The Platform Economy

The Platform Economy
Executive Summary
December 2020 The technology platform has emerged as the preeminent business model a er many
years in ascent. We use natural language processing to identify platform companies
Kai Wu and show that they have significantly outperformed the stock market. Platforms’
Founder & Chief Investment Officer powerful network effects generate positive feedback and monopoly dynamics, which
kai@sparklinecapital.com are disrupting traditional valuation approaches.
The Rise of the Platform Exhibit 2

The Platform Parade 🎉
March of the Unicorns 🦄
Last week, Airbnb and DoorDash had their highly anticipated
initial public offerings (IPOs). DoorDash went first. The IPO
priced at the top of the range and rocketed +86% on the first
day of trading to a $72 billion valuation. Airbnb followed the
next day to even greater euphoria, jumping +113% to a $100
billion valuation. Even Airbnb’s CEO was le speechless.
Exhibit 1 Source: S&P, Sparkline

And There Was Much Rejoicing!
Despite all the fanfare, these newly public companies are
dwarfed by their platform predecessors. The five largest US
companies (Apple, Microso , Amazon, Google, Facebook) all
operate extremely successful technology platforms. These
tech giants represent $7.4 trillion in market cap and 22% of
the S&P 500. The unique features of the platform model
have helped these firms reach an unprecedented scale.
The “platformization" of the stock market is not just about
splashy IPOs and tech giants. The number of platforms in
the top 500 US stocks has steadily grown from 40 to 100.
Exhibit 3
It Feels Good to Be a Platform
Source: Pitchbook, Bloomberg, Sparkline
While operating in very different industries, these companies
employ a common business model. They are both platform
companies. Platform companies externalize the means of
production. They do not own homes or employ drivers.
Instead, they profit from orchestrating networks of external
consumers and producers.
This marks only one of many milestones in the platform
parade. Many iconic platform companies such as Uber and
Slack went public in the past couple years. Even Warren
Buffett, who publicly eschewed IPOs for decades, invested
$735 million in Snowflake’s public debut. Source: SEC, Sparkline
1

The Platform Economy | Dec 2020

We expect this trend to continue. There are 500 private Exhibit 5
companies valued over $1 billion. Many of these “unicorns” Platform Taxonomy
are platform businesses ripe to join the public markets in the
next several years.

Exhibit 4
Platform Unicorns
Source: Sparkline

It is important to note that we use the term “platform” to
mean the business model. This is distinct from the term
“technology platform,” which describes pretty much every
Source: CBInsights, Sparkline so ware company. E-commerce companies such as Chewy
are retailers that happen to utilize technology platforms for
more efficient distribution over the internet. Similarly, SaaS
The Many Shapes of Platforms firms are just product companies that use the cloud to
Platforms can take many shapes. They can connect a single streamline deployment.
type of user (Facebook) or different types (iOS App Store).
Producers and consumers can be the same individuals Why Now?
(LinkedIn) or different (YouTube). The item exchanged can
The platform business model is not new. Marketplaces have
range from rides to photos to so ware applications.
existed since ancient times to facilitate the exchange of

crops, cattle, spices, commodity futures, stocks, and other
In order to make sense of this wonderful diversity, we’ll use
goods. Even the shopping mall is a type of platform. So why
the taxonomy created by Cucumano, Gawar, and Yoffie
are platform companies only having their moment now?
(2018) to categorize platforms into two broad types:

💱 Transaction platforms serve as intermediaries for
The value of platforms is a nonlinear function of the size of
their networks (more on this later). Thus, their potential has
direct exchanges or transactions between users
increased exponentially as globalization and technology
🔬 Innovation platforms provide a technological have connected the world. Railroads, electricity, air travel,
foundation upon which other firms develop telephones, the Internet, and smartphones have made the
complementary innovations world a much smaller place. The ancient spice market may
have at best connected thousands of merchants. Today,
The next exhibit shows a few examples of platforms and how Facebook alone serves 2.7 billion active monthly users.
they fit into this taxonomy.
The modern digital platform has taken this primordial
concept to epic proportions. The dominance of firms such as
Google and Amazon is just another of the many side effects
of the megatrends of globalization and technology.

2


Platform Stock Returns However, the biggest problem with human-compiled lists is
the introduction of survivorship bias. While it is easy to think
of success stories, we have a tendency to forget the losers.
Identifying Platforms
The historical performance of platform companies will be
So how have the stocks of platforms done? In order to grossly overstated if these companies are selected with the
conduct a rigorous study, we need a way to classify platform benefit of hindsight.
companies. Despite their dominant economic position,
empirical work on platform companies is extremely limited.
Company Embeddings
Most writing on platforms is tailored toward entrepreneurs,
managers, and venture investors. We can address these challenges using machine learning.

The few attempts to study the empirical characteristics of In Investment Management in the Machine Learning Age, we
publicly traded platform companies begin with manually introduced “company embeddings,” which we used to
defined lists. The next exhibit compares the constituents of compute the semantic distance between descriptions of
three such lists. For consistency, we only include stocks that companies’ businesses in their 10-K filings. In Value
were in the S&P 500 in 2015 (when the lists were compiled). Investing Is Short Tech Disruption, we used this technique to
identify companies that are described similarly to the FAANG
Exhibit 6 stocks. While we could use this approach to find failed
Identifying Platforms Is Subjective platform companies, it would still require manually defining
a list of platform companies.

Thus, we will instead introduce a new way to use company
embeddings. The key insight is that while we have so far
only compared companies’ 10-K business descriptions to
each other, there is nothing preventing us from comparing
these passages to any arbitrary text document.

We will compare the 10-Ks to two articles from platform
experts Eisennman, Parker and Van Alstyne (2006, 2007). We
use both articles as one deals with transaction platforms
and the other innovation platforms. We also mask company
names so that our algorithm focuses only on the general
features of the platform business model itself.
Exhibit 7
Platform Articles for Calculating Similarity
Source: Cusumano, Gawar, and Yoffie (2018), Moazed and Johnson (2016),
Ravichandran, Lu and Lee (2017), Sparkline

It is clear that identifying platforms is subjective. While eBay
and Facebook are clearly platforms, there is less consensus
around innovation platforms such as Oracle and Adobe.
Researchers also seem to be less interested in old-economy
platforms such as financial exchanges and credit card
companies despite their ample network effects. Determining
where to draw the line for hybrid companies such as Apple
that operate both platforms and traditional businesses
introduces additional subjectivity. Source: Harvard, MIT

3


We deliberately chose older articles to avoid hindsight bias. Exhibit 9
When they were published, the authors had no idea that Examples of Platforms Identified
Facebook and Amazon would go on to run the world. Also,
note that we could have chosen from many other writings
on platforms and our results would have been similar.

The next exhibit shows the similarity scores of excerpts from
two recent 10-K filings to our platform articles. Platform
scores can range from a high of 1 to low of -1. Despite both
being technology companies, their annual filings reveal very
different business models. Uber is a platform company,
while Lam Research employs a traditional product model.
Source: SEC, Sparkline
Exhibit 8
Technology Is Not a Business Model In addition to stalwarts such as Microso and PayPal, the
embeddings dig up newer platforms such as Match Group
Uber Technologies (online dating), Slack (enterprise messaging), and Zoom
We are a technology platform that uses a massive network, leading (video conferencing).
technology, operational excellence and product expertise to power
movement from point A to point B. We develop and operate Since the algorithm is run on a rolling basis, we can track
proprietary technology applications supporting a variety of offerings how companies’ business models evolve over time. Amazon
on our platform (“platform(s)” or “Platform(s)”). We connect
started as an online bookstore. However, over time it added
consumers (“Rider(s)”) with independent providers of ride services
(“Rides Driver(s)”) for ridesharing services, and connect consumers platforms such as a third-party marketplace, Amazon Web
(“Eater(s)”) with restaurants (“Restaurant(s)”) and food delivery Services, Kindle Direct Publishing, and Alexa. Amazon’s
service providers (“Delivery People”) for meal preparation and similarity score has increased over time as these platforms
delivery services. have grown in importance.
Platform Score = 0.48
Exhibit 10
Lam Research The Platformization of Amazon
We are a global supplier of innovative wafer fabrication equipment
and services to the semiconductor industry. We have built a strong
global presence with core competencies in areas like nanoscale
applications enablement, chemistry, plasma and fluidics, advanced
systems engineering and a broad range of operational disciplines. Our
products and services are designed to help our customers build
smaller, faster, and better performing devices that are used in a variety
of electronic products, including mobile phones, personal computers,
servers, wearables, automotive vehicles, and data storage devices.
Platform Score = 0.06


We can apply this technique to the annual filings of all public
companies through time, saving us from having to read
through tens of thousands of pages of SEC filings. The next

exhibit shows some of the companies that our algorithm has
While we compute a continuous platform score for each
identified as platforms.
company, for the rest of the analysis we will use a simpler
binary categorization. We will employ a threshold of 0.33 to
balance the rate of false positives and negatives.
4


Platform Performance Exhibit 12
Platforms Have Outperformed
With our new definition of platform in hand, we can now
explore the characteristics of platform companies.

First, we find that the popularity of the platform business
model has been steadily rising. Among the top 500 US
stocks, the number of platforms has increased from 40 to
100. The percentage of market cap owned by platforms has
increased even more dramatically, swelling from 10 to 37%.

Exhibit 11
Platforms Rising
Source: S&P, SEC, Sparkline

This backtest uses a market cap weighting scheme, which
gives more weight to larger companies. In order to address
the concern that our results are driven solely by Big Tech, we
also create an equal-weighted version. This portfolio
delivers similar returns, confirming that the platform effect
is broad-based.

Exhibit 13
Equal-Weighted Platform Outperformance

The platform effect is concentrated in the communication
and technology sectors. 76% of platforms today are in these
industries. In addition, the share of these sectors that are
platforms has increased from 30 to 60% over this period.

Second, we compare the performance of platform stocks to
the rest of the largest 500 US stocks. The portfolio is
reconstituted each month in order to compute performance
the next month. We find that platform companies have
outperformed by an impressive 8.7% per year. Source: S&P, SEC, Sparkline

Third, we want to determine the source of platform profits.
We can decompose the returns of any investment strategy
into returns from sales growth, dividend yield and multiple
expansion. These components mathematically constitute
total performance. The next chart decomposes the returns
of platforms relative to non-platform companies.

5


Exhibit 14 All four groups have handily outperformed the market,
Platforms Fueled by Growth although smaller legacy platforms have lagged their larger
and newer platform brethren. Platforms have been a solid
investment regardless of vintage.

Why Have Platforms Thrived?

Platforms have managed to deliver strong growth through
two channels: disrupting existing industries and unlocking
new sources of value.

Disruption
The platform business model has been highly disruptive. As
the world has become more heavily networked, platforms
Source: S&P, SEC, Sparkline have been able to take market share from less efficient
incumbents. Uber and Ly are decimating the taxi industry;
This analysis reveals platforms have won by consistently Google and Facebook are squeezing the advertising
outgrowing the market. Interestingly, they did not suffer the industry; and Airbnb is shaking up the hotel industry.
losses from P/S contraction that would be expected if the
market “efficiently” priced in higher sales growth. In fact, In Value Investors Are Short Tech Disruption, we used
valuation change has not contributed meaningfully to natural language processing (NLP) to classify companies as
returns with the exception of the current year. disruptive or non-disruptive. As with platforms, our
disruption factor transcends the limitations of rigid industry
Finally, we want to test if performance is driven by startups classifications.
or mature firms. We divide our universe into companies that
were founded before and a er 1998. The older sample The next exhibit shows the overlap between platforms and
skews toward innovation platforms and includes Apple, disruption. We show the number of firms that fall into each
Amazon and Microso . The newer sample consists mainly of bucket with some examples (for illustrative purposes only).
transaction platforms and includes Google and Facebook.
We compute both cap- and equal-weighted versions. Exhibit 16
Platforms and Disruption
Exhibit 15
Mature and Startup Firms
Source: Sparkline

Source: S&P, SEC, Sparkline We find that not all disruptive companies are platforms, but
that most platforms are disruptive. In other words, while the
6


platform business model is disruptive, disruption can also By unifying China’s fragmented small business market, he
be driven by other forces such as technological innovation created a $680 billion behemoth.
(e.g., cloud computing, e-commerce).
Untapped Supply
How Platforms Create Value Similar to the way Jack Ma brought Chinese suppliers into
the global market, platform companies have managed to tap
Platforms have been most successful disrupting industries many previously idle sources of supply.
with the following characteristics:
YouTube, TikTok, and Twitch have launched a generation of
1. Inefficient gatekeepers
superstars that would likely not have had a platform in the
2. High fragmentation traditional entertainment world. Ly has helped increase
3. Untapped supply the utilization of cars that would otherwise sit idle in the
driveway. Airbnb has increased the utilization of spare
These are three areas where platforms are especially well rooms and homes. Etsy, 99designs, and the iOS App Store
suited to create value. have created opportunities for independent cra speople,
designers and developers.
Inefficient Gatekeepers
In order to access consumers, producers o en have to pass By replacing inefficient gatekeepers, aggregating farflung
through gatekeepers. Gatekeepers play an important role in markets, and tapping underutilized supply, platforms have
curation and quality control. However, they o en also have created robust communities and growth.
conflicts of interest and take a large cut of value. Moreover,
human gatekeepers do not scale, leading them to generally
ignore smaller producers who are not worth their time. Platform Monopolies

Rather than remove traditional gatekeepers, platforms have
replaced them. We have seen reintermediation rather than Network Effects
disintermediation. Platforms such as Spotify, Yelp, Kindle In order to understand why platforms have enjoyed such
Direct Publishing, and YouTube allow just about anyone to success, we must study their core asset, network effects.
create and distribute content. Platforms lower the barriers
to producer access and then solve the now greater curation Network effects are a phenomenon whereby additional
problem with algorithms and crowdsourced reviews. users enhance the value of a product to its existing users.
The mathematical value of network effects is commonly
Platforms not only broaden access but also tend to reduce modeled by Metcalfe’s Law, which states that the potential
transaction costs and friction. This helps grow the size of the number of connections in a network increases with the
overall market. For example, the ease of calling an Uber has square of the number of users (i.e., V ∝ n2 ).
grown the overall market for car services.
Exhibit 17
High Fragmentation Metcalfe’s Law
A highly fragmented market consists of many small,
dispersed participants. Fragmented markets suffer from a
lack of liquidity and limited network effects.

Platforms aggregate fragmented supply into a single, large
market. In building Alibaba, Jack Ma intentionally focused
on the fragmented Chinese exporter, saying: “Among the
great density of small companies here, most do not have
channels to the large trading companies. Most have no way
to reach a market. Simply by going through our Alibaba Source: SeaRates
network, they can get access to all of American and Europe.”
7


While not a law of nature, researchers have found that Positive Feedback
Metcalfe’s Law provides a good fit of the growth trajectories
of social networks such as Facebook and Tencent. Network effects generate positive feedback loops. As users
join a network, the value of that network increases. This
Metcalfe’s Law models the maximum number of potential entices more users to the platform, further increasing its
connections in a network. In practice, not all users will value. Meanwhile, as competing networks lose users, they
connect with each other and networks instead form around become less valuable, leading to further defections.
local clusters. Uber’s network consists of thousands of
isolated city-based clusters. Twitter has dozens of These virtuous and vicious cycles are a natural outgrowth of
sub-communities organized around topics such as finance the convexity of Metcalfe’s Law. This dynamic is common in
and climate change, each with its own influencers. the technology and communications industries. Famous
historical examples include 1980s video recorders (VHS vs.
The social graph below illustrates how distinct clusters tend Beta), 1990s PC operating systems (Windows vs. Apple), and
to form around the various communities in a Facebook 2000s social networks (Facebook vs. Myspace).
user’s network (e.g., high school, college, and work friends).
W. Brian Arthur (1996) famously coined the term “increasing
Exhibit 18 returns” to describe this positive feedback effect.
Real Life Facebook Friends
“Increasing returns are the tendency for that which is
ahead to get further ahead, for that which loses
advantage to lose further advantage. They are
mechanisms of positive feedback that operate — within
markets, businesses, and industries — to reinforce that
which gains success or aggravate that which suffers loss.
Increasing returns generate not equilibrium but
instability: If a product or a company or a technology —
one of many competing in a market — gets ahead by
chance or clever strategy, increasing returns can
magnify this advantage, and the product or company or
technology can go on to lock in the market.”

Monopoly
The endgame of positive feedback is monopoly. It is not
efficient to have liquidity fragmented across hundreds of
Source: a16z, Barabasi, Griffen
stock exchanges, social networks, or mobile operating

systems. Ultimately, positive feedback will tip the market
We have so far only considered “same-side” network effects.
toward one or two dominant players.
“Cross-side” network effects exist when, say, attracting more

developers to the Playstation ecosystem creates value for
We saw this dynamic play out in mobile operating systems.
gamers by providing a wider variety of games. Finally,
While originally split five ways, the market eventually tipped
complementary network effects crop up when one set of
toward Android. Apple’s iOS survives in second place based
users creates value for users of a separate product (e.g.,
on its tight integration with the upmarket iPhone.
Microso Windows and Office).

While network effects are nuanced, the most important
thing to remember is that bigger is better.

8


Exhibit 19 Scale
Mobile OS Market Share
Economists distinguish between supply- and demand-side
economies of scale. Traditional industrial monopolists
enjoyed supply-side scale, in which they leveraged their
massive size to squeeze out cost efficiencies. In contrast,
demand-side scale stems from the incremental value
consumers derive from greater network effects.

Platform monopolies generally benefit from both supply-
and demand-side economies of scale. Not only do they have
network effects, but marginal cost is nearly zero. Platforms
are the ultimate asset-light businesses. Airbnb does not own
any apartments and Uber does not own any vehicles.
Scaling does not require physical investment.

In their pioneering book on the information economy,
Source: Wikipedia, IDC, Statistica, Sparkline Shapiro and Varian (1999) write:

In Monopolies Are Distorting the Stock Market, we examined “Despite its supply-side economies of scale, General
the rise of superstar firms with supernormal profits and Motors never grew to take over the entire automobile
deep moats. Platform giants are the ultimate superstar market… [b]ecause traditional economies of scale
company. Network effects provide one of the deepest moats. based on manufacturing have generally been exhausted
Once the market is “locked,” it is very challenging for upstart at scales well below total market dominance… .”
competitors to lure away users, resulting in supernormal
profits. In a competitive market, these profits would be In contrast, asset-light platforms with network effects are
competed away. However, network effects prevent or at able to transcend the limits of the industrial economy. This
least greatly slow this mean reversion. enables them to flourish at unprecedented scale.

At the extreme, platforms can only be unseated by the The flipside is that platforms will struggle while subscale.
obsolescence of their market. While Microso still controls Companies must grow their networks to a critical mass
the personal computer OS market, it failed to establish itself before they can deliver enough value to users to earn a
in the now more important smartphone OS market. Intel, profit. Facebook with a single user is pointless. Nobody
seems to have made a similar mistake by passing on a deal wants to wait 30 minutes for an Uber. Achieving scale is not a
to produce the iPhone chip. In a classic example of luxury but a necessity. The opposite of big is failure.
disruption, these firms have ceded critical markets to their
rivals (Apple, Google, ARM and TSMC). Scale economies make it extra important to carefully size up
a company's total addressable market (TAM). If the market is
However, in spite of their missteps, these companies still got too small, even achieving total dominance will not be
to enjoy decades of monopoly profits. Winning a platform enough to justify the fixed costs of building the network.
battle is sort of like getting a patent. Patents incentivize Think twice before launching the “Uber for frozen yogurt!”
companies to invest in R&D by offering the chance of a
temporary legal monopoly. Similarly, platforms invest Competition in the network economy is quite different from
heavily in R&D and user subsidies in hopes of earning a the perfect competition in economics textbooks. Investors
period of monopoly profits. need to adopt a venture capitalist mindset, where they are
effectively buying a call option on a company achieving
market dominance. Platform competition is “go big or go
home!” 🏂🌊

9


Investing in Platforms Admittedly, value hasn’t worked very well in general since
2008. However, these results are even worse than value
investors have experienced in the broader market. We
Platforms Distort Value Investing
believe the unique dynamics of network effects - positive
In Investing in the Intangible Economy, we discuss the feedback and monopoly - are subverting mean reversion in
increasing importance of intangible assets such as this particular subset of stocks.
intellectual property, brands and network effects.
Network Valuation
Despite their rise, financial accounting practices largely We believe the problem is that financial accounting metrics
ignore intangible assets. This omission leads investors to do not capture the intangible value of network effects. If we
systematically undervalue intangible-rich companies. We want to succeed in a market increasingly dominated by
believe this is the primary reason for the broad-based platform stocks, we need ways to value networks.
outperformance of platform companies we saw earlier.
There are many ways to quantify network health. Of course,
Moreover, this leads traditional quant value strategies to the most important metric of network strength is growth,
struggle with platform companies. The companies with the which can be measured both in terms of the number of users
strongest network effects tend to look the most expensive and the overall level of activity. The second most important
on metrics such as price-to-book or price-to-sales. metric is user lock-in, which measures how hard it is for
users to leave for competing networks.
Normally, value investors profit from the tendency of these
metrics to subsequently mean revert. However, network There are many other metrics that can help further refine
effects create positive feedback loops. Thus, companies with our view of network performance. Are users being acquired
strong network effects tend to get stronger while the weak organically? What is the cost of customer acquisition? How
get weaker. engaged are users? How have these metrics evolved by
cohort? How do they differ by geography? For marketplaces,
The following exhibit shows the performance of a simple what is the match rate and are supply and demand
value strategy in the platform company universe. The results balanced?
are the exact opposite of what value investors want. The
more expensive the company, the better it does! We will now address the two most important measures of
network health: growth and lock-in.
Exhibit 20
Value Working in Reverse
Growth
Positive feedback means that investors must carefully
monitor network growth. Accelerating growth could indicate
a platform is tipping the market while deceleration could
portend an impending death spiral.

One comment that gets tossed around is that “venture
capitalists are momentum investors and public market
investors are value investors.” For platforms, the venture
strategy of buying winners and selling losers makes sense in
the presence of positive feedback.

Unfortunately, the metrics used to track network growth

vary widely by company. For instance, Uber reports monthly
active users, trips and gross bookings while Facebook

discloses their monthly and daily average users and average
revenue per user. As these are non-GAAP metrics, they are
10


o en buried in the unstructured portion of companies’ Exhibit 22
quarterly filings and are not always disclosed. Food Delivery Market Share

While we could use NLP to extract this unstructured data, an
even better approach is to leverage now ubiquitous
alternative data. These data can provide a higher frequency
measure of network health.

We’ll use restaurant delivery as a test case. While there are
specialized services that measure website traffic, app usage
and credit card purchases, we will use search query data
from Google. This provides a reasonable metric inasmuch as
hungry users begin their quest for food by typing a term like
“Grubhub” into Google.

Exhibit 21
Source: Google, Sparkline
Google Search Interest

This example provides a quick illustration of the utility of
alternative data in providing a high frequency handle on
growth and competitive positioning. Positive momentum is
particularly important for platforms given the “go big or go
home” nature of network effects.

As a fun bonus, the following exhibit shows the average
strength of each service by state over the past six years.
Seamless has been strong in New York City and Postmates
flourished in Las Vegas.

Exhibit 23
Local Network Effects


We find that all the major food delivery services attracted
increasing interest over the past six years with the exception
of Seamless. Demand spiked in March and April 2020 as
Covid-19 shut down restaurants and forced people to eat at
home.

Normalizing this data to 100% gives us a chart of market
share. While Grubhub and Seamless were early market

leaders, DoorDash has managed to capture around half the
market. Although it only provides data for two dates,

DoorDash’s S-1 confirms our data is in line with their own
This provides a great real life example of local network
estimates.
effects. While DoorDash has the strongest overall network of

drivers, restaurants, and diners, its competitors maintain
dense clusters in particular geographies.

11


Lock-In Exhibit 24
Platform Adoption By Industry (US Top 500)
In order to assess the defensibility of a company’s network,
investors need to understand the concepts of switching
costs and multihoming.

Switching costs are the costs faced by users leaving for a
competing network. They include the cost of breaking legal
contracts; buying expensive platform-specific durable
hardware; learning to use a new system; migrating data from
a proprietary format (e.g., ERP systems); rebuilding positive
reputation (e.g., Yelp, Amazon); and rebuilding a subscriber
base (e.g., Twitter).

Multihoming is a related concept that refers to the tendency
for users to also utilize competing products. For example, it

is common for drivers to work for both Uber and Ly .
While the tech industry will likely become even more
However, it is less common for someone to own both an
saturated with platforms, in order for platforms to further
iPhone and Android phone and even less common for a
impact the stock market, they will need to break into other
hospital to have two EMRs. Multihoming weakens networks
industries.
by making it easier for users to defect to competitors.

One innovation that might produce this outcome is the
“Envelopment” is a competitive tactic platforms use against
Internet of Things. Platforms facilitating the communication
each other. It involves taking users from a competing
of machines (e.g., self-driving cars, buildings, wearable
platform by combining one’s own functionality with the
electronics, heavy machinery) with each other have
target’s to create a multi-platform bundle. For example,
considerable potential to transform the industrial economy.
Microso used envelopment to defeat Netscape in the

browser wars of the 1990s. More recently, Facebook used
In financials, we are starting to see fintech platforms gain
this tactic on Snapchat with its “Stories” feature. Firms with
traction in payments, peer-to-peer lending and insurance.
networks overlapping (Snapchat) or even simply adjacent
Blockchain technology could provide a further opening for
(Netscape) to those of formidable competitors are more
the platformization of the financial sector.
vulnerable, all things equal.

Platforms offer a potentially more efficient way to organize
The Future of Platforms economic activity. Thus, it is likely they will play at least
some role as technology ultimately infiltrates finance,
Cross-Industry Adoption industrials and other asset-heavy sectors.

While platforms now dominate the communications and
technology industries, they have not experienced
Rising Valuation
widespread adoption in less information-intensive We found that platforms have historically been undervalued
industries. It is hard to think of many successful platforms in relative to the strength of their businesses. However, the
industrials, energy, and materials. market seems to be gradually waking up to the power of
network effects.

12


Exhibit 25 concerns have also arisen over the laws and regulations that
Relative Valuation of Platform Companies have not yet adapted to this new economic model.

One important issue is the employment status of external
workers. The most politicized example is of Uber drivers.
Uber wants these workers classified as contractors in order
to avoid having to pay payroll taxes, minimum wage, and
benefits.

A second issue is the responsibility of publishing platforms
like Facebook and YouTube to police third-party content.
While controversy first centered on copyright infringement,
the most important issue today involves fake news and hate
speech. As we saw with the recent presidential elections,
extremists and state actors will go to great lengths to
weaponize tech platforms.

Source: S&P, SEC, Sparkline Platforms came of age in a period of lax regulation as
regulators and public opinion slowly came to grasp their
Over the past few years, the relative valuation of platform profound societal impact. However, we are likely nearing the
companies has increased by around +85% when measured end of this grace period. We expect platforms will ultimately
against sales or book value. However, the increase is a more have to take on more responsibility for the participants in
muted +33% compared to earnings as platforms have also their communities. Platforms are o en compared to nation
increased their profitability advantage over the market. states and may soon be expected to practice governance
principles befitting this role.
Much of this effect is not specific to platforms but is the
consequence of the market's recent obsession with hot IPOs. Mega-Platforms and Antitrust
It just so happens that many extremely richly-valued IPOs
are also platforms, such as Snowflake (P/S of 360), DoorDash Until recently, the trend has been toward the consolidation
(P/S of 26) and Airbnb (P/S of 56). Removing these stocks of platforms into a few mega-platforms that bundle together
cuts the valuation increase in half. However, established a wide variety of services. Google users can get email, cloud
platforms have still seen relative valuations rise modestly. storage, navigation, search, browser, spreadsheets, and
video conferencing without having to leave the ecosystem.
While we believe platforms generally remain undervalued,
we need to be a bit more discerning. In particular, investors We expect this trend to slow and possibly reverse due to
should be careful not to blindly buy into platforms that are regulatory backlash. Over the past few months, Google and
also hot IPOs pricing in unrealistic growth. Facebook have been served with antitrust lawsuits. Pressure
is mounting for Facebook to divest Instagram and WhatsApp.
Governance and the Gig Economy
Even the Chinese government has started cracking down on
The most significant societal impact of platforms stems from tech monopolies. China is perhaps even more platformized
their externalization of the means of production. Rather than the US, with China’s BAT (Baidu, Alibaba and Tencent)
than hire employees, these companies manage vast enjoying an even greater concentration of power than our
networks of independent contractors. This has fueled the FAANG. Among other antitrust moves, the Chinese
growth of the so-called “gig economy.” authorities scuttled the $37 billion IPO of fintech giant Ant
Group a er publicly chastising its founder, Jack Ma.
On one hand, the gig economy offers high levels of flexibility
and autonomy and opportunities to workers who might Overall, it feels that the political winds are shi ing against
otherwise be excluded from the labor market. However, giant tech companies. While this will likely limit the growth
13


of mega-platforms, it may foster the emergence of more
independent single-product platforms.

Conclusion Kai Wu
Founder & CIO, Sparkline Capital LP
Platforms have become a dominant economic force. Their
ascent has disrupted dozens of industries, created massive
concentrations of power, and transformed many aspects of Kai Wu is the founder and Chief Investment Officer of
society. Sparkline Capital, an investment management firm applying
state-of-the-art machine learning and computing to uncover
However, investors have been slow to recognize the power alpha in large, unstructured data sets.
of the platform business model. Platforms have handily
outperformed the stock market for over a decade. We Prior to Sparkline, Kai co-founded and co-managed
believe this is due to the complexity of cleanly identifying Kaleidoscope Capital, a quantitative hedge fund in Boston.
platform companies and valuing intangible network effects. With one other partner, he grew Kaleidoscope to $350
million in assets from institutional investors. Kai jointly
Network effects lead to positive feedback and monopoly, managed all aspects of the company, including technology,
which disrupt traditional investment approaches. Given investments, operations, trading, investor relations, and
their increasing importance, investors would be well served recruiting.
to devote resources to deeply understanding the platform
business model. Previously, Kai worked at GMO, where he was a member of
Jeremy Grantham’s $40 billion asset allocation team. He
also worked closely with the firm's equity and macro
investment teams in Boston, San Francisco, London, and
Sydney.

Kai graduated from Harvard College Magna Cum Laude and
Phi Beta Kappa.
Disclaimer
This paper is solely for informational purposes and is not an offer
or solicitation for the purchase or sale of any security, nor is it to be
construed as legal or tax advice. References to securities and
strategies are for illustrative purposes only and do not constitute
buy or sell recommendations. The information in this report should
not be used as the basis for any investment decisions.

We make no representation or warranty as to the accuracy or
completeness of the information contained in this report, including
third-party data sources. The views expressed are as of the
publication date and subject to change at any time.

Hypothetical performance has many significant limitations and no
representation is being made that such performance is achievable
in the future. Past performance is no guarantee of future
performance.
14

The Platform Economy

Uploaded by

Copyright:

Available Formats

The Platform Economy

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Platform Economy

Uploaded by

Copyright:

Available Formats

The Platform Economy

The Rise of the Platform Exhibit 2

Exhibit 1 Source: S&P, Sparkline

Platform Score = 0.06

Source: S&P, SEC, Sparkline

Source: S&P, SEC, Sparkline

Why Have Platforms Thrived?

You might also like