Discover more from European Straits
An Investment Thesis: Help Software Digest the World
Today: We all know the idea of “software eating the world”. Can we develop digestive pills to facilitate that process?
The Agenda 👇
Software eating the world is the mother of all paradigm shifts
All software companies follow the same direction, but some go further than others
Facilitating software eating the world requires dealing with regulations and tangibility
Three examples: Apple and the iPhone; Tesla and electric vehicles; Satoshi Nakamoto
Expect more investors to walk into Andreessen Horowitz’s footsteps in the future
1/ Software is eating the world
I’ve had many occasions to witness the profound impact of Marc Andreessen’s brilliant image about the current paradigm shift. In just a few words, he made people realize the breadth and depth of the widespread transformation that’s currently at work across the global economy.
Let me simply quote what I think is the most revealing paragraph (remember, this was written in 2011):
More and more major businesses and industries are being run on software and delivered as online services — from movies to agriculture to national defense. Many of the winners are Silicon Valley-style entrepreneurial technology companies that are invading and overturning established industry structures. Over the next 10 years, I expect many more industries to be disrupted by software, with new world-beating Silicon Valley companies doing the disruption in more cases than not.
One problem we have is that people see this phenomenon of “software eating the world” as one shift among many others. This is typically what you’ll hear from policymakers: “We’re going through three transitions: a technological transition, an ecological transition, and a geopolitical transition”.
Yet this juxtaposition of transitions not only creates misunderstandings, it also makes it impossible to act! If you have three challenges to tackle at the same time, all of which seem equally threatening, how should you manage the situation? Fighting on three fronts is not a strategy; deciding to focus on one front, on the other hand, is a strategy, and it makes it easier to reach effective decisions.
2/ The mother of all paradigm shifts
My strategy is based on the assumption that the technological transition of “software eating the world” is the shift that underpins all the others. It’s based on the idea that technology and commerce are the main contributors to how history unfolds, echoed in the works of historians such as Fernand Braudel and economists such as Carlota Perez.
After having studied many industries over the years, I even designed my own framework to explain how software eating the world happens in a given industry—a sequence that I detailed in this paper: The Five Stages of Denial. The idea is that all industries go from non-tech to tech via five consecutive stages:
Stage 1: A first wave of startups appears in a given industry.
Stage 2: One of these emerges as a potential winner.
Stage 3: The incumbents strike back, notably on the regulatory front.
Stage 4: Incumbents merge, trying to acquire critical mass against tech rivals.
Stage 5: Tech companies diversify up the stream, cementing a new balance of power.
As does every framework, this one simplifies reality. But having used it since 2015, I have found it tracks with what’s been happening in every industry. There’s a general direction that’s always the same for the process of software eating the world, even though the pace and substance of that process are different depending on the specifics of a given industry.
3/ Same direction, different destinations
As I’ve written many times, there are two phenomena that mix together and explain why it’s harder for software to eat certain industries: tangibility (‘atoms’ rather than ‘bits’) and regulations.
Both make it more difficult for software to eat the world as they tend to make it harder to generate precious increasing returns to scale.
The ideal software company (from a growth perspective) is one that holds no tangible assets, employs no one and serves a market that’s neither fragmented nor limited by regulations. In practice, however, we know that there are two problems:
If the situation is just as I described, the competitive pressure will eventually be too great to withstand, and no contender will end up making money. After all, who wouldn’t want to enter that market if such an ‘easy’ business is a possibility?
Much more likely, the business hits the wall of reality and ends up having to do the same things as any other business: hiring people, investing in tangible assets, and complying with regulations—all of which tend to replace increasing returns with diminishing returns to scale.
Obviously the most interesting businesses belong to the second category, hence my interest in the turning point at which returns start to diminish rather than increasing. In each sector, software eats as much of the world as it can, but at some point it has to stop. The question is: how far can software go before it hits the wall of indigestion?
In my essay Principles for Capital Allocation (Round 1), I introduced what I then called a “sliding scale, from Category 1 (no software) to Category 5 (pure software)”:
Category 1 is about essentially tangible activities with a little bit of software that’s almost immaterial: a website to advertise the company’s value proposition and phone number, or a newsletter to announce summer sales and boost the product when Christmas is approaching.
Category 2 is essentially the same but with a software-driven distribution channel deployed in addition to tangible distribution. This is what all traditional retailers are doing on the grocery market these days with delivery and e-commerce.
Category 3 is where many startups land these days, as software is now eating more tangible industries. Companies at this level have tangible assets and/or employees that make their business more defensible, but their margins are low and they have a rather hard time scaling up.
Category 4 is a software-driven business that relies on tangible assets owned and operated by others. This is the case of marketplaces, managed or not, such as Uber and Airbnb. But this kind of setup also used to exist in the past with franchise businesses (McDonald’s).
Category 5 is a business that relies so much on software that it doesn’t have tangible assets on its balance sheet (not even servers, since those are provided by either Amazon or Microsoft) and not many employees (except for maybe a salesforce whose payroll is a fraction of sales).
The shift always goes in the same direction (software eating as much as it can), but the difference from one industry to another is how much of the industry it can digest.
This, by the way, is the same as during the previous paradigm shift: mass production was the equivalent of software back then, and it was bound to eat the world, too. Yet it got mixed results depending on the industry. The car industry was easily digested; other industries, such as construction, resisted mass production for most of the 20th century.
4/ How to deal with regulations: lobbying or...more software
Regulations are a bit of a paradox: in theory, it would be enough to change them so as to make it easier for software to eat the world. Recently I shared a very interesting discussion by Chris Yiu of the Tony Blair Institute about how we could do that so as to facilitate the growth of the gig economy while advancing the interests of workers who join it:
In practice, however, changing regulations is very hard for reasons we’re all familiar with: more often than not regulations are a reflection of the local culture, which makes it difficult to bring the local population and policymakers together around a new consensus. Even worse, old regulations are usually mastered by a few incumbents with quite a lot of power, and those players are perfectly content with keeping them in place.
That being said, there are different levers that can be pulled to help software digest the world in the realm of regulated industries.
One is plain and simple lobbying, which should be adapted for the startup world. I wrote a handbook for founders on the matter here: A Founder's Handbook for Lobbying the Government.
Another approach is to expand from a geographic perspective so as to be present in different countries and pit them against each other. That was brilliantly executed by Tesla and Waymo, for instance, when it came to getting self-driving cars on the road in various American states; the same approach has been implemented with some success by founders in the cannabis industry. In short: be present in multiple places, shop for the most welcoming one, and make it known that you’re doing so!
Finally, the third approach was alluded to in my The Digital World Is Not a Flat Circle. It consists in using the power of software to make compliance easier and more affordable. We have more than one startup that has been focusing on that in The Family’s portfolio, and this approach was touted by Marc Andreessen and Ben Horowitz in a podcast published in 2014:
Compliance is getting more and more commoditized. As Marc Andreessen and Ben Horowitz suggested here, artificial intelligence is so powerful now that software could swallow any regulation and implement compliance right into any application, making it easy to comply with local regulations almost in real-time and for a minimal cost. So instead of changing regulations and waiting for the states to harmonize them in the entire world, it’s actually easier to develop a proxy that would allow any startup to comply in exchange for almost nothing. Full stack startups in highly regulated sectors such as healthcare or education are currently laying the groundwork.
5/ How to deal with tangibility: enter deep tech
I first came up with this idea of helping software digest the world in my 2019 essay The Hard Truth About Deep Tech:
Deep tech is about helping software digest the world. Software alone can only go so far. Its eating the world requires advanced technological breakthroughs in certain industries. The car industry is more easily digested by software if vehicles are electric, and so you need R&D to come up with these electric vehicles. Likewise, healthcare is more easily digested if you can customize treatment for any patient, and so you need to come up with radically new ways of designing and implementing treatments…
It’s no wonder why Silicon Valley is so passionate about the global deep tech effort. They have their sights set on specific problems in certain industries, and they know that they need breakthrough technologies so that software can more easily digest those.
Indeed, deep tech is all about making it easier for a tangible industry to go through the shift to the Entrepreneurial Age. Like longer-lasting batteries for the car industry, deep tech provides digestive pills to help software eat more of that industry!
I’m especially interested in this because I’ve expressed skepticism about deep tech in the past, but this new thesis of helping software digest the world sheds new light on deep tech as an opportunity to generate hefty returns on invested capital.
Still the question remains: Where do these digestive pills come from? Are there opportunities for investors to be exposed to the value they help create?
6/ Apple and the invention of the smartphone
When people think about deep tech, they think about a laboratory populated with scientists that’s working on a piece of hardware or some obscure molecule. What they don’t always realize is that successful deep tech eventually turns (or is embedded) into an enabling product that becomes so ubiquitous that it effectively helps software digest the world.
A very good example of such a product is the iPhone. It was not the first smartphone per se: when the iPhone launched, many corporate executives were already reading their emails on a BlackBerry. But there were two things with the iPhone that made a difference:
First, it took after its previous, more rudimentary version, the iPod, which in turn came with a powerful application: iTunes. Now, I agree that iTunes today doesn’t meet what users expect from a software product, but at the time it was revolutionary because it came with an account you had to create, and for that account to work you had to submit your name, your address, and your credit card. As I’ve written in the past, this unprecedented ability to monitor users in a regular and systematic way is what effectively turned Apple into a tech company (before that, it was nothing more than a manufacturer of well-designed computers).
Then, one year after the launch of the iPhone in 2007, Apple launched the App Store, which instantly turned into a platform for software developers across the world. Indeed, for them the iPhone was much more than a computer. First, it was a device that people had with them and could potentially use all the time, not only as they were seated at their desk. Second, over time the iPhone provided software developers with resources well beyond an ordinary computer’s capacity—things such as taking pictures (the camera) and tracking movement (GPS and the accelerometers).
This greatly broadened the scope of software applications. For instance, it made it possible to expand the delivery business. Before the smartphone, delivery was hard because it was difficult for the courier and the customer to synchronize and meet at a given place and time, hence the failure of first-generation online delivery companies such as Kozmo. After the smartphone, delivery became easy because the smartphone made it sustainable and value-adding to apply software to the delivery business. Just ask Uber, Deliveroo, Meituang, and many others!
So that’s one configuration: a company creates a generic digestive pill that in turns helps many, many other companies to build new software destined to eat the world. And an investor’s goal should be to back such a company early on so as to reap the benefits over time.
7/ Tesla and the full-stack strategy
Another example of deep tech helping software digest the world is what’s happening with Tesla. Elon Musk’s electric car company fits well into my 5-stage framework:
Stage 1: A first wave of startups appeared in the space of electric vehicles/self-driving cars
Stage 2: Tesla emerged as a potential winner
Stage 3: The incumbents struck back, notably to try and prevent Tesla from selling cars directly to consumers.
Stage 4: Incumbents merged, trying to acquire critical mass against tech rivals. This has been seen with the consolidation in the automotive industry.
Stage 5: Tesla diversified up the stream (with powerwalls, solar panels, gigafactories), cementing a new balance of power.
Now Tesla hasn’t really made it all the way yet (it’s a Category-3 company at best on the scale of software eating the world), but it’s still the winner and, thanks to tremendous progress in non-software technology such as batteries, it has made it easier for software to digest the car industry. The model is vertical integration (or ‘full-stack’—a software company investing in deep tech for itself rather than deploying a platform for others like Apple did with its own achievement on the deep tech front).
About that, let me quote Benedict Evans’s excellent Is Tesla disruptive?
We will go from complex cars with simple software to simple cars with complex software. Instead of many stand-alone embedded systems each doing one thing, we’ll have cheap dumb sensors and actuators controlled by software on a single central control board, running some sort of operating system, with many different threads (there are a few candidates). This is partly driven by electric, but becomes essential for autonomy.
This echoes my favorite of Jeff Bezos’s annual letters to shareholders, the one he sent in 2010:
While many of our systems are based on the latest in computer science research, this often hasn’t been sufficient: our architects and engineers have had to advance research in directions that no academic had yet taken. Many of the problems we face have no textbook solutions, and so we -- happily -- invent new approaches.
In this letter, Bezos explains that Amazon needs to implement computer science R&D to tackle certain challenges, but the same reasoning could be applied beyond the realm of software. “New approaches” in non-software technological fields can make sense from a software perspective, assuming they help software eat the world and thus reap the benefits of related increasing returns to scale.
8/ Crypto and the collective taking charge
A third example is one in which the deep tech component doesn’t come from a company deploying a platform such as Apple or one that explores new approaches on the inside so as to leverage software, such as Tesla or Amazon. It’s bitcoin—a deep tech technology that was designed by a collective of anonymous architects and developers so as to make it easier for software to digest money.
Clearly money is not about tangibility, though—it’s much more about regulations. As I wrote in Facebook's Libra ≋ (June 2019):
Money comes with [several] problems. First, it’s one of the most regulated areas in the global economy, and governments were swift in answering with hostility, constraints, mandates, and regulations.
There are two reasons why it didn’t take an established corporation such as Apple, Tesla, or Amazon to tackle this particular challenge. First, corporate CEOs might be visionaries with a ‘reality distortion field’, but they’re not crazy to the point of tackling the world’s financial regulators and central banks all at once. Second, it really took the political motivation of a collective of individuals in the aftermath of the financial crisis to decide that facilitating the digestion of money by software was needed. As I wrote in the Libra piece,
We can understand why pioneers in the crypto world chose money as the first challenge to tackle—as opposed to, say, email or personal data, two fields riddled by problems that could be solved by new protocols. After all, it was right after the financial crisis, and the need to build a new financial system seemed obvious to many.
9/ Andreessen Horowitz’s positioning
Once you realize that software needs digestive pills so as to eat industries that are more tangible and/or more regulated, the infinite diversification of Andreessen Horowitz, Marc Andreessen’s VC firm, starts to make sense. They are far from being present in all corners of the world (yet), but they have set up dedicated vehicles specifically targeted at funding digestive pills in (at least) two interesting fields.
One is a crypto fund, led by veteran entrepreneur and investor Chris Dixon and former financial regulator Katie Haun. As explained in the announcement of raising Crypto Fund II,
These are a few areas where people are already building, but it only scratches the surface of the yet-to-be-imagined applications that entrepreneurs will dream up. In the same way that it wasn’t obvious in 2007 how applications on top of mobile phones would change so many aspects of the ways in which we move, consume, travel, communicate, and even date, it’s hard to imagine what the very best apps and use cases will be for blockchain-based computing platforms.
The other field I have in mind is biotech. Venture capital has existed in biotech forever, but its historical focus has long been to fund research and development of new drugs, not things related to software. This is why biotech is dominated by very different VC firms than those present in computing and networks (‘tech’), and many firms that were trying to do both under the same roof eventually had to split (as was the case when the biotech branch of Index Ventures became Medicxi in 2016).
Well, the exact opposite has happened with Andreessen Horowitz’s bio funds, led by Vijay Pande. As he and his colleagues explained in the announcement of raising Bio Fund III,
Four years ago, when we launched our first bio fund, the idea that software would make an impact in drug discovery sounded like science fiction; the idea it could transform the patient experience in healthcare a fool’s errand. Fast forward to 2020, and it’s not just accepted but broadly embraced, igniting an explosion of new startups and big investments from established incumbents. Tech, biotech, and our healthcare system are merging — into what we call simply “bio”. And whether for pharma, hospitals, or investors, bio is now officially the hot new thing.
You get the marketing trick: it’s introduced as an effort to help biotech by leveraging software. But my reading in the broader context of Andreessen Horowitz is that it’s more an effort to help software (finally) eat biotech, and eventually healthcare in general!
I wouldn’t be surprised to see Andreessen Horowitz launching more specialized funds in the future, with a focus on industries that software still has a hard time eating, such as energy, agriculture, construction, space. In any case, theirs is an example that more investors across the world could follow.
10/ A compelling investment thesis
Indeed, once it becomes clear that software entrepreneurs need digestive pills for their startups to succeed in ‘hard’ industries, we don’t need to wait for platform companies such as Apple, full-stack companies such as Tesla, or a collective of insurgents led by another Satoshi Nakamoto to make things easier. The following eventually becomes an obvious investment thesis:
Spot an industry in which software will eventually be in need of digestive pills.
Identify entrepreneurs that have the potential to develop these pills.
Back them and make sure they emerge on the radar as indispensable contributors.
Interestingly, this seems to be exactly what Cathie Wood and her team are doing at ARK Invest. Here’s a list of publicly listed funds that they manage:
Autonomous Technology & Robotics ETF
Next Generation Internet ETF (“artificial intelligence, big data, cloud computing, cybersecurity, and blockchain technology”)
Genomic Revolution ETF
Fintech Innovation ETF
The 3D Printing ETF
Forthcoming space ETF (a new one dedicated to space)
Now, this is mostly deep tech, designed to help software digest tangible industries. But as for regulated industries, there are similar approaches:
One is to set up a specialized investment firm that backs software entrepreneurs who try to enter these industries and provide value-added services to clear the way for them. A well-known example is Tusk Ventures. Another promising one, closer to home, is run by my friends in London Leo Ringer and Patrick Newton, who launched Form Ventures very recently.
Another approach is an emerging sector known as “RegTech”, about which I wrote a few lines three years ago in my Thoughts on Switzerland (September 2018):
I was especially interested in a segment described to me as “RegTech”, which means startups focused on making compliance easier. One example is Apiax, of which I met Andrew Campbell, a seasoned business developer. Apiax’s value proposition is to provide their customer in the financial services industry with advanced tools to master complex regulations, thus making “compliance lean and efficient again”. I’m glad I now have this example in mind, as it perfectly illustrates an idea first voiced by Marc Andreessen in this podcast: technology is so powerful now that software could swallow any rule and implement compliance right into any application, making it easy to satisfy local regulations in almost real-time and for a minimal cost.
And here’s a bonus: Benedict Evans’s recent take on Outgrowing software 🤗
What do you think of this ‘digestive pills’ thesis? Do you have investment firms in mind that you think are representative of that trend? What are, according to you, the next industries to be eaten by software? And what are the assets, either on the deep tech front or the regulatory side, that need to be developed to facilitate the process? Looking forward to reading your thoughts!
If you enjoyed this edition of European Straits, you should subscribe so as not to miss the next ones.
From Munich, Germany 🇩🇪