday Brunch that makes it easy to keep up with the posts on Breakfast Bytes from the previous week. Sign up by clicking on sign me up.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.