Sunday, August 2, 2020

Stay away from popular tech stocks

Today, America's five largest companies by market capitalization are all well-known technology and internet businesses: in descending order, Apple, Amazon, Microsoft, Alphabet (the parent company of Google), and Facebook. Historically, being one of the top companies by market cap has been a contrary indicator, both for the company itself and for the industry to which it belongs. In 1980, right before a decade-long decline in oil prices, six of the top ten were oil companies. In 2000, at the peak of the dotcom bubble, six of the top ten were computer and internet companies.

I think history will repeat and that none of today's Big Five will grow enough to justify its current market cap. So I want to lay out the risks, as I see them, of investing in these companies and popular tech stocks in general.

The Economic Risks

As Facebook and Google have grown, they've become more exposed to the overall economy. During the 2007-09 recession, online advertising was a relatively small part of the total ad market and was steadily gaining market share from other advertising media. These share gains offset the recession's effects and allowed them to keep growing despite broader economic turmoil.

Today Facebook and Google are huge advertising players. In 2017, one survey put their worldwide market share at 20% and another suggested it was as high as 25%. The percentages are presumably even higher today. And as impressive as those figures are, they actually understate the extent to which Facebook and Google have saturated their niche, because online ads are not well-suited to certain kinds of advertising.

As "Ad Contrarian" argues, online ads are particularly ill-suited to brand-building. Much of the appeal of brands is their signaling value. For instance, Nike spends a lot of money on advertising that associates its products with athletic prowess. So people buy Nike apparel because it's a way of showing that they're committed athletes, if only as amateurs. And for this kind of advertising to be effective, "it's not enough for it to be seen by a single person or even by many people. Someone has to know that everyone else has seen it, too." Online advertising generally isn't capable of offering that because it tends to be personalized, for example ads that are generated in response to a search query.

So online advertising's total addressable market may be smaller than investors think, and accordingly Facebook and Google may be closer to saturating it. And the purpose for which online ads are most effective--targeted "direct response" advertising--is particularly attractive to local businesses and niche vendors that want to reach a specific audience. Many of these small businesses have suffered immensely during the COVID-19 epidemic, and some are likely to close permanently.

And they aren't the only online-advertising customers who are at risk of failing. The venture capitalist Chamath Palihapitiya, in his 2018 shareholder letter, claimed that "startups spend almost 40 cents of every VC dollar on Google, Facebook, and Amazon." Most of these startups are unprofitable, some structurally so. Even if Facebook and Google have great business models, many of their customers don't.

Amazon's retail operation, like Facebook's and Google's advertising platforms, is far more mature now than it was during prior recessions and thus more exposed to economic fluctuations. That is a risk because the retail operation has negative working capital: after a customer purchases something on Amazon, it pays its supplier with a significant lag. The lag means that suppliers are effectively financing Amazon, and that, combined with low spending on property and equipment--since as an online retailer, Amazon doesn't need to invest in physical stores--means that its cash balance increases as its revenue grows. Negative working capital is considered a very attractive trait for businesses, and it is most of the time. The exception is when revenue declines, in which instance cash inflows turn into potentially large cash outflows.

Regarding Amazon as a retailer, investors also need to consider the possibility that online commerce has made retailing structurally more competitive and less profitable by increasing price transparency. In a 2018 academic paper, Alberto Cavallo states that "in the past 10 years online competition has raised both the frequency of price changes and the degree of uniform pricing across locations." As a commenter on Value Investors Club writes, "Many innovations have destroyed industries without capturing a fraction of the value destroyed. The surplus goes to consumers and is one reason for increased standards of living in a free-market economy."

Amazon trades at a high price/earnings multiple because it has trained investors to think that it minimizes current earnings in favor of growth. Maybe the minimal earnings are not a choice but structural feature of online retailing.

The Cloud: Actually a Bubble

"Cloud computing," the provisioning of computing services over the internet, is one of the fastest-growing parts of the technology sector. Traditionally, companies that wanted to host websites, store computer data, or run software applications bought and maintained their own computers for those purposes. Cloud computing lets them outsource all of these functions to "cloud providers" that have more resources and computing expertise. 

The largest cloud provider is Amazon's Amazon Web Services (AWS), which has been growing at 30% per year. Microsoft's Azure, the second-largest provider, has grown at 50% per year. Google's Google Cloud Platform, the third-largest, has grown more than 40%. These growth rates--and the expectation that they will continue for many years as cloud computing becomes more prevalent--have convinced investors to grant high valuations to these and other companies that are involved in the cloud. Amazon devoted significant resources to developing AWS before most investors realized how pervasive cloud computing would become, and this has made it a particular favorite of investors. They see AWS not just as a leader in an attractive sector but as a reflection of Amazon CEO Jeff Bezos's shrewdness and foresight.

I think that cloud computing is over-hyped and that the market's ultimate size and profitability are smaller than investors realize., a news and discussion site for tech workers, has many comment threads devoted to cloud computing. Commenters on the site frequently state that using the cloud is far more expensive than using one's own computers and colocation hosting. Here are some representative comments:

Even the cheapest cloud (seems to be Google at the moment) is 5x or more of the price of an equivalent dedicated server (2x for "compute", >20x for bandwidth, so 5x is somewhat average.

I've done the math many times and it's orders of magnitude cheaper to colocate as long as you can afford an IT guy and the upfront cost of hardware.

[C]loud bandwidth is insanely expensive. For example, Hetzner offers 1.3$/TB (if you happen to exceed their generous 30 TB quota). In comparison, Amazon is 70x more expensive at 90$/TB.

There is no economy of scale with the cloud. As you scale up your service, Amazon/Google happily scale your bill right up with it. When you roll your own datacenter, things get much cheaper per unit (bandwidth, storage, etc) as you get larger.

Dedicated servers, and colocation are going to be far cheaper than the cloud, and worse, the savings directly relate to the size of the infrastructure you need.

So why do companies use cloud providers if they're so much more expensive? ycombinator commenters suggest several reasons:

1. The company's use of computing power is highly variable, so it needs to be able to increase its computing resources on short notice. For instance, an online retailer that receives a flood of orders during the holiday season, or that has a fad product that suddenly becomes popular, may use the cloud to ensure that it can keep its website running smoothly as usage surges.

2. The company is growing quickly, either because it's a successful startup or because it's an established company that's significantly increasing its use of the internet and computing power, and cloud providers let it add capacity almost instantaneously. One commenter states that the cloud "makes it a lot easier for a small startup to build out their services without thinking about scaling challenges."

3. Many corporate information-technology departments are inefficient and unresponsive. They may take weeks to fulfill simple requests for new equipment or software. Using a cloud provider lets corporate managers bypass the IT bureaucracy.

4. Cloud providers let companies replace capital expenditures with operating expenditures. This can appeal to both small businesses that are strapped for cash and publicly-traded companies that want to massage their financial statements.

5. Corporate executives with no technical expertise want to use a cloud provider because it's "the new thing" and they aren't aware of the relative cost. One commenter writes, "[a] lot of companies shouldn't ever have scaling issues and are buying into the cloud to be hip. Just like Hadoop and other big data technologies they didn’t need in 2014."

1 is a valid reason to use the cloud, but it is a niche use, and the potential demand for cloud computing as swing capacity is too small to justify the huge valuations that investors have awarded Amazon etc. 

ycombinator commenters suggest that 2 and 3 are the main reasons why cloud computing is popular. If so, then cloud providers are vulnerable to a slowdown of computing growth. If the cloud is more flexible than internally-maintained computing but much costlier, then companies are likely to prefer the cloud when they are growing their computing power rapidly, but when their growth slows, they should find the lower cost of do-it-yourself more appealing. Likewise, the IT bureaucracy's inefficiency is less of a problem when corporate managers have less need to grow and acquire and install new equipment. One commenter writes, "The cloud isn't a money-saving tactic. It enables you to test demand for a service for a very low absolute cost... As soon as you can confidently predict sufficient demand, the economically rational decision is to hire a good ops team and run real hardware."

Demand from 4 and 5 is hard to quantify but vulnerable to changes in perception. During a recession, or even a growth slowdown, companies typically look for ways to reduce their costs. A company that goes with the cloud because it was the easiest and most visible choice may rethink that decision in a more austere environment.

Many investors assume that cloud computing is a secular trend: corporate use of the internet and computing is quickly increasing, and cloud providers should grow even more quickly as they gain market share from do-it-yourself. Investors need to consider the alternative possibility that, rather than being a play on total demand for computing/internet services, the cloud is a play on the growth of computing demand. And if that growth rate falls, even if overall computing usage continues to rise, then the total addressable market for cloud hosting may be much smaller than commonly assumed.

Capital Allocation and Short-Termism

Amazon, Apple, Facebook, and Google grew rapidly by displacing older firms: newspapers, department stores and mall retailers, Nokia and Research in Motion. As a result, they have corporate cultures that are oriented toward growth. With the older firms vanquished, and Amazon etc. dominating their respective markets, there is a risk that they will try to continue growing by investing in crowded fields where they lack any competitive advantage. That's happening in streaming video, where Amazon and Apple are spending billions to compete with Netflix, Disney, HBO, and others. 

Being a monopoly and having high margins doesn't guarantee good shareholder returns. The monopolist may squander its profits by making bad acquisitions, wasting money on unfocused capex and research, or buying back overvalued shares. In Warren Buffett's Agony and Ecstasy speech, the company he used to illustrate agony was the telephone monopoly AT&T, which, "at the end of 1979, was selling for $10 billion less than the shareholders had either put in or left in the business."

Every one of the Big Five--and especially Amazon, Facebook and Google--gives its employees stock compensation that's significant relative to both the company's operating income and the recipients' base salaries. I believe that this stock comp encourages executives at these companies to prioritize a high stock price in the short term over the long-term health of the business.

Amazon's executives get a particularly high share of their compensation as stock, and Amazon is also the company for which there's the most anecdotal evidence of short-termism. Vox claims that "Amazon is stuffing its search results pages with" sponsored listings, while The Wall Street Journal claims that "Amazon changed [a] search algorithm in ways that boost its own products," and Bloomberg reports that Nike has stopped selling its products directly on Amazon after struggling with counterfeits that third-party merchants were selling on the site.

Amazon's worst decision has been allowing large numbers of Chinese vendors to join the site as third-party merchants. Mainland China has extremely low social trust, and Chinese merchants who sell to US customers are beyond the reach of the US judicial system. So unsurprisingly, many of Amazon's new Chinese merchants are selling counterfeit and defective products.

There are anecdotes across the internet suggesting that Amazon has become less competitive on price. That has been my experience. For years, I bought two niche products on Amazon that I assumed would be hard to find in physical stores. In the past year, both products' categories have been inundated with fakes on Amazon, which prompted me to look for other places to buy them. To my surprise, I found both at a local physical health/beauty retailer: one cost slightly more at the physical retailer and the other was 20% cheaper. 

Many Americans do all their online shopping at Amazon because over time it developed a reputation for reliability and low prices. Knowing that Amazon's products were good and competitively priced reduced the "search costs" of shopping on the Internet. Between bad Chinese merchants, sponsored listings that reduce the accuracy of search results, and higher prices, Amazon is increasing its customers' search costs. Trust is hard to gain and easy to lose, and Amazon is at risk of losing their trust.

The Political Risks

The Big Five face three large and growing political risks: nationalism, populism, and resentful competitors. 

First, nationalism: If I were a politician in Brazil, I would wonder why a couple of companies in California are allowed to earn enormous profits connecting Brazilian businesses to Brazilian consumers. I expect that, over time, countries that are large enough and have enough intellectual capital to develop native internet champions will aggressively regulate American tech giants in order to privilege the local champions. Amazon has invested more than $5 billion in India, and in the past couple years the Indian government has passed laws and regulations targeting Amazon and other foreign companies operating in the country.

Second, populism: The Big Five have created a lot of wealth, but most of that wealth has gone to a relatively small group of founders, highly-skilled employees, venture capitalists, and shareholders. Many large tech companies outsource their menial work, meaning that low-level workers don't share in the companies' profits and lack opportunities for advancement. Relevant to this, The New York Times published an interesting article contrasting the employment practices of Kodak during its glory days with employment at Apple today. 

Third, resentful competitors: The Big Five's scale and ubiquity give them power over smaller tech companies and startups. Promising startups may face pressure to sell out to a tech giant, with the threat that otherwise it will drive the startup out of business. After Snapchat created a social-media platform with novel features, Facebook expressed an interest in acquiring it. When Snapchat rebuffed Facebook, it proceeded to copy many of Snapchat's features on Instagram.

So far most investors have ignored these political risks. They are being too blase. Facebook benefited significantly from acquiring Instagram, which has continued to grow robustly as Facebook's user base in the U.S. and Canada has essentially stopped growing. A similar acquisition today would not be politically feasible. And the European Union has levied €8.2bn of fines against Google over the past few years.

The tech giants' size guarantees that, if the political environment becomes more hostile, there are numerous things that politicians and regulators can use to justify punishing them with onerous new regulations. For example, Google tweaks its search results in various ways, and according to The Wall Street Journal, "In one change hotly contested within Google, engineers opted to tilt results to favor prominent businesses over smaller ones, based on the argument that customers were more likely to get what they wanted at larger outlets." In another article, the Journal claims that at Amazon, contrary to claims it has made before Congress, "employees have used data about independent sellers on the company's platform to develop competing products."

Speculation as a Service

So far I've written exclusively about the five largest technology and internet companies, but software as a service (SaaS) also deserves a mention. Similar to cloud hosts, SaaS companies offer their services remotely: unlike traditional software that customers install on their own computers, SaaS runs on the vendor's own servers, and customers access it over the internet.

While SaaS companies are all smaller than the Big Five, they are far more numerous, and as a group they are big part of the tech/internet sector's aggregate market capitalization. They have also been among the stock market's best performers. But in contrast to their stock performance, I believe that SaaS vendors offer poor economic returns to public-market investors.

SaaS vendors typically have very high gross margins, but most of that margin gets captured by employees, notably in the form of stock compensation. Salesforce (CRM), which is one of the leading SaaS companies, had a gross margin of 74% in the financial quarter ended April 30, 2020, but its operating margin was slightly negative. That isn't a fluke; other large SaaS companies like Workday and Servicenow have similar financial profiles. I believe that two things give SaaS employees significant bargaining power:

1. Few people have software development and engineering skills, and even fewer are elite programmers, so their skills have scarcity value.
2. Most tech/software companies are concentrated in Silicon Valley and a few other places like Seattle. The geographic concentration makes switching employers easy.

A few technology companies like Facebook and Google are able to pay significant compensation and still earn high profits because they've established lucrative monopolies. So the question a SaaS investor has to ask is: how many SaaS companies will achieve that kind of monopoly? Judging by their valuations--the companies typically trade above 10x sales--the answer is "most of them."

In part, investors have awarded SaaS vendors high valuations because the first generation of software behemoths like Microsoft, Oracle, and SAP did achieve lucrative monopolies. All three companies have products with high switching costs, and Microsoft's Windows and Office also have significant network effects. So the assumption is that many of today's SaaS offerings will replicate that success. But the comparison is inapt.

When Microsoft, Oracle, and SAP became tech superstars in the 1980s and early 1990s, the kinds of products they sold were unprecedented. The power of computer hardware had risen, and its cost had fallen, enough that for the first time companies could cost-effectively run advanced business software. And since most of the companies that bought such software had never used anything like it before, they didn't consider the possibility that they would be "locked in" to using Oracle or SAP. They were just happy to acquire software that could significantly reduce their costs and increase their efficiency and capabilities. 

Today every large company has experience buying software and understands the risk of being locked in, and they will do what they can to avoid it. In an interview with Patrick O'Shaughnessy, Hamilton Helmer articulates this argument:

It's very rare, once you have a business up and running, to introduce switching costs, because what happens is that customers at that point are smart and they realize what you're up to... It's hard to establish it after the fact. And so that's why usually for switching-costs businesses, you need to get customers during the take-off phase, when things are growing so rapidly that customers are just worried about "Can I get the product?" and they're not worried about all the things associated with it, like switching costs.

And unlike Microsoft, Oracle, and SAP during their glory days, most SaaS is not revolutionary. It's an incremental improvement over existing software, not the kind of boon that would cause software customers to throw caution to the wind.

Also, in the first generation of software, there were numerous losers for every winner. For every Oracle, there was an Ingres, an Informix, a Sybase. I don't expect many SaaS vendors to achieve monopolies with high switching costs, but even if some do, there will be corresponding losers. Yet every SaaS stock is priced for ultimate success.

Which brings me to one final point: In 1994, when Microsoft was the leader in operating systems but had yet to vanquish IBM's OS/2 and completely dominate the market, it traded at 5-6x sales net of excess cash and was nicely profitable. In the same year, Oracle traded at 5x sales and also had a high profit margin. Contrast that with SaaS vendors trading at 10-20x sales with minimal earnings.