Google and Ambient Computing

The most surprising revelation from yesterday’s Made by Google 19 keynote came in Google Senior Vice President of Devices and Services Rick Osterloh’s opening remarks:

If you look across all of Google’s products, from Search to Maps, Gmail to Photos, our mission is to bring a more helpful Google for you. Creating tools that help you increase your knowledge, success, health, and happiness. Now when we apply that mission to hardware and services, it means creating products like…Pixel phones, wearables, laptops, and Nest devices for the home. Each one is thoughtfully and responsibly designed to help you day to day without intruding on your life.

Did you catch that? Apparently Google has a new mission — to bring a more helpful Google to you. So much for organizing the world’s information!

To be clear, I’m overplaying what was surely a misstatement; five months ago, at the beginning of May’s Google I/O keynote, CEO Sundar Pichai both reiterated the company’s longstanding mission statement while also introducing the “helpful” phrasing that Osterloh used:

It all begins with our mission to organize the world’s information and make it universally accessible and useful. Today our mission feels as relevant as ever, but the way we approach it is constantly evolving. We are moving from a company that helps you find answers to a company that helps you get things done. This morning we’ll introduce you to many products built on a foundation of user trust and privacy…we want our products to work harder for you in the context of your job, your home, and your life. They all share a single goal: to be helpful, so we can be there for you big and small over the course of your day.

So “being helpful” is the company’s goal, not its mission statement. A fine distinction, perhaps, but I’m grateful for the misstatement: going back to Pichai’s comments was how I made sense of what was, at first viewing, a pretty boring and self-satisfied event.

Google’s Announcements

Google announced, in order:

  • That Stadia, the company’s video game streaming service, would launch on November 19th
  • Pixel Buds, the company’s AirPods competitor, which will ship in “Spring 2020”; there weren’t even working models for the press to try
  • Pixelbook Go, the company’s third Chromebook, which will start shipping October 28
  • New pricing for Nest Aware, the cloud recording service for Nest devices; instead of charging a fee per device Google will charge a flat fee per household. The new plans will launch in “early 2020”
  • Nest Wifi, a mashup of its Google Wifi mesh router with Google Home speakers, which will start shipping November 4
  • A new Nest Mini, a replacement for the Google Home Mini, which will start shipping on October 22
  • The Pixel 4 smartphone, with radar chips, new cameras, and enhanced Google Assistant capabilities; it will start shipping on October 24

The first thing that is striking about this list is how many of the announcements won’t ship for quite some time. The second thing is that most of the products were not announced on their own merits, but rather after long interludes about Google’s product development process. Like I said, boring and self-satisfied.

Pichai’s articulation of the company’s new goal, though, is helpful to understand what I believe the company was driving towards: to “be helpful” Google needs to be everywhere, which by extension means the company needs to be trusted. Thus the announcement of a wide array of products — whether ready to launch or not — that covered a multitude of places of where you might need Google’s assistance, done in the context of explaining how Google really does have its users best interests at heart.

Google’s Vision

Osterloh described this vision as “ambient computing”. From the keynote:

In the mobile era, smartphones changed the world. It’s super useful to have a powerful computer everywhere you are. But it’s even more useful when computing is anywhere you need it, always available to help. Now you heard me talk about this idea with Baratunde, that helpful computing can be all around you — ambient computing. Your devices work together with services and AI, so help is anywhere you want it, and it’s fluid. The technology just fades into the background when you don’t need it. So the devices aren’t the center of the system, you are. That’s our vision for ambient computing.

Frankly, it’s a compelling vision on multiple dimensions:

  • First, it is a vision for the future that actually seems larger than the smartphone reality we live in. Alternatives like augmented reality or wearables feel smaller.
  • Second, it is a vision that does not compete with the smartphone, but rather leverages it. The smartphone is so useful for so many things that any directly competitive technology would have to cover an impossible number of use cases to displace it; ambient computing, though, simply conceives of the smart phone as one of several means to deliver on its promise.
  • Third, it is a vision that Google is uniquely suited to pursue. The company is a services company incentivized to serve the maximum number of customers no matter the means (i.e. device), and it already has a head start in providing services that contain and accumulate essential information about people’s lives.

Note how much better Google is placed than Facebook or Amazon, both of which I wrote about two weeks ago. The latter two companies are hindered by their lack of a smartphone, and their beachheads in the consumer space — Oculus and Alexa, respectively — are constrained by specialization in the case of Facebook and location in the case of Amazon. Ambient computing that goes away when you turn off a headset or leave your house is not truly ambient. Osterloh made this point:

The Google Assistant plays a critical role here. It pulls everything together and gives you a familiar, natural way to get the help you need. Our users tell us they find the Google Assistant to be smart, user-friendly, and reliable, and that’s so important for ambient technology. Interactions need to feel natural and intuitive. Here’s an example: if you want to listen to music, the experience should be the same whether you are in the kitchen, you are driving in your car, or hanging out with friends. No matter what you are doing, you should be able to just say the name of the song and the music just plays without you having to pull out a phone and tap on screens or push buttons.

Only companies with smartphone platforms can deliver the same experience everywhere. That is to say, only Google and Apple, and the latter seems to be barely trying in the home in particular.

Google’s Integration

This also explains why Google, despite being a Services company, is investing in hardware. Clayton Christensen in The Innovator’s Solution, explained what it took to win in new markets:

When there is a performance gap — when product functionality and reliability are not yet good enough to address the needs of customers in a given tier of the market — companies must compete by making the best possible products. In the race to do this, firms that build their products around proprietary, interdependent architectures enjoy an important competitive advantage against competitors whose product architectures are modular, because the standardization inherent in modularity takes too many degrees of design freedom away from engineers, and they cannot not optimize performance.

To close the performance gap with each new product generation, competitive forces compel engineers to fit the pieces of their systems together in ever-more-efficient ways in order to wring the most performance possible out of the technology that is available. When firms must compete by making the best possible products, they cannot not simply assemble standardized components, because from an engineering point of view, standardization of interfaces (meaning fewer degrees of design freedom) would force them to back away from the frontier of what is technologically possible. When the product is not good enough, backing off from the best that can be done means that you’ll fall behind.

In the case of ambient computing, “integration” does not refer to an individual device and its associated software. Rather, the integration that matters is between all of the various devices that exist in every part of your life — home, work, play, and everywhere-in-between — and the service that links them together. Thus all of Google’s various hardware offerings: without question the best solution for ambient computing by some time next year will be Nest devices in your house, a Google Pixel in your pocket, Pixel Buds in your ears, and a Pixelbook at work.

Google's Ambient Computing

I don’t think this is Google’s long-run goal, though, nor should it be. While the company has at times been drawn into the trap of prioritizing and differentiating Android with its services, the fundamental services nature of Google means that its ambient computing offering will leverage any OEM that wishes to take part, even Apple. For now, though, the technology just isn’t good-enough, which is why Google is doing a lot of the work itself.

Google’s Challenges

Despite how well-placed Google is to execute on this vision, it is not a certainty that the company will win, for reasons both structural and also internal to Google itself.

First, the customers most likely to not only be interested in the idea of ambient computing but to also have the significant funds necessary to buy all of the various gadgets required to make it a reality probably use iPhones. Apple was the high-end integrated player in smartphones, and contra-Christensen, that was a sustainably large portion of the market. Google, meanwhile, was the modular player in smartphones, which meant it had the most affordable smartphone offerings and by far the largest marketshare. The challenge the company faces is that its modular customer base is less likely to spend on the integrated solution that Google is selling.

Second, while Siri will likely never reach the reliability and usability of Google Assistant — Apple has its own internal challenges — Apple continues to increase the switching cost from iPhone by doubling down on devices. AirPods are infinitely better than Pixel Buds in that they actually exist and have for three years, and the Apple Watch continues to grow strongly. Both devices, particularly when used together, also give you ambient computing beyond the smartphone (and yes, HomePod is still muddling along).

Third, as I noted above, Google spent so much time yesterday framing its approach in terms of user-centricity for a very good reason: its core advertising business is under attack for treating users and their data as a commodity. This raises the question as to whether customers will be comfortable having Google involved in even more aspects of their life, a point that Apple has and will continue to make regularly (Google, as I wrote after I/O, is fighting back by touting the benefits that come from it having so much data).

Fourth, Google has a business model problem. Yes, per the previous point, being a continuous presence in people’s lives will bring in even more data for ever more finely targeted advertisements, but there is no place for advertising in ambient computing generally. The Google Assistant can only give one answer, and it had better be the best one, not one that is paid for, if Google wishes to retain trust.

Fifth, to the extent the previous point does not matter, simply because Search and Display and YouTube make so much money, is the extent to which Google can be lackadaisical about execution. It doesn’t really matter that a good portion of the products announced yesterday won’t be ready until next month or next year because they are a rounding error on Google’s income statement. That may seem like a luxury, but in fact needing to succeed or die is one of the greatest advantages a company can have, particularly while trying to enter a new market.

Google’s Culture

One thing Google can absolutely work on is their messaging: I found yesterday’s presentation dreadfully boring, and only picked up on what Google was trying to convey on a second viewing.

That, though, isn’t necessarily a surprise. Google from its founding has succeeded simply by being better and letting the masses figure it out for themselves. It completely worked too: Google search was better than anything else on the market, and by virtue of being on the Internet it was immediately accessible to anyone anywhere on a zero marginal cost basis.

The company struggles, though, when it has to actually sell something. Look no further than Google Cloud Platform, which is a distant third to Amazon (which was first) and Microsoft (which can sell, particularly to existing customers). The company is currently trying to brute force its way into contention, hiring a VP from Oracle and a whole bunch of salespeople, but those efforts will run up against the company’s sense that simply building better stuff should be enough.

The challenges in ambient computing will be different given the differences between the consumer and enterprise markets, but no less significant: to succeed, particularly with its integrated offering, Google has to get better at all parts of the funnel, from initial awareness to education to conversion to channel to distribution to support. However, there is not much evidence the company has made progress in any of these areas, and, given how strong the company’s core business remains, not much motivation to either.

That’s the thing with visions: they are easy to come up with, harder to articulate, and even more difficult to build. It is the selling, though, that truly requires dedication.

The China Cultural Clash

It all started with a tweet:

Daryl Morey's since-deleted tweet

“It” refers to the current imbroglio surrounding Daryl Morey, the General Manager for the Houston Rockets of the National Basketball Association (NBA), and the latter’s dealings with China. The tweet, a reference to the ongoing protests in Hong Kong,1 “hurt the feelings of the Chinese people” (a rather frequent occurrence). The Global Times, a Chinese government-run English-language newspaper, stated in an editorial:

Daryl Morey, general manager of the NBA team the Houston Rockets, has obviously gotten himself into trouble. He tweeted a photo saying “fight for freedom, stand with Hong Kong” on Saturday while accompanying his team in Tokyo. The tweet soon set the team’s Chinese fans ablaze. It can be imagined how Morey’s tweet made them disappointed and furious. Shortly afterward, CCTV sports channel and Tencent sports channel both announced they would suspend broadcasting Rockets’ games. Some of the team’s Chinese sponsors and business partners also started to suspend cooperation with the Rockets.

There’s one rather glaring hole in this story of immediate outrage from Chinese fans over Morey’s tweet: Twitter is banned in China.

China Started It

Earlier this year I wrote about the uneven playing field between the U.S. and China when it comes to technology companies. From China, Leverage, and Values:

This is where I take the biggest issue with Culpan labeling this past week’s actions as the start of a tech cold war: China took the first shots, and they took them a long time ago. For over a decade U.S. services companies have been unilaterally shut out of the China market, even as Chinese alternatives had full reign, running on servers built with U.S. components (and likely using U.S. intellectual property)…

The truth is that the U.S. China relationship has been extremely one-sided for a very long time now: China buys the hardware it needs, and keeps all of the software opportunities for itself — and, of course, pursues software opportunities abroad.

This understated the case: not only were Chinese companies allowed into the U.S. while U.S. companies remained locked out of China, Chinese attacks on U.S. tech companies were allowed by China’s censors, and in fact even augmented by the Great Firewall. James Griffiths wrote about the 2015 attack on Github earlier this year:

In a paper coauthored with researchers at Citizen Lab, an activist and research group at the University of Toronto, Weaver described a new Chinese cyberweapon that he dubbed the “Great Cannon.” The “Great Firewall” — an elaborate scheme of interrelated technologies for censoring internet content coming from outside China—was already well-known. Weaver and the Citizen Lab researchers found that not only was China blocking bits and bytes of data that were trying to make their way into China, but it was also channeling the flow of data out of China.

Whoever was controlling the Great Cannon would use it to selectively insert malicious JavaScript code into search queries and advertisements served by Baidu, a popular Chinese search engine. That code then directed enormous amounts of traffic to the cannon’s targets…The cannon could also be used for other malware attacks besides denial-of-service attacks. It was a powerful new tool: “Deploying the Great Cannon is a major shift in tactics, and has a highly visible impact,” Weaver and his coauthors wrote.

The attack went on for days. The Citizen Lab team said they were able to observe its effects for two weeks after GitHub’s alarms first went off. Afterward, as the GitHub developers struggled to make sense of the attack and come up with a road map for future incidents, there was confusion within the cybersecurity community. Why had China launched so public an attack, in such a blunt fashion? “It was overkill,” Weaver told me. “They kept the attack going long after it had ceased working.”

It was a message: a shot across the bow from the architects of the Great Firewall, who—having conquered the internet at home—were now increasingly taking aim overseas, unwilling to brook challenges to their system of control and censorship, no matter where they came from.

The projects China was presumably targeting were Chinese versions of GreatFire.org, which documents censorship by the Great Firewall, and the New York Times, both of which were hosted on Github. Given the importance of Github to software development, China could not block the site completely, so instead they tried to hold it hostage. It was a harbinger of what happened this week.

Dreams Versus Reality

The story about engagement with China, both in terms of the U.S. generally but also tech specifically, has long been a belief that some engagement was better than no engagement, and that the shift to more freedom was inevitable. President Bill Clinton stated when the U.S. established Permanent Normal Trade Relations with China:

The change this agreement can bring from outside is quite extraordinary, but I think you could make an argument that it will be nothing compared to the changes that this agreement will spark from the inside out in China. By joining the W.T.O., China is not simply agreeing to import more of our products; it is agreeing to import one of democracy’s most cherished values: economic freedom. The more China liberalizes its economy, the more fully it will liberate the potential of its people — their initiative, their imagination, their remarkable spirit of enterprise.

He added with regards to the Internet:

In the new century, liberty will spread by cell phone and cable modem. In the past year, the number of Internet addresses in China has more than quadrupled, from 2 million to 9 million. This year the number is expected to grow to over 20 million. When China joins the W.T.O., by 2005 it will eliminate tariffs on information technology products, making the tools of communication even cheaper, better, and more widely available. We know how much the Internet has changed America, and we are already an open society. Imagine how much it could change China.

Now there’s no question China has been trying to crack down on the Internet. Good luck! That’s sort of like trying to nail jello to the wall. But I would argue to you that their effort to do that just proves how real these changes are and how much they threaten the status quo. It’s not an argument for slowing down the effort to bring China into the world, it’s an argument for accelerating that effort. In the knowledge economy, economic innovation and political empowerment, whether anyone likes it or not, will inevitably go hand in hand.

In fact, it turned out that China was able to first contain the Internet, blocking sites outside the Great Firewall, then control the Internet, censoring content on social networks like Weibo and WeChat, and, as this New York Times article explains, even leverage the Internet:

The Communist Party indeed doesn’t hesitate to use state power to tell the Chinese people how they should think. But the displays of patriotism, especially from young people, also show that the party’s propaganda machine has mastered the power of symbol and symbolism in the mass media and social media era…While imposing tight censorship, the Communist Party has also learned to lean on the most popular artists and the most experienced internet companies to help it instill Chinese with patriotic zeal. It’s propaganda for the Instagram age, if Instagram were allowed in China.

The problem from a Western perspective is that the links Clinton was so sure would push in only one direction — towards political freedom — turned out to be two-way streets: China is not simply resisting Western ideals of freedom, but seeking to impose their own. Note this statement from state-owned broadcast CCTV, as it announced that it would not televise NBA games:

NBA Commissioner Adam Silver defended Morey. “I think as a values-based organization that I want to make it clear…that Daryl Morey is supported in terms of his ability to exercise his freedom of expression,” Silver said in an interview with Kyodo News in Tokyo Japan. CCTV did not agree with Silver’s remarks.

“We are strongly dissatisfied and we oppose Silver’s claim to support Morey’s right of free expression. We believe that any speech that challenges national sovereignty and social stability is not within the scope of freedom of speech,” CCTV said in its statement in Chinese, which was translated by CNBC.

The lever for this rather radical definition of “freedom of speech” is the China market. The Global Times editorial I linked to above could not have been more explicit on this point:

Respecting customers is a universal business rule. Morey has to choose between safeguarding his individual freedom of speech and protecting the Rockets’ commercial interests by respecting the feelings of Chinese fans. When he opted for the former, the Rockets will have to make a second choice from the perspective of the team.

In other words, Morey, a private U.S. citizen posting an image on a social network already banned in China, had to be fired, or the Rockets and the NBA would quite literally pay the price. Abide by China’s standards, or else.

The TikTok Question

China’s exportation of its standards goes beyond brute force. Consider TikTok, the short-form video app owned by the $75 billion Chinese startup ByteDance, which has exploded onto Western markets over the last year. The Guardian reported last last month:

TikTok, the popular Chinese-owned social network, instructs its moderators to censor videos that mention Tiananmen Square, Tibetan independence, or the banned religious group Falun Gong, according to leaked documents detailing the site’s moderation guidelines. The documents, revealed by the Guardian for the first time, lay out how ByteDance, the Beijing-headquartered technology company that owns TikTok, is advancing Chinese foreign policy aims abroad through the app.

The revelations come amid rising suspicion that discussion of the Hong Kong protests on TikTok is being censored for political reasons: a Washington Post report earlier this month noted that a search on the site for the city-state revealed “barely a hint of unrest in sight”.

In fact, at least as of this afternoon, there is a hint of unrest on the site: while searches for “Hong Kong” show city views and high school students playing along with the latest TikTok meme, searching for Hong Kong in Chinese (香港) brings up a video that shows the protestors as hooligans and vandals (this was the first result as of this afternoon, and the only video relating to the protests):

Honk Kong on TikTok

There appear to be similar efforts in the case of the NBA controversy. Searching for the “Warriors”, “Lakers”, and “Rockets” brings up the sort of content you would expect:

NBA teams in Chinese on TikTok

However, searching for the same team names in Chinese (“勇士”, “湖人”, and “火箭”, respectively) shows basketball-related results for the first two and nothing related for the third:

Tiktok searches for NBA teams

This should raise serious concern in the United States and other Western countries: is it at all acceptable to have a social network that has a demonstrated willingness to censor content under the control of a country that has clearly different views on what constitutes free speech?

There is an established route for undoing this state of affairs: earlier this summer China’s Kunlun Tech Company agreed to divest Grindr under pressure from the Committee on Foreign Investment in the United States (CFIUS); Kunlun Tech had acquired Grindr without undergoing CFIUS review. TikTok similarly acquired Musical.ly without oversight and relaunched it as TikTok for the Western market; it is worth at least considering the possibility of a review given TikTok’s apparent willingness to censor content for Western audiences according to Chinese government wishes.

The NBA’s Example

Adam Silver, the commissioner of the NBA, ultimately did do the right thing. In response to that CCTV cancellation Silver released a new statement that stated:

Values of equality, respect and freedom of expression have long defined the NBA — and will continue to do so. As an American-based basketball league operating globally, among our greatest contributions are these values of the game¬

It is inevitable that people around the world — including from America and China — will have different viewpoints over different issues. It is not the role of the NBA to adjudicate those differences. However, the NBA will not put itself in a position of regulating what players, employees and team owners say or will not say on these issues. We simply could not operate that way.

Silver added in a press conference following the statement:

Part of the reason I issued the statement I did is because this afternoon, CCTV announced that because of my remarks supporting Daryl Morey’s freedom of expression, not the substance of this statement but his freedom of expression, they were no longer going to air the Lakers-Nets preseason games that are scheduled for later this week. Again, it’s not something we expected to happen. I think it’s unfortunate. But if that’s the consequence of us adhering to our values, we still feel it’s critically important we adhere to those values.

I am increasingly convinced this is the point every company dealing with China will reach: what matters more, money or values?

China Responses

I am not particularly excited to write this article. My instinct is towards free trade, my affinity for Asia generally and Greater China specifically, my welfare enhanced by staying off China’s radar. And yet, for all that the idea of being a global citizen is an alluring concept and largely my lived experience, I find in situations like this that I am undoubtedly a child of the West. I do believe in the individual, in free speech, and in democracy, no matter how poorly practiced in the United States or elsewhere. And, in situations like this weekend, when values meet money, I worry just how many companies are capable of choosing the former?

The NBA, to its immense credit, appears to have done just that. Will technology companies be so brave? Certainly Google did so once before, exiting China in 2010 (albeit after both losing share to Baidu and being attacked by Chinese hackers). At the same time, the company appeared eager to reverse its decision, only terminating “Project Dragonfly” earlier this year; similarly, Facebook worked earnestly for approval — its product team built a censorship apparatus and CEO Mark Zuckerberg learned Chinese — only to give up last year. Both decisions appear motivated by the certainty of failure as opposed to core values.

And then there is Apple: the company is deeply exposed to China both in terms of sales and especially when it comes to manufacturing. The reality is that, particularly when it comes to the latter, Apple doesn’t have anywhere else to go. That, though, is where the company’s massive cash stockpile and ability to generate more comes in handy: it is past time for the company to start spending heavily to build up alternatives. Sticking one’s corporate head in the sand, praying that President Trump will not be re-elected and that everything will go back to normal, is deeply irresponsible both to shareholders and to the values Apple claims motivates them.

The government response is also critical: I already argued that CFIUS should revisit TikTok’s acquisition of Musical.ly; the current skepticism around all Chinese investment in the United States should be continued if not increased. Attempts by China to leverage market access into self-censorship by U.S. companies should also be treated as trade violations that are subject to retaliation. Make no mistake, what happened to the NBA this weekend is nothing new: similar pressure has befallen multiple U.S. companies, often about content that is outside of China’s borders (Taiwan and Hong Kong, for example, being listed in drop-down menus for hotels or airlines).

The biggest, shift, though, is a mindset one. First, the Internet is an amoral force that reduces friction, not an inevitable force for good. Second, sometimes different cultures simply have fundamentally different values. Third, if values are going to be preserved, they must be a leading factor in economic entanglement, not a trailing one. This is the point that Clinton got the most wrong: money, like tech, is amoral. If we insist it matters most our own morals will inevitably disappear.

I wrote a follow-up to this article in this Daily Update.

  1. This post is not about the specific issues driving the Hong Kong protests; it is useful to understand that China, particularly internally, has characterized the protests, which began when the Hong Kong government attempted to pass an extradition bill that would allow extradition to China, as a separatist movement driven by foreign powers. The protesters state their goals are not independence but rather that China honor its promises surrounding the transfer of Hong Kong back to Chinese rule, particularly in terms of universal suffrage []

Beachheads and Obstacles

The fact that Amazon held its annual hardware event the same day as the keynote for Facebook’s Oculus Connect conference is almost certainly a coincidence. It was, though, a happy one, at least as far as Stratechery is concerned: these two events, wildly disparate in terms of presentation and content, have more in common than it might seem.

Revisiting the Smartphone Wars

In 2013, when Stratechery started, the widely held belief was that the iPhone, innovative though it may have been, was in serious trouble in the face of Android’s increasing marketshare. Henry Blodget wrote a useful articulation of the bear case on Business Insider:

If smartphones and tablets were not a platform — if the only thing that mattered to the value of the product and a customer’s purchase decision was the gadget itself — then Apple’s loss of market share would not make a difference. Apple zealots would be correct when they smugly assert that what matters is Apple’s “profit share” not “market share.”

But smartphones and tablets are a platform. Third-party companies are building apps and services to run on smartphone and tablet platforms. These apps and services, in turn, are making the platforms more valuable. Consumers are standardizing their lives around the apps and services that run on smartphone and tablet platforms. Because of these “network effects,” in platform markets, dominant market share is huge competitive advantage. In platform markets, as the often-hated but always insanely powerful Microsoft demonstrated for decades in the PC market, the vast majority of the power and profits eventually accrue to the market-share leader.

In this view there is still a premium market, but only within the dominant ecosystem. This, Blodget argued, was Apple’s problem: soon the company would have no market, because Android would have the ecosystem, and by extension all of Apple’s premium customers.

Of course this turned out to be mistaken, for reasons I laid out in What Clayton Christensen Got Wrong.

  • First, integration provided user experience benefits that premium customers valued
  • Second, those premium users were more likely to pay for apps, which increased the attraction of iOS to developers
  • Third, the absolute size of the smartphone market was so big that both iOS and Android were large enough to be attractive to developers

Note, though, that just because Blodget and company were wrong about the iPhone’s prospects does not mean they were wrong conceptually: ecosystems do matter. However, instead of one ecosystem devouring the entire premium versus ubiquity landscape, Apple and Google split it up rather neatly:

Apple and Google's Ecosystem Duopoly

Amazon and Facebook were two of the more prominent companies that found out this reality the hard way.

Mobile Successes and Failures

Apple and Google may be the first companies people think of when you ask who won mobile, but Amazon and Facebook were not far behind.

Amazon spent the smartphone era not only building out Amazon.com, but also Amazon Web Services (AWS). AWS was just as much a critical platform for the smartphone revolution as were iOS and Android: many apps ran on the phone with data or compute on Amazon’s cloud; mobile also created a vacuum in the enterprise for SaaS companies eager to take advantage of Microsoft’s desire to prop up its own mobile platforms instead of supporting iOS and Android, and those SaaS companies were built on AWS.

Smartphones, meanwhile, saved Facebook from itself: instead of a futile attempt to be a platform within the browser, mobile made Facebook just an app, and it was the best possible thing that could have happened to the company. Facebook was freed to focus solely on content its users wanted and advertising to go along with it, generating billions of dollars and a deep moat in targeting advertising along the way.

What is not clear is if Amazon’s and Facebook’s management teams agree. After all, both launched smartphones of their own, and both failed spectacularly.

Facebook’s attempt was rather half-assed (to use the technical term). Instead of writing their own operating system, Facebook Home was a launcher that sat on top of Android; instead of designing their own hardware, the Facebook One was built by HTC. Both decisions ended up being good ones because they made failure less expensive.

Amazon, meanwhile, went all out to build the Fire Phone: a new operating system (based on Android, but incompatible with it), new hardware, including a complicated camera system that included four front-facing cameras, and a sky-high price to match. It fared about as well as the Facebook One, which is to say not well at all.

That, though, is what made last week’s events so interesting: it is these two failures that seemed to play a bigger role in what was announced than did the successes.

Amazon and Facebook’s Announcements

Start with Amazon: the company announced a full fifteen hardware products. In order: Echo Dot with Clock, a new Echo, Echo Studio (an Echo with a high-end speaker system), Echo Show 8 (a third-size of the Echo with a screen), Echo Glow (a lamp), new Eero routers, Echo Flex (a microphone only Echo that hangs off an outlet), Ring Retrofit Alarm Kit (that lets you leverage your preinstalled alarm), Ring Stick Up Cam (a smaller Ring camera), Ring Indoor Cam (an even smaller Ring camera), Amazon Smart Oven (an oven that integrates with Alexa), Fetch (a pet tracker), Echo Buds (wireless headphones with Alexa), Echo Frames (eyeglasses with Alexa), and Echo Loop (a ring with Alexa). Whew!

This is an approach that is the exact opposite of the Fire Phone: instead of pouring all of its resources into one high-priced device, Amazon is making just about every device it can think of, and seeing if they sell. Moreover, they are doing so at prices that significantly undercut the competition: the Echo Studio is $150 cheaper than a HomePod, the Echo Show 8 is $60 cheaper than the Google Nest Hub, and the new Eero is $150 cheaper than the product Eero sold as an independent company. Amazon is clearly pushing for ubiquity; a whale strategy this is not.

Facebook, meanwhile, effectively consolidated its Oculus product line from three to one: the mid-tier Oculus Quest, a standalone virtual reality (VR) unit, gained the capability to connect to a gaming PC in order to play high-end Oculus Rift games; Oculus Go apps, meanwhile, gained the capability to run on the relatively higher-specced Oculus Quest. It is not clear why either the Go or Rift should be a target for developers or customers going forward.

The broader goal, though, remains the same: Facebook is determined to own a platform; the lesson the company seems to have drawn from its smartphone experience is the importance of doing it all.

Beachheads and Obstacles

What Amazon and Facebook do have in common — and perhaps this is why both seem to look back at their very successful smartphone eras with regret — is that Apple and Google are their biggest obstacles to success, and it’s because of their smartphone platforms.

Amazon to its great credit — and perhaps because the company did not have a smartphone to rely on — found a beachhead in the home, the one place where your phone may not be with you. Now it is trying to not only saturate the home but also extend beyond it, both through on-body accessories and also an expanding number of deals with automakers.

Facebook, meanwhile, is searching for a beachhead of its own in virtual reality. That, the company believes, will give it the track to augmented reality, and by extension, usefulness in the real world.

Facebook and Amazon are building beachheads to take on Apple and Google

Amazon’s challenge is Google: Android phones are already everywhere, and Google is catching up in the home more quickly and more effectively than Amazon is pushing outside of it. Google also has a much stronger position when it comes to the sort of Internet services that provide the rough grist of intelligence of virtual assistants: emails, calendars, and maps.

Facebook, meanwhile, is ultimately challenging Apple: augmented reality is going to start at the high end with an integrated solution, and Apple has considerably more experience building physical products for the real world, and a major lead in chip design and miniaturization, not to mention consumer trust. Moreover, while there is obviously technical overlap when it comes to creating virtual reality and augmented reality headsets, the product experience is fundamentally distinct.

Lessons Learned

I’ve been pretty skeptical about Facebook and Oculus all along, both at the time of purchase and last year. I’d like to say I’ve changed my mind, but frankly, last week’s keynote made me question whether Facebook learned any lessons from mobile at all. Zuckerberg said in the keynote opening:

We experience the world through this feeling of presence and the interactions that we get with other people, which is why Facebook’s technology vision has always been about putting people at the center of your computing experience. We’ve mostly done that so far through building apps. I don’t think it’s an accident that a lot of the top-used and biggest apps that are out there are social experiences that put people at the center of the experience, because that’s how we process things.

But there is only so much you can do with apps, without also shaping and improving the underlying platform. I find it shocking that we’re here in 2019 and our phones and our computers are still organized around apps and tasks and not people that we are actually present with. I feel like we can help all of us together deliver a unique contribution to this field by helping to ensure that the next platform changes this.

Zuckerberg is, in effect, saying that he finds it shocking that Facebook Home didn’t succeed. I think the reasons were pretty clear, and a lack of distribution or high-end hardware was not the primary problem. The fact of the matter is that while social connection on our phones is important — perhaps the most important — it is not the only job we ask phones to do. That is why Facebook is an app and not a platform, and that’s ok! Apps, particularly those of Facebook’s scale and advertising prowess, are fantastic businesses. And apps shouldn’t be platforms.

Amazon, on the other hand, seems to have learned the right lessons from its mobile failures; what is notable about the company’s approach to Alexa is that it leverages and learns from the mobile era. Alexa benefits from Amazon’s investments in data centers and networking, interacts with both iOS and Android to the greatest extent possible, and is roughly inline with Amazon’s overall business — making buying things that much more convenient. Alexa is an operating system for the home, and perhaps beyond.

This isn’t a guarantee of success, of course. Google is a formidable competitor, with multiple advantages. It is particularly hard to see Alexa gaining traction outside the home. The only reason Amazon has a chance is because building on strengths is always better than doing something completely new and different from what has made you successful in the past.

Neither, and New: Lessons from Uber and Vision Fund

The first time I wrote about Uber was in June, 2014. The Wall Street Journal had posted a column entitled Uber’s $18.2B Valuation is a Head Scratcher, which led to an easy rejoinder: Why Uber is Worth $18.2 Billion. Given that Uber is today worth $53.2 billion on the open market, that one turned out pretty well.

A month later I felt even better about my piece when Bill Gurley, the legendary venture capitalist, wrote his own rebuttal of an Uber skeptic. Gurley was gentle in his takedown of NYU Stern professor Aswath Damodaran, writing in the introduction:

It is not my aim to specifically convince anyone that Uber is worth any specific valuation. What Professor Damodaran thinks, or what anyone who is not a buyer or seller of stocks thinks, is fairly immaterial. I am also not out to prove him wrong. I am much more interested in the subject of critical reasoning and predictions, and how certain assumptions can lead to gravely different outcomes. As such, my goal is to offer a plausible argument that the core assumptions used in Damodaran’s analysis may be off by a factor of 25 times, perhaps even more. And I hope the analysis is judged on whether the arguments I make are reasonable and feasible.

Gurley’s arguments, which focused on Damodaran’s assumptions around Uber’s total addressable market ($100 billion, the same as taxis) and terminal market share (10%) were clearly correct: Uber is already at a $50+ billion gross bookings run rate, and has around 70% of the market. Damodaran’s assumptions, rooted in the analog world, were sorely mistaken.

At the same time, while Gurley didn’t make any specific assertions about Uber’s valuation, surely he must have expected it would have increased by more than 192% in the following five years; I certainly did. To be sure, there were rather significant intervening events, specifically Uber’s disastrous 2017, where the company endured seemingly endless scandals, lost its CEO, and worst of all, gave life to Lyft, its most important competitor which, at the beginning of that year, was on the verge of going out of business. It is very fair to argue that Uber without an at-scale competitor is a much more valuable company.

That noted, this line from Gurley’s article stands out to me today more than ever:

I am much more interested in the subject of critical reasoning and predictions, and how certain assumptions can lead to gravely different outcomes.

Just because Uber’s critics were wrong to assume that the service was analogous to taxis does not mean that those of us on the other side — not only of the Uber question but of a host of other similar companies that straddle the physical and digital worlds — were completely right in our assumptions either. The opposite of an old-world company is not necessarily a tech company. It is something we haven’t quite seen before, and applying either old-world rules or tech rules is a mistake.

AB 5 and Worker Classification

This idea of the old classifications not quite making sense, and the need for something new, should feel quite familiar in the context of Uber: it is precisely the issue surrounding Uber’s drivers.

Earlier this month California passed AB 5, which codified a California Supreme Court decision setting forward a three-part test to determine whether or not a worker is an independent contractor or an employee (with all of the attendant regulation and taxes that go along with that classification). From the decision:

Under this test, a worker is properly considered an independent contractor to whom a wage order does not apply only if the hiring entity establishes: (A) that the worker is free from the control and direction of the hirer in connection with the performance of the work, both under the contract for the performance of such work and in fact; (B) that the worker performs work that is outside the usual course of the hiring entity’s business; and (C) that the worker is customarily engaged in an independently established trade, occupation, or business of the same nature as the work performed for the hiring entity.

The question as to whether the new law applies is closer than it seems: on one hand, Uber et al1 really do give drivers, who use their own equipment, flexibility as far as hours go, and while there are rules to be followed while on the job, it is the former that is usually the more important standard. Plus drivers famously drive for multiple companies; the need to compete for their presence on the platform (more on this in a bit) is one of the big reasons why Uber is so unprofitable.

That means that (B) is the question: if Uber is in the transportation business, then drivers are workers; Uber, though claims its business “is serving as a technology platform for several different types of digital marketplaces.” As I wrote in a Daily Update:

It’s not an entirely irrational argument. For example, consider the rate: Uber’s point is that not that it sets the rate, but rather the rate is the market-clearing price that maximizes the amount of revenue drivers earn. The idea is that if drivers could set their own prices — a common objection to drivers being independent contractors is that they cannot — a negotiation would occur between customers and drivers until a price was agreed upon; over time this price would be equalized across drivers and riders. Uber’s argument is that it dramatically accelerates this process and in fact makes the market possible, since the level of coordination necessary to reach a market-clearing price at scale would be impossible otherwise.

At the same time, this sort of argument, technically correct from an economic modeling perspective, suffers from the same flaws as most economic models: the lack of any sort of accounting for the human component. In this case the missing bit, though, is not in the model’s outcome, but rather in the manifestation: the way that Uber is experienced by riders and especially riders is that “Drivers are the face of Uber to consumers” (that quote is from Uber’s S-1, by the way). Drivers are also indispensable to how Uber actually generates revenue: sure, drivers can and do come and go as they please, and work simultaneously for Uber’s competitors, but to suggest they are a not a part of the “usual course” of Uber’s business seems off.

That is why the best solution to the employment classification question is to realize that neither of the old categorizations fit: Uber drivers are not employees, nor are they contractors; they are neither, and new. A much better law would define this category in a new way that provides the protections and revenue-collection apparatus that California deems necessary while still preserving the flexibility and market-driven scalability that make these consumer welfare-generating platforms possible.

What is Uber?

So what of Uber itself? It is not a taxi company, as noted above, but is it a tech company? I suggested it was a few weeks ago in What Is a Tech Company?:

Uber…checks most of the same boxes:

  • There is a software-created ecosystem of drivers and riders.
  • Like Airbnb, Uber reports its revenue as if it has low marginal costs, but a holistic view of rides shows that the company pays drivers around 80 percent of total revenue; this isn’t a world of zero marginal costs.
  • Uber’s platform improves over time.
  • Uber is able to serve the entire world, giving it maximum leverage.
  • Uber can transact with anyone with a self-serve model.

A major question about Uber concerns transaction costs: bringing and keeping drivers on the platform is very expensive. This doesn’t mean that Uber isn’t a tech company, but it does underscore the degree to which its model is dependent on factors that don’t have zero costs attached to them.

In fact, I’ve changed my mind: I was right to mention Uber’s costs, and wrong to dismiss them and call Uber a tech company. At the same time, Uber clearly has no analog in the physical world. It is neither, and new — and Uber’s drivers help explain why.

That magical marketplace I described above, where Uber effectively simulates countless one-on-one negotiations between drivers and riders that, on an infinite timescale and with infinite patience, would arrive at the market-clearing price, is very much a technological product. This marketplace leverages today’s paradigm-shifting technologies — smartphones and cloud computing — and is itself software, and thus infinitely leverageable and always improving.

Uber’s financials reflect this: last quarter the company had a gross margin of 51%. That is a fair bit lower than a typical SaaS company’s 70%+ gross margins, but that is primarily because the company’s cost of revenue includes insurance, which scales linearly with revenue. The software behind Uber’s marketplaces scales perfectly.

The problem, though, is that Uber’s financials are an incomplete view on the overall Uber experience, because riders don’t simply pay Uber: they also pay the drivers. And, if you look at Uber’s financials from a rider perspective,2 the situation looks a lot worse; consider last quarter:

in millions Uber’s Financials The Rider Perspective
Revenue $2,768 $15,574
Cost of Revenue $1,342 $14,148
Gross Profit $1,426 $1,426
Gross Margin 51.5% 9.2%

Suddenly that gross margin looks nothing like a software company — and keep in mind this is all Uber has to work with before it gets to its fixed costs.3 The only way this company works is if it grows to a truly mammoth size such that it has sufficient gross margin to cover fixed costs, but it is that much more difficult to acquire a marginal new customer when you simply don’t have that much margin to play with; spending on sales and marketing simply increases the hill you need to climb!

None of this is to say that Uber is not a viable business: all of Gurley’s arguments about the total addressable market and Uber’s ability to dominate that market still apply, because of technology. Uber is not a taxi company! At the same time, a different sort of valuation metric than that usually applied to tech companies was clearly appropriate as well, as Uber’s adventures on the public market demonstrate. In short, the company was neither, and new.

The Uber Anomaly

The corresponding article to What Is a Tech Company? could very well be What Is a Venture Capital Firm?. If tech companies are characterized by zero marginal costs, increased returns to scale, and ecosystems, venture capital firms match with equity financing (which means capped downside and infinite upside), a Babe Ruth portfolio management approach that focuses on home runs despite the increase in strikeouts, and a focus on iterated games when it comes to exerting power.

The synergy between tech companies and venture capitalists

That last point is worth dwelling on; in 2017 I described why an iterated game approach mattered for venture capitalists in the context of — you guessed it! — Uber. That was when Gurley’s Benchmark, then Uber’s largest investor, first forced out and then sued Uber’s former CEO Travis Kalanick.

A venture capitalist will invest in tens if not hundreds of companies over their career, while most founders will only ever start one company; that means that for the venture capitalist investing is an iterated game. Sure, there may be short-term gain in screwing over a founder or bailing on a floundering company, but it simply is not worth it in the long-run: word will spread, and a venture capitalists’ deal flow is only as good as their reputation…

The entire point of venture investing is to hit grand slams, and that calls for more swings of the bat. After all, the most a venture capitalist might lose on a deal — beyond time and opportunity cost, of course — is however much they invested; the downside is capped. Potential returns, though, can be many multiples of that investment. That is why, particularly as capital has flooded the Valley over the last decade, preserving the chance to make grand slam investments has been paramount. No venture capitalist wants to repeat Sequoia’s mistake: better to be “nice”, or, as they say in the Valley, “founder friendly.”

Uber, though, was different:

Uber’s most recent valuation of $68.5 billion nearly matches the worth of every successful Benchmark-funded startup since 2007. Sure, it might make sense to treat company X and founder Y with deference; after all, there are other fish in the pond. Uber, though, is not another fish: it is the catch of a lifetime.

That almost assuredly changed Benchmark’s internal calculus when it came to filing this lawsuit. Does it give the firm a bad reputation, potentially keeping it out of the next Facebook? Unquestionably. The sheer size of Uber though, and the potential return it represents, means that Benchmark is no longer playing an iterated game. The point now is not to get access to the next Facebook: it is to ensure the firm captures its share of the current one.

As I’ve noted, that valuation proved to be faulty; at the same time, $53.2 billion is still a huge amount of money, and probably wouldn’t have changed Benchmark’s calculation. The real takeaway, though, is that Uber was not a typical Silicon Valley startup. No, they weren’t a taxi company, but they weren’t a tech company either, they were something new, and that meant a new kind of investor. Enter SoftBank.

Vision Fund

Masayoshi Son, Softbank’s CEO and the driving force behind Vision Fund, told Bloomberg a year ago that he wanted to “go big bang”:

SoftBank’s massive bet in WeWork is emblematic of Son’s overall approach. “Why don’t we go big bang?” he told Bloomberg in an interview last year when asked about his investing style, and added that other venture capitalists tend to think too small. His goal of swaying the course of history by backing potentially world-changing companies requires that those companies make large outlays in areas from customer acquisition to hiring talent to research and development, a spending tactic that he acknowledged sometimes brings him into conflict with other investors.

“The other shareholders, they try to create clean, polished little companies,” Son said. “And I say: ‘Let’s go rough. We don’t need to polish. We don’t need efficiency right now. Let’s make a big fight. Let’s make a big, successful—a big win.’”

In fact, the “other shareholders” that Son derides are trying to create tech companies: up-front fixed costs to develop software, with high gross margins once it is sold. These are the companies that require investors that have all of the qualities I detailed above: a desire for equity, a willingness to risk strikeouts while swinging for home runs, and the decency that comes from playing an iterated game.

Vision Fund is none of these things. It doesn’t just want equity, it wants preferred equity with a ratchet, to guarantee they get theirs first. Moreover, it seeks to not only invest in winners, but also to leverage its capital to make winners, by forcing competing companies to merge. And, because of this, Vision Fund is very much not playing an iterative game: it will do whatever it takes to win the markets it invests in, including deposing of founders who become a liability.

The problem, though, is Vision Fund may have confused “big capital needs” with “big opportunity”. What is striking about the firm’s portfolio is the paucity of “tech companies”. Almost everything falls in the “Neither and New” category defined by Uber: entire categories like real estate and logistics are defined by their interaction with the physical world, almost everything in the consumer category uses technology to enable real-world services, and the other major category, fintech, by definition needs huge amounts of capital. Most of these companies may have income statements that seem attractive in isolation, but when viewed from a total revenue perspective4 in fact have extremely low gross margins (relative to tech companies) and very high marginal costs.

The question for Softbank then is how many markets are there the size of transportation, with the possibility of taking a large enough chunk to make the economics work (leaving aside the fact that Softbank is underwater on its Uber investment)? The Vision Fund is invested in OpenDoor, for example, which is in an even larger market than transportation (residential real estate), but with much less potential transaction volume; Zillow, which followed OpenDoor into the “iBuyer” market, has a market cap of only $6 billion, in part because of investor skepticism about margins.

This is the challenge for Vision Fund: yes, these companies have huge capital needs, and yes, the only way they can become successful is if they become so big that their small margins are sufficient to cover their fixed cost, but does that necessarily mean big returns? Or did Son anchor on “big” without making sure that his adjective of choice attached to the noun — “returns”, as opposed to “needs” or “markets” — that his investors are expecting?

Moreover, it’s not clear how many misses Vision Fund can afford: the Wall Street Journal reported earlier this week that Vision Fund has promised a 7% return a year to 40% of its investors, which means that SoftBank has limited capacity to be patient and wait for home runs — particularly if WeWork starts dragging down the whole fund.

Worse, it’s not clear how many home runs Softbank has. Looking at 29 U.S. tech IPOs since the beginning of 2018, 20 have increased in market cap over their offering price, and all of them are pure tech companies with high margins.5 Of the nine that have fallen in value, four are marketplace companies6, two are hardware companies7, and only three are pure tech companies8. Son, though, sees pure technology companies as “clean, polished little companies” that are not big enough for Vision Fund.9

Vision Fund is not a venture capital firm, nor is it a public market-focused hedge fund: it is neither, and new, but it very much remains to be seen if “new” is valuable.

New Lessons

At the same time, this is good news for the tech ecosystem: there is clearly still tremendous opportunity to build “tech companies”, primarily for the enterprise, and Vision Fund won’t be an obstacle. True, there are fewer opportunities in the consumer space, but that is more a consequence of big company dominance than Venture Fund stealing away opportunities with outsized returns relative to capital invested. If anything Vision Fund is stealing duds.

This is also good news for public market investors: despite all of the press about Uber and WeWork, more companies are up post-IPO than down — and the gains are much larger in percentage terms than are the losses. The tech company formula still works.

This is also a lesson for me: I started with an article that I got right, but in retrospect I was only halfway correct. Uber had a large market and there were tech-like dynamics that meant it could get a big part of that market, but margins — both reported, but especially relative to the customer transaction — still matter. I didn’t pay enough attention to them.

It also means I should have been more explicitly skeptical about WeWork; my goal was to write a contrarian piece exploring the upside, while still being clear that I wouldn’t invest. I did state that, but I wasn’t nearly clear enough about just how absurd the valuation was, because I didn’t spend enough time discussing margins.10

Going forward I plan to be a lot more skeptical about other tech startups that interface with the real world and the attendant drag on margins that follows; I am not saying that the category isn’t viable, and technology truly makes these companies different than the incumbents in their space, but they are not necessarily tech companies either.

Neither, and new.

I wrote a follow-up to this article in this Daily Update.

  1. I am going to use Uber as a stand-in for companies like Lyft, DoorDash, Instacart, etc. for the rest of this article, but everything applies to all of the companies that use “gig” workers []
  2. In this case, the rider perspective includes both Uber riders and also UberEats customers; the two categories are not separated in Uber’s financials []
  3. It should be noted that Uber, rightfully, accounts for driver incentives either as contra-revenue (most of them) or as a cost of revenue (for driver incentives it is unlikely to earn back); the only driver incentives that fall under fixed costs are bonuses to existing drivers for driver referrals []
  4. I.e. the equivalent of gross bookings in the case of Uber []
  5. In order of returns, Zscaler, Anaplan, Smartsheet, Zoom, DocuSign, CrowdStrike, Fastly, SurveyMonkey, Pinterest, Health Catalyst, Medallia, Cloudflare, Carbon Black, Dynatrace, Datadog, PagerDuty, EverQuote, Zuora, Tenable []
  6. UpWork, Eventbrite, Uber, and Lyft []
  7. Sonos and Arlo Technologies []
  8. Pivotal, Dropbox, and Slack []
  9. Interestingly, the one tech company on this list that the Vision Fund owns, Slack, is the worst performing of all the SaaS companies []
  10. I did, though, make clear in a (free) follow-up that the AWS comparison was never intended to be a direct one []

Exponent Podcast: The Exponent IPO

On Exponent, the weekly podcast I host with James Allworth, we discuss Cloudflare and what it is like going through an IPO, as well as what has gone wrong with WeWork.

Listen to it here.

Day Two to One Day

Jeff Bezos opened his 2016 letter to Amazon shareholders like this:

“Jeff, what does Day 2 look like?”

That’s a question I just got at our most recent all-hands meeting. I’ve been reminding people that it’s Day 1 for a couple of decades. I work in an Amazon building named Day 1, and when I moved buildings, I took the name with me. I spend time thinking about this topic.

“Day 2 is stasis. Followed by irrelevance. Followed by excruciating, painful decline. Followed by death. And that is why it is always Day 1.”

To be sure, this kind of decline would happen in extreme slow motion. An established company might harvest Day 2 for decades, but the final result would still come.

Bezos went on to give advice about how to avoid Day 2, including “True Customer Obsession”, “Resist Proxies”, “Embrace External Trends”, and “High-Velocity Decision Making”. The company he manages then spent the next several years looking like it was in fact Day 2.

Tilting the Scales

Consider this story in the Wall Street Journal about how Amazon reportedly adjusted its search algorithm to favor its own products:

Amazon.com Inc. has adjusted its product-search system to more prominently feature listings that are more profitable for the company, said people who worked on the project—a move, contested internally, that could favor Amazon’s own brands…The adjustment, which the world’s biggest online retailer hasn’t publicized, followed a yearslong battle between executives who run Amazon’s retail businesses in Seattle and the company’s search team, dubbed A9, in Palo Alto, Calif., which opposed the move, the people said.

Note how badly this decision fares relative to Bezos’ advice:

  • Shifting results away from relevance towards factors that benefits Amazon’s bottom line is not a decision that results from “true customer obsession”.
  • Goal-seeking for profit is a poor proxy for that customer obsession that Bezos focused on in the 1997 shareholder letter attached to every subsequent letter.
  • Amazon allegedly spent “years” deciding whether or not to do this, which is definitely not “high-velocity decision making”.

To be fair, Amazon is embracing the external trend of raising antitrust concerns, which are probably overblown given the company’s single-digit share of retail in the United States. There are no objections to Walmart, for example, having store brands or pay-for-placement programs, despite the fact that Walmart’s share of retail is about 33% larger than Amazon’s (8.9% of consumer retail spending in the U.S. versus 6.4 percent), so it’s not clear on what basis the digital equivalents of these programs would be prosecuted.

With regards to Bezos’ warning, though, the antitrust discussion is a moot point: companies that spend months or years arguing about the legality and customer friendliness of tilting the scales are usually well into Day Two.

Squeezing Suppliers

This is hardly the only example of Amazon becoming obsessed with profitability on the margins in its retail operation over the last few years. For example, from Recode last November:

Over the past few months, Amazon has applied intense pressure to consumer brands across different product categories — seizing more control over what, where and how they can sell their goods on the so-called everything store, these people say. One apparent goal: To take more control over the price of goods on Amazon so the company can better compete with retailers. The power moves are also believed to be a prelude to a new internal system that Amazon has yet to launch called One Vendor. The new initiative will essentially funnel big brands and independent sellers alike through the same back-end system in a supposed effort to improve the uniformity of the shopping experience across Amazon on the public-facing side.

From the Wall Street Journal in December:

As Amazon focuses more on its bottom line in addition to its rapid growth, it is increasingly taking aim at CRaP products [“Can’t Realize a Profit”], according to major brand executives and people familiar with the company’s thinking. In recent months, it has been eliminating unprofitable items and pressing manufacturers to change their packaging to better sell online, according to brands that sell on Amazon and consultants who work with them.

From CNBC in March:

In recent months, Amazon has been telling more vendors, or brand owners who sell their goods wholesale, that if Amazon can’t sell those products to consumers at a profit, it won’t let them pay to promote the items. For example, if a $5 water bottle costs Amazon that amount to store, pack and ship, the maker of the water bottle won’t be allowed to advertise it.

From Bloomberg, also in March:

Amazon.com Inc. has abruptly stopped buying products from many of its wholesalers, sowing panic. The company is encouraging vendors to instead sell directly to consumers on its marketplace. Amazon makes more money that way by offloading the cost of purchasing, storing and shipping products. Meanwhile, Amazon can charge suppliers for these services and take a commission on each transaction, which is much less risky than buying goods outright.

From Bloomberg in May:

In the next few months, bulk orders will dry up for thousands of mostly smaller suppliers, according to three people familiar with the plan. Amazon’s aim is to cut costs and focus wholesale purchasing on major brands like Procter & Gamble, Sony and Lego, the people said. That will ensure the company has adequate supplies of must-have merchandise and help it compete with the likes of Walmart, Target and Best Buy.

The vendor purge is the latest step in Amazon’s “hands off the wheel” initiative, an effort to keep expanding product selection on its website without spending more money on managers to oversee it all. The project entails automating tasks like forecasting demand and negotiating prices which were predominantly done by Amazon employees. It also involves pushing more Amazon suppliers to sell goods themselves so Amazon doesn’t have to pay people to do it for them.

None of these decisions are necessarily wrong in a vacuum; what has been striking, though, is the drumbeat of Amazon Retail changes that seem primarily concerned about Amazon’s profitability. And, for the record, it has worked:

Amazon.com's North American Results

Throughout the second half of 2018 and the first part of 2019, Amazon flipped revenue and expense growth by just a smidge, which caused income to skyrocket on a year-over-year basis. It may have been Day 2 as far as Amazon’s prioritization of profitability above everything was concerned, but at least the company was, in Bezos’ words, “harvesting”.

One Day Shipping

Note, though, that the chart above is missing last quarter’s results, and for good reason:

Amazon.com's North American Results

It’s a bit hard to make out, particularly because I am using trailing twelve-month averages (because of Amazon’s high seasonality), but expenses increased a lot more than revenue last quarter; in fact year-over-year income growth on a quarterly basis was actually -15%.

What changed is that Amazon decided to travel back in time — to Day One — and invest in what the company does best: massively difficult logistical problems that customers love having solved. Originally that was access to any book, then access to anything period, then access in two days, and now Amazon is committed to one.

First, from the company’s Q1 2019 earnings call announcing Amazon’s ambition:

We’re currently working on evolving our Prime free Two-Day Shipping program to be a free One-Day Shipping program. We’re able to do this, because we spent 20 plus years expanding our fulfillment and logistics network, but this is still a big investment and a lot of work to do ahead of us.

For Q2 guidance, we’ve included approximately $800 million of incremental spend related to this investment. And just to clarify, to give a little more information, we have been offering, obviously, faster than Two-Day Shipping for Prime members for years, one day, same day, even down to one to two hour delivery for Prime Now. So we’re going to continue to offer same day and Prime Now selection in an accelerated basis.

But this is all about the core free Two-Day offer evolving into a free One-Day offer. We’ve already started down this path. We’ve in the past months significantly expanded our one-day eligible selection and also expanded the number of zip codes eligible for one-day shipping.

The costs started to show up last quarter. From the company’s Q2 2019 earnings call, a quarter where Amazon missed on profits for the first time in several years:

In Q2, we had a meaningful step up in the one-day shipments, primarily in North America, and one-day volume was accelerating throughout the quarter…On the cost side, we talked last time about $800 million estimate of transportation cost to supply one day, the additional one day in Q2. We were a little bit higher than that number in total cost.

We saw some additional transition costs in our warehouses. We saw some lower productivity as we were expanding rather quickly, both local capacity in the off-season also in our delivery networks. We also saw some costs were moving: buying more inventory and moving inventory around in our network to have it be closer to customers. And we built not only that cost structure, but an accelerating cost penalty into our Q3 guidance that was released with our earnings today.

This is an initiative that clearly passes Bezos’ test:

  • Customers love getting items in one day instead of two.
  • One day shipping is a clear goal.
  • Increased convenience will always be the ultimate external trend.
  • Ramping up one-day shipping in a manner of weeks by definition requires high-velocity decision making.

It is also the opposite of harvesting: it is investing, and it seems more likely than not that Amazon’s upcoming results will look much more like the “Day One” company it was for years, with rapidly growing revenue and costs to match.


This Article comes at a bit of a weird time: in truth I had been considering writing a bearish Amazon article for several months as the penny-pinching anecdotes started to pile up. Tech companies rarely find sustainable growth by focusing on costs; if anything they find antitrust violations.

That announcement about one-day shipping, though, made me hold my fire. Spending a lot of resources on incredibly difficult logistical problems is precisely what makes Amazon so valuable, which means that the commitment to do just that — even with higher costs — is a reason to be bullish. The only problem is that the revenues I anticipate have not yet appeared in the quarterly results.

Still, this search news made me revisit the issue a bit early: tilting the field to favor the bottom line instead of doing what is best for customers is the surest sign of harvesting instead of investing, and it reminded me of my bearish thesis. I wonder if Amazon might not reconsider their approach to search now that the company is demonstrating a recommitment to growing the top line instead of the bottom.

That’s also why last week’s Apple event was encouraging: Apple may have its work cut out to be an effective services company, but by cutting iPhone prices and pricing its services offerings aggressively it is making its own moves towards investing, not simply harvesting. This is also why Facebook’s commitment to Stories was a good sign even if it entailed an earnings hit; its various attempts to wring engagement out of its core app through things like forced Instagram integrations and dating services run in the opposite direction. For Microsoft, meanwhile, The End of Windows meant the end of harvesting and a return to investing, much to investors’ benefit.

Perhaps the biggest question mark, though, is around Google: the company has gotten far more mileage than I ever expected out of mobile generally and cramming more ads into mobile search results specifically; both, though, particularly the latter, seem more like harvesting than investing. And, even when Google does invest, it is too often in projects far removed from customers and the forcing function that going to market entails.

This also may be why Google is the most susceptible to antitrust action of all the major consumer tech companies; the question as to what comes first, harvesting instead of investing or behaving anticompetitively, ceases to matter when you are operating at the scale of any of these companies. And, on the flipside, it strongly suggests that antitrust actions are a trailing indicator of a company that has peaked,1 not a causal force of decline.

  1. YouTube remains a tremendously important counterweight to any bearish Google story []

The iPhone and Apple’s Services Strategy

Editor’s Note: Stratechery was referenced in yesterday’s keynote. I had no knowledge of or awareness of this reference, and have no relationship with Apple, up-to-and-including not owning their stock individually, as explained in my ethics policy.

It is the normal course for Apple events to come and go and people to complain about how boring it all was, particularly when the company announces said event like this:

Apple Event Invitation: "By Innovation Only"

Apple reporter extraordinaire Mark Gurman was not impressed:

Gurman isn’t necessarily wrong about the highly iterative nature of the hardware announcements (although I think that an always-on Apple Watch is a big deal), but that doesn’t necessarily mean he is right about the innovation question. To figure that out we need to first define what exactly innovation is.

Beyond the iPhone, Revisited

Another Apple keynote that was greeted with a similar collective yawn was in 2016, when the company announced the iPhone 7 and Series 2 Apple Watch. Farhad Manjoo wrote at the time in the New York Times:

Apple has squandered its once-commanding lead in hardware and software design. Though the new iPhones include several new features, including water resistance and upgraded cameras, they look pretty much the same as the old ones. The new Apple Watch does too. And as competitors have borrowed and even begun to surpass Apple’s best designs, what was iconic about the company’s phones, computers, tablets and other products has come to seem generic…

I quoted Manjoo’s piece at the time and went on to explain why I thought that year’s keynote was more meaningful than it seemed, particularly because of the AirPods introduction:

What is most intriguing, though, is that “truly wireless future” Ive talked about. What happens if we presume that the same sort of advancement that led from Touch ID to Apple Pay will apply to the AirPods? Remember, one of the devices that pairs with AirPods is the Apple Watch, which received its own update, including GPS. The GPS addition was part of a heavy focus on health-and-fitness, but it is also another step down the road towards a Watch that has its own cellular connection, and when that future arrives the iPhone will quite suddenly shift from indispensable to optional. Simply strap on your Watch, put in your AirPods, and, thanks to Siri, you have everything you need.

That future is here, although the edges are still rough (particularly Siri, which was a major focus of that article); Apple’s financial results have certainly benefited. Over the last three years the company’s “Wearables, Home and Accessories” category, which is dominated by the Apple Watch and AirPods, has nearly doubled from $11.8 billion on a trailing twelve-month (TTM) basis1 to $22.2 billion over the last twelve months. In other words, according to the metric that all businesses are ultimately measured on, that 2016 keynote and the future it pointed to was very innovative indeed.

Apple’s Services Narrative

Wearables have not been Apple’s only growth area: over the same three-year span Services revenue has increased by almost the exact same rate — 89% versus 88% — from $23.1 billion TTM to $43.8 billion TTM. At the same time, it feels a bit icky to call that innovation, particularly given the anticompetitive nature of the App Store.

That’s not totally fair of course: the App Store was one of the most innovative things that Apple ever created from a product perspective; that the company has positioned itself to profit from that innovation indefinitely is innovative in its own right, at least if you go back to measuring via revenue and profits.

Still, the idea of Apple being a Services company is one that has long been hard to grok. When the company first started pushing the “Services Narrative” I declared that Apple is not a Services Company:

Services (horizontal) and hardware (vertical) companies have very different strategic priorities: the former ought to maximize their addressable market (by, say, making a cheaper iPhone), while the latter ought to maximize their differentiation. And, Cook’s answer made clear what Apple’s focus remains.

That answer was about continuing Apple’s pricing approach, which at that time was $649+ for new iPhones, with old iPhones discounted by $100 for every year they were on the market, and Cook’s specific words were “I don’t see us deviating from that approach.”

In fact, Apple did deviate, but in the opposite direction: in 2017 the company launched the $999+ iPhone X at the high end and bumped the price of the now mid-tier iPhone 8 to $699+. I wrote at the time:

The iPhone X sells to two of the markets I identified above:

  • Customers who want the best possible phone
  • Customers who want the prestige of owning the highest-status phone on the market

Note that both of these markets are relatively price-insensitive; to that end, $999 (or, more realistically, $1149 for the 256 GB model), isn’t really an obstacle. For the latter market, it’s arguably a positive.

What this strategy was absolutely not about was expanding the addressable market for Services. Apple was definitely not a Services company when it came to their strategic direction (even if, as I conceded in 2017, it was increasingly fair to evaluate the financial results in that way).

The iPhone’s Price Cut

This leads to what is in my mind the biggest news from yesterday’s event: Apple cut prices.

It was easy to miss, given that the iPhone 11 Pro, the successor to the iPhone X and then XS, hasn’t changed in price: it still starts at $999 ($1,099 for the larger model), and tops out at $1,449; if you want the best you are going to pay for it.

Perhaps the most interesting aside in the keynote, though, is that for the first time a majority of Apple’s customers weren’t willing to pay for the best. Tim Cook said:

Last year we launched three incredible iPhones. The iPhone XR became the most popular iPhone and the most popular smartphone in the world. We also launched the iPhone XS and iPhone XS Max, the most advanced iPhones we have ever created.

In a vacuum there is nothing surprising about this. The iPhone XR was an extremely capable phone, with the same industrial design, the same Face ID, and the same processor as the iPhone XS; the primary differences were an in-between size, one less camera, and an LCD screen instead of OLED. That doesn’t seem like much of a sacrifice for a savings of $250.

And yet, even while I said Apple’s strategy “bordered on over-confidence”, I still fully expected the iPhone XS to be the best-selling phone like the iPhone X before it; that is how committed Apple’s customers have been to buying the flagship iPhone. Even Apple, though, can’t escape the gravitational pull of “good enough” — which is why the price cuts, which happened further down the line — were so important.

There are two ways to see Apple’s price cuts. First, by iPhone model:

Launch 1 year old 2 years old
iPhone 7 $649 $549 $449
iPhone 8 $699 $599 $449
iPhone XR $749 $599
iPhone 11 $699

Secondly by year:

Flagship Mid-tier 1 year old 2 years old
2016 $649 $549 $449
2017 $999 $699 $549 $449
2018 $999 $749 $599 $449
2019 $999 $699 $599 $449

In the second chart you can see how Apple in 2017 not only raised prices dramatically on its flagship models, but also on the mid-tier model relative to previous flagships. This was important because it was these mid-tier models that replaced previous flagships in Apple’s usual “sell the old flagship for $100 less per year” approach. That meant that 2017’s price hike filtered through to 2018’s 1-year-old model, which increased from $549 to $599.

That means that this year actually saw three price cuts:

  • First, the iPhone 11 — this year’s mid-tier model — costs $50 less than the iPhone XR it is replacing.
  • Second, the iPhone XR’s price is being cut by $150 a year after launch, not $100 as Apple has previously done.
  • Third, the iPhone 8’s price is also being cut by $150 two years after launch, not $100 as Apple has previously done.

To be fair, this doesn’t necessarily mean the line looks much different today than it did yesterday: the only price point that is different is the iPhone 11 relative to the XR. That, though, is because it will take time for those previous price hikes to work their way out of the system, presuming Apple wants to stay on this path in the future.

They should. The success of the iPhone XR strongly suggests that there is more elasticity in the iPhone market than ever before. Apple also cut prices in China earlier this year with great success; I wrote after Apple’s FY2019 Q2 earnings:

The available evidence strongly suggests that iPhone demand in China is very elastic: if the iPhone is cheaper, Apple sells more; if it is more expensive, Apple sells less. This is, of course, unsurprising, at least for a commodity, and right there is Apple’s issue in China: the iPhone is simply less differentiated in China than it is elsewhere, leaving it more sensitive to factors like new designs and price than it is elsewhere.

As I note in that excerpt, China is unique, but the commodity argument is a variant of the “good-enough” argument I made above: while Apple doesn’t necessarily need to worry about iPhone customers outside of China switching to Android, they are very much competing with the iPhones people already have, and, as the XR demonstrated, their own new, cheaper phones.

That’s ok, though, and the final step in Apple truly becoming a Services company, not just in its financial results but also in its strategic thinking. More phones sold, no matter their price point, means more Services revenue in the long run (and Wearables revenue too).

Apple’s Services Announcement

Apple’s two service-related announcements are also good reasons to pursue this strategy. Perhaps the most compelling from a financial perspective is Apple Arcade. For $4.99/month a family gets access to a collection of games featured on their own tab in the App Store:

What makes this compelling from Apple’s perspective is that the company is paying a fixed amount for those games overall, which means that once the company covers the costs of those games, every incremental subscription is pure profit. Contrast this to something like Apple Music, where costs scale inline with revenue; no wonder the service is getting such prime real estate — and no wonder Apple suddenly seems interested in selling more iPhones, even if they earn less revenue up-front.

Similar dynamics apply to Apple TV+: once content costs are covered, incremental customers are pure profit. That noted, I’m not convinced that Apple TV+’s ultimate purpose is to be a profit driver by itself; I explained after Apple’s services event earlier this year:

To be very clear about my analysis of Apple TV+, I don’t think it is a Netflix competitor. I see it as a customer acquisition cost for the Apple TV app; it is Apple TV Channels that will make the real money, and this is not an unreasonable expectation. Roku’s entire business is predicated on the same model; the hardware is basically sold at cost, while the “platform” last year had $417 million in revenue and $296 million in profit, which equates to a tidy 71% gross margin.

Apple TV Channels is a means to buy subscriptions to other streaming services, which makes a lot of money for Roku and Amazon in particular; Apple TV+ content is a reason to make Apple TV the default interface for video leading to more subscriptions via Apple TV Channels.2 This view also explains why Apple is going to bundle a year of Apple TV+ with all new Apple device purchases (which is also very much in line with the idea of Apple giving up short-term revenue on its products — or incurring contra-revenue in this case — for long-term subscription revenue).

iPhone as a Service

It does feel like there is one more shoe yet to drop when it comes to Apple’s strategic shift. The fact that Apple is bundling a for-pay service (Apple TV+) with a product purchase is interesting, but what if Apple started including products with paid subscriptions?

That may be closer than it seems. It seemed strange yesterday’s keynote included an Apple Retail update at the very end of the keynote, but I think this slide explained why:

iPhone monthly pricing

Not only can you get a new iPhone for less if you trade in your old one, you can also pay for it on a monthly basis (this applies to phones without a trade-in as well). So, in the case of this slide, you can get an iPhone 11 and Apple TV+ for $17/month.

Apple also adjusted their AppleCare+ terms yesterday: now you can subscribe monthly and AppleCare+ will carry on until you cancel, just as other Apple services like Apple Music or Apple Arcade do. The company already has the iPhone Upgrade Program, that bundles a yearly iPhone and AppleCare+, but this shift for AppleCare+ purchased on its own is another step towards assuming that Apple’s relationship with its customers will be a subscription-based one.

To that end, how long until there is a variant of the iPhone Upgrade Program that is simply an all-up Apple subscription? Pay one monthly fee, and get everything Apple has to offer. Indeed, nothing would show that Apple is a Services company more than making the iPhone itself a service, at least as far as the customer relationship goes. You might even say it is innovative.

  1. Apple’s product numbers are always best represented on a trailing twelve-month basis given the huge amount of seasonality in their revenue []
  2. I also believe this is now the strategic rationale behind Amazon Prime Video []

What Is a Tech Company?

At first glance, WeWork and Peloton, which both released their S-1s in recent weeks, don’t have much in common: one company rents empty buildings and converts them into office space, and the other sells home fitness equipment and streaming classes. Both, though, have prompted the same question: is this a tech company?

Of course, it is fair to ask, “What isn’t a tech company?” Surely that is the endpoint of software eating the world; I think, though, to classify a company as a tech company because it utilizes software is just as unhelpful today as it would have been decades ago.

IBM and Tech-Centered Ecosystems

Fifty years ago, what is a tech company was an easy question to answer: IBM was the tech company, and everybody else was IBM’s customers. That may be a slight exaggeration, but not by much: IBM built the hardware (at that time the System/360), wrote the software, including the operating system and applications, and provided services, including training, ongoing maintenance, and custom line-of-business software.

All kinds of industries benefited from IBM’s technology, including financial services, large manufacturers, retailers, etc., and, of course, the military. Functions like accounting, resource management, and record-keeping automated and centralized activities that used to be done by hand, dramatically increasing the efficiency of existing activities and making new kinds of activities possible.

Increased efficiency and new business opportunities, though, didn’t make J.P. Morgan or General Electric or Sears tech companies. Technology simply became one piece of a greater whole. Yes, it was essential, but that essentialness exposed technology’s banality: companies were only differentiated to the extent they did not use computers, and then to the downside.

IBM, though, was different: every part of the company was about technology — indeed, IBM was an entire ecosystem onto itself: hardware, software, and services, all tied together with a subscription payment model strikingly similar to today’s dominant software-as-a-service approach. In short, being a tech company meant being IBM, which meant creating and participating in an ecosystem built around technology.

Venture Capital and Zero Marginal Costs

The story of IBM handing Microsoft the contract for the PC operating system and, by extension, the dominant position in computing for the next fifteen years, is a well-known one. The context for that decision, though, is best seen by the very different business model Microsoft pursued for its software.

What made subscriptions work for IBM was that the mainframe maker was offering the entire technological stack, and thus had reason to be in direct ongoing contact with its customers. In 1968, though, in an effort to escape an antitrust lawsuit from the federal government, IBM unbundled their hardware, software, and services. This created a new market for software, which was sold on a somewhat ad hoc basis; at the time software didn’t even have copyright protection.

Then, in 1980, Congress added “computer program” to the definition list of U.S. copyright law, and software licensing was born: now companies could maintain legal ownership of software and grant an effectively infinite number of licenses to individuals or corporations to use that software. Thus it was that Microsoft could charge for every copy of Windows or Visual Basic without needing to sell or service the underlying hardware it ran on.

This highlighted another critical factor that makes tech companies unique: the zero marginal cost nature of software. To be sure, this wasn’t a new concept: Silicon Valley received its name because silicon-based chips have similar characteristics; there are massive up-front costs to develop and build a working chip, but once built additional chips can be manufactured for basically nothing. It was this economic reality that gave rise to venture capital, which is about providing money ahead of a viable product for the chance at effectively infinite returns should the product and associated company be successful.

Indeed, this is why software companies have traditionally been so concentrated in Silicon Valley, and not, say, upstate New York, where IBM was located. William Shockley, one of the inventors of the transistor at Bell Labs, was originally from Palo Alto and wanted to take care of his ailing mother even as he was starting his own semiconductor company; eight of his researchers, known as the “traitorous eight”, would flee his tyrannical management to form Fairchild Semiconductor, the employees of which would go on to start over 65 new companies, including Intel.

It was Intel that set the model for venture capital in Silicon Valley, as Arthur Rock put in $10,000 of his own money and convinced his contacts to add an additional $2.5 million to get Intel off the ground; the company would IPO three years later for $8.225 million. Today the timelines are certainly longer but the idea is the same: raise money to start a company predicated on zero marginal costs, and, if you are successful, exit with an excellent return for shareholders. In other words, it is the venture capitalists that ensured software followed silicon, not the inherent nature of silicon itself.

To summarize: venture capitalist fund tech companies, which are characterized by a zero marginal cost component that allows for uncapped returns on investment.

Microsoft and Subscription Pricing

Probably the most overlooked and underrated era of tech history was the on-premises era dominated by software companies like Microsoft, Oracle, and SAP, and hardware from not only IBM but also Sun, HP, and later Dell. This era was characterized by a mix of up-front revenue for the original installation of hardware or software, plus ongoing services revenue. This model is hardly unique to software: lots of large machinery is sold on a similar basis.

The zero marginal cost nature of software, however, made it possible to cut out the up-front cost completely; Microsoft started pushing this model heavily to large enterprise in 2001 with version 6 of its Enterprise Agreement. Instead of paying for perpetual licenses for software that inevitably needed to be upgraded in a few years, enterprises could pay a monthly fee; this had the advantage of not only operationalizing former capital costs but also increasing flexibility. No longer would enterprises have to negotiate expensive “true-up” agreements if they grew; they were also protected on the downside if their workforce shrunk.

Microsoft, meanwhile, was able to convert its up-front software investment from a one-time payment to regular payments over time that were not only perpetual in nature (because to stop payment was to stop using the software, which wasn’t a viable option for most of Microsoft’s customers) but also more closely matched Microsoft’s own development schedule.

This wasn’t a new idea, as IBM had shown several decades earlier; moreover, it is worth pointing out that the entire function of depreciation when it comes to accounting is to properly attribute capital expenditures across the time periods those expenditures are leveraged. What made Microsoft’s approach unique, though, is that over time the product enterprises were paying for was improving. This is in direct contrast to a physical asset that deteriorates, or a traditional software support contract that is limited to a specific version.

Today this is the expectation for software generally: whatever you pay for today will be better in the future, not worse, and tech companies are increasingly organized around this idea of both constant improvements and constant revenue streams.

Salesforce and Cloud Computing

Still, Microsoft products had to actually be installed in the first place: much of the benefit of Enterprise Agreements accrued to companies that had already gone through that pain.

Salesforce, founded in 1999, sought to extend that same convenience to all companies: instead of having to go through long and painful installation processes that were inevitably buggy and over-budget, customers could simply access Salesforce on Salesforce’s own servers. The company branded it “No Software”, because software installations had such negative connotations, but in fact this was the ultimate expression of software. Now, instead of one copy of software replicated endlessly and distributed anywhere, Salesforce would simply run one piece of software and give anyone anywhere access to it. This did increase fixed costs — running servers and paying for bandwidth is expensive — but the increase was more than made up for by the decrease in upfront costs for customers.

This also increased the importance of scale for tech companies: now not only did the cost of software development need to be spread out over the greatest number of customers, so did the ongoing costs of building and running large centralized servers (of course Amazon operationalized these costs as well with AWS). That, though, became another characteristic of tech companies: scale not only pays the bills, it actually improves the service as large expenditures are leveraged across that many more customers.

Atlassian and Zero Transaction Costs

Still, Salesforce was still selling to large corporations. What has changed over the last ten years in particular is the rise of freemium and self-serve, but the origins of this model go back a decade earlier.

The early 2000s were a dire time in tech: the bubble had burst, and it was nearly impossible to raise money in Silicon Valley, much less anywhere else in the world — including Sydney, Australia. So, in 2001, when Scott Farquhar and Mike Cannon-Brookes, whose only goals was to make $35,000 a year and not have to wear a suit, couldn’t afford a sales force for the collaboration software they had developed called Jira they simply put it on the web for anyone to trial, with a payment form to unlock the full program.

This wasn’t necessarily new: “shareware” and “trialware” had existed since the 1980s, and were particularly popular for games, but Atlassian, thanks to being in the right place (selling Agile project management software) at the right time (the explosion of Agile as a development methodology) was using essentially the same model to sell into enterprise.

What made this possible was the combination of zero marginal costs (which meant that distributing software didn’t cost anything) and zero transaction costs: thanks to the web and rudimentary payment processors it was possible for Atlassian to sell to companies without ever talking to them. Indeed, for many years the only sales people Atlassian had were those tasked with reducing churn: all in-bound sales were self-serve.

This model, when combined with Salesforce’s cloud-based model (which Atlassian eventually moved to), is the foundation of today’s SaaS companies: customers can try out software with nothing more than an email address, and pay for it with nothing more than a credit card. This too is a characteristic of tech companies: free-to-try, and easy-to-buy, by anyone, from anywhere.

The Question of the Real World

So what about companies like WeWork and Peloton that interact with the real world? Note the centrality of software in all of these characteristics:

  • Software creates ecosystems.
  • Software has zero marginal costs.
  • Software improves over time.
  • Software offers infinite leverage.
  • Software enables zero transaction costs.

The question of whether companies are tech companies, then, depends on how much of their business is governed by software’s unique characteristics, and how much is limited by real world factors. Consider Netflix, a company that both competes with traditional television and movie companies yet is also considered a tech company:

  • There is no real software-created ecosystem.
  • Netflix shows are delivered at zero marginal costs without the need to pay distributors (although bandwidth bills are significant).
  • Netflix’s product improves over time.
  • Netflix is able to serve the entire world because of software, giving them far more leverage than much of their competition.
  • Netflix can transact with anyone with a self-serve model.

Netflix checks four of the five boxes.

Airbnb, which has yet to go public, is also often thought of as a tech company, even though they deal with lodging:

  • There is a software-created ecosystem of hosts and renters.
  • While Airbnb’s accounting suggests that its revenue has minimal marginal costs, a holistic view of Airbnb’s market shows that the company effectively pays hosts 86 percent of total revenue: the price of an “asset-lite” model is that real world costs dominate in terms of the overall transaction.
  • Airbnb’s platform improves over time.
  • Airbnb is able to serve the entire world, giving it maximum leverage.
  • Airbnb can transact with anyone with a self-serve model.

Uber, meanwhile, has long been mentioned in the same breath as Airbnb, and for good reason: it checks most of the same boxes:

  • There is a software-created ecosystem of drivers and riders.
  • Like Airbnb, Uber reports its revenue as if it has low marginal costs, but a holistic view of rides shows that the company pays drivers around 80 percent of total revenue; this isn’t a world of zero marginal costs.
  • Uber’s platform improves over time.
  • Uber is able to serve the entire world, giving it maximum leverage.
  • Uber can transact with anyone with a self-serve model.

A major question about Uber concerns transaction costs: bringing and keeping drivers on the platform is very expensive. This doesn’t mean that Uber isn’t a tech company, but it does underscore the degree to which its model is dependent on factors that don’t have zero costs attached to them.

Now for the two companies with which I opened the article. First, WeWork (which I wrote about here and here):

  • WeWork claims it has a software-created ecosystem that connect companies and employees across locations, but it is difficult to find evidence that this is a driving factor for WeWork’s business.
  • WeWork pays a huge percentage of its revenue in rent.
  • WeWork’s offering certainly has the potential to improve over time.
  • WeWork is limited by the number of locations it builds out.
  • WeWork requires a consultation for even a one-person rental, and relies heavily on brokers for larger businesses.

Frankly, it is hard to see how WeWork is a tech company in any way.

Finally Peloton (which I wrote about here):

  • Peloton does have social network-type qualities, as well as strong gamification.
  • While Peloton is available as just an app, the full experience requires a four-figure investment in a bike or treadmill; that, needless to say, is not a zero marginal cost offering. The service itself, though, is zero marginal cost.
  • Peloton’s product improves over time.
  • The size, weight, and installation requirements for Peloton’s hardware mean the company is limited to the United States and the just-added United Kingdom and Germany.
  • Peloton has a high-touch installation process

Peloton is also iffy as far these five factors go, but then again, so is Apple: software-differentiated hardware is in many respects its own category. And, there is one more definition that is worth highlighting.

Peloton and Disruption

The term “technology” is an old one, far older than Silicon Valley. It means anything that helps us produce things more efficiently, and it is what drives human progress. In that respect, all successful companies, at least in a free market, are tech companies: they do something more efficiently than anyone else, on whatever product vector matters to their customers.

To that end, technology is best understood with qualifiers, and one of the most useful sets comes from Clayton Christensen and The Innovator’s Dilemma:

Most new technologies foster improved product performance. I call these sustaining technologies. Some sustaining technologies can be discontinuous or radical in character, while others are of an incremental nature. What all sustaining technologies have in common is that they improve the performance of established products, along the dimensions of performance that mainstream customers in major markets have historically valued. Most technological advances in a given industry are sustaining in character…

Disruptive technologies bring to a market a very different value proposition than had been available previously. Generally, disruptive technologies underperform established products in mainstream markets. But they have other features that a few fringe (and generally new) customers value. Products based on disruptive technologies are typically cheaper, simpler, smaller, and, frequently, more convenient to use.

Sustaining technologies make existing firms better, but it doesn’t change the competitive landscape. By extension, if adopting technology simply strengthens your current business, as opposed to making it uniquely possible, you are not a tech company. That, for example, is why IBM’s customers were no more tech companies than are users of the most modern SaaS applications.

Disruptive technologies, though, make something possible that wasn’t previously, or at a price point that wasn’t viable. This is where Peloton earns the “tech company” label from me: compared to spin classes at a dedicated gym, Peloton is cheap, and it scales far better. Sure, looking at a screen isn’t as good as being in the same room with an instructor and other cyclists, but it is massively more convenient and opens the market to a completely new customer base. Moreover, it scales in a way a gym never could: classes are held once and available forever on-demand; the company has not only digitized space but also time, thanks to technology. This is a tech company.

This definition also applies to Netflix, Airbnb, and Uber; all digitized something essential to their competitors, whether it be time or trust. I’m not sure, though, that it applies to WeWork: to the extent the company is unique it seems to rely primarily on unprecedented access to capital. That may be enough, but it does not mean WeWork is a tech company.

And, on the flipside, being a tech company does not guarantee success: the curse of tech companies is that while they generate massive value, capturing that value is extremely difficult. Here Peloton’s hardware is, like Apple’s, a significant advantage.

On the other hand, asset-lite models, like ride-sharing, are very attractive, but can Uber capture sufficient value to make a profit? What will Airbnb’s numbers look like when it finally IPOs? Indeed, the primary reason Peloton’s numbers look good is because they are selling physical products, differentiated by software, at a massive profit!

Still, definitions are helpful, even if they are not predictive. Software is used by all companies, but it completely transforms tech companies and should reshape consideration of their long-term upside — and downside.

I wrote a follow-up to this article in this Daily Update.