The Illogic of the Data Wheel:
It will not save data analytic or AI startups
Many VCs, startups, and consultants believe that today’s unprofitable Unicorns will eventually make enormous profits because of the Data Wheel. Building from another unproved theory, Data is the New Oil, the Data Wheel claims that more data leads to better algorithms, more value and thus more customers, a kind of perpetual motion machine that will lead to winner-take all outcomes. Other factors can be included in the wheel, such as more and better talent, but the logic or should I say illogic is the same: data analytic and AI startups will eventually make profits because of the data wheel.
The key problem is a lack of successful examples. Uber, Lyft, Palantir, Snowflake, Nest Labs, and DeepMind have gathered huge amounts of data yet have losses greater than 50% of revenues despite being founded between eight and 17 years ago. And these examples of huge loss makers are just the tip of the iceberg. Not a single ride sharing company in the world is profitable, only one of 27 ex-Unicorns are profitable in business software and 20 of 21 in consumer Internet, yet most if not all these startups have gathered huge amounts of data. Some peer-to peer loan Unicorns are profitable, but their profits have disappeared as the economic lockdown has made their algorithms, created in a growing economy, mostly useless, thus causing profits to disappear.
Believers in the data wheel might point to Microsoft, Google, Amazon, and Facebook as success stories for the data wheel, but their success is better explained by traditional network economics. Their early advantages in installed base led to greater value, thus creating positive feedback between the number of users and value. Later, switching costs were created that reduced the chances of users moving to another platform, thus leading to “winner-take-all” by one or two firms.
For instance, Microsoft can charge high prices for its software because users do not want to pay the switching costs associated with moving their files to other word processing, spreadsheet, or power point software, costs that were even higher 30 years ago. With large amounts of files created in Microsoft’s formats, moving to different software would require users to make large expenditures in time and money. Many of these users are now “locked-in,” a process that began with the network effects associated with Microsoft’s software; the more people who could share Microsoft-compatible files, the greater the benefit from using Microsoft-compatible files. Thus, most of the world, particularly the corporate world, use Microsoft products and are willing to pay high prices that enabled Microsoft to become one of the most valuable firms in the world.
Similarly, Facebook users are attracted by the large number of Facebook users and probably would not pay the cost of transferring their profiles and associated data to a new social networking provider, even if a new one began to succeed. Amazon’s users are attracted by the wide variety of third-party products sold on Amazon and the third-party sellers would not like to pay the cost of transferring their websites to another host or to become independent. Its cloud users would also not like to incur the costs associated with switching to a new host, costs associated with “dev workflows, documentation, education around processes, [and] the engineering work needed to transition providers. “ For Google, advertisers are attracted by its domination of the search business and many of them do not want learn new tools and analytics, and content providers do not want to modify their Web sites, even if another search provider became popular.
Where does the data wheel fit in these stories? The simple answer is that it doesn’t, but it might in the future. All these companies are big users of data analytics and AI, but their use of data analytics and AI do not explain their success. They may claim it is, or they may tell a similar story to explain their success in order to deflect attention away from the traditional network economics that enable them to maintain their monopolies. They want us to believe that their success comes from relentless innovation and thus they should not be stopped because stopping them will stop America’s economy.
Will the data wheel help today’s Unicorns? The simple answer is maybe. We don’t know yet. Some of these Unicorns will succeed, despite their worse performance than startups of decades ago, demonstrated in previous articles in this series on startups and technologies. Only one startup founded since 2000 has achieved top 100 market capitalization versus six in the 1970s, nine in the 1980s, and eight in the 1990s despite the fact that most of these 24 achieved top 100 status very quickly; three achieved this status within 10 years of founding, six more by 15 years, and seven more by 20 years of their founding. This suggests that more than one founded since 2000 should have achieved top 100 status.
Moreover, none of America’s profitable ex-Unicorns are even close to being ranked in the top 100, as shown in the second article in this series. Being in the top 100 required a market capitalization of $98 billion in 2019 and the highest market cap in 2019 among profitable unicorns was Zoom with a market cap of about $20 billion. Uber had a market capitalization of about $60 billion, but it was highly unprofitable with losses of about $7.4 billion in 2019 and cumulative losses exceeding $20 billion. Other ex-Unicorns are also much less profitable than those of 20 to 50 years ago. The third article in this series found that only six of 45 ex-Unicorns had profits in 2019 despite most of them being founded before 2010, or more than 10 years ago.
It is a mistake to assume the data wheel will save the Unicorns, a kind of hype or hopium, an interesting term coined by Paul Martin. Venture capitalists (VCs) building an investment portfolio around the data wheel is like them building an investment portfolio around the so-called “new economy” of the dotcom bubble, a new economy that purportedly followed different rules than did the bricks and mortar economy. Of course, it didn’t, well shown by Carl Shapiro and Hal Varian in their seminal book Information Rules. The Internet succeeded through exponential improvements in microprocessors, memory chips, computers, and Internet data speeds, improvements that enabled huge amounts of value to be created by the Internet of which some was captured by Facebook, Apple, Amazon, Microsoft, Netflix, Google (FAAMNG), and other startups. But the Internet did not create new economic rules or better metrics just like the current, bubble is not creating new rules or better metrics.
VCs should go back to the basics: create value for users, value that enables users to enjoy better products and services, or to improve productivity in their internal processes. The most successful startups of 20 to 50 years ago, including FAAMNG, did this by exploiting breakthrough technologies and their huge market capitalizations are evidence of the great value they created from these breakthrough technologies. Network economics, including network effects, switching costs, and winner-take all, also played a role, but these factors will have less impact on today’s Unicorns than yesterday’s startups, a topic covered in my 12th article in this series.
Today’s VCs need to create value for users, preferably from the use of breakthrough technologies. This requires much more careful searches for good opportunities, and not charging down the path of hype. This can be done, but it requires much more careful attention to industry and technology economics, something I have covered in other articles.
 https://medium.com/@jeffreyleefunk/are-there-any-industries-in-which-ex-unicorns-are-profitable-747eca652170 https://medium.com/@jeffreyleefunk/how-successful-are-todays-startup-unicorns-893043f32d24