lion drivers, this meant Uber had to find
375,000 new drivers every month and replace all its drivers every eight months.
Driver costs before commissions on rides
were almost $250 million per month or
nearly $3 billion per year.
2 In short, Uber
has been massively subsidizing both
sides of its platform—large payments to
drivers as well as low fares to riders. This
is a great way to lose a lot of money.
Lyft loses less money ($911 million
in 2018 on sales of $2.2 billion, and
$1.14 billion on sales of $776 million in
the first quarter of 2019, including IPO-related stock-compensation charges)
mainly because it is smaller.
lost $1.9 billion in 2018 on $1.8 billion
in revenues and postponed its IPO after losses of $690 million on revenues
of $1.5 billion in the first six months of
2019.6 By contrast, Airbnb, which plans
to go public in 2020, consistently reports profits or at least a positive cash
9 Why? Because Airbnb can charge
both sides of its platform. It does not pay
people to list rooms or to rent real estate
on its supply side, and it allows renters
to charge market prices. Uber pays drivers even when they do not have riders.
We Work pays for real estate even when
it does not have renters (and, like Uber,
it generally has charged well below the
market price in order to grow quickly).
Why would investors put their money
into businesses that consistently lose
money? In the case of Uber, investors
must be hoping for a “winner take all
or most” outcome. Once competitors
have disappeared, then Uber can raise
prices to riders and reduce payments
to drivers. But Uber already has 70% of
the U.S. ride-sharing market and it still
does not have enough market power or
operating efficiencies to turn a profit.
At least in part this is because of high
driver turnover and the fact that, in
most cities, there are still too many
transportation options (including
people using their own vehicles). Uber
makes a profit mainly in a few cities
where taxis and other transport options
are limited and expensive.
Uber has also attracted investors with
the promise it will become the “Amazon
of transportation” and move into nearly
every transportation segment.
1 But ex-
pansion into more “bad” businesses
simply means the bigger the platform
gets, the more money it will lose. Ama-
zon took many years to make a profit and
number of employees, generated nearly
twice the operating profits, had market
values more than twice as high, and were
growing about twice as fast (see the ac-
Not surprisingly, investors and entrepreneurs keep looking for the next
blockbuster platform. In fact, we estimate 60% to 70% of the billion-dollar private “unicorn” companies are platform
5 But “platformizing” (creating
a platform) in a “bad” business (an industry with low profit margins due to
high costs, low entry barriers, or other
structural factors) does not make it a
good business. Profits depend on supply and demand, which impact prices,
as well as economies of scale and scope
or other efficiencies, which impact
costs. Moreover, a business that delivers a physical good or service is unlikely
to generate the same high profits as a
platform that sells digital goods such
as software, content, or advertisements,
which have close to zero marginal costs.
Let’s look more carefully at Uber, the
most valuable sharing-economy plat-
form. It went public in May 2019 and has
since seen its stock price drop sharply,
even though Uber remains a compelling
and convenient service. Uber users can
summon a ride, usually within minutes,
track the driver’s progress, and then pay,
all via a smartphone app. Uber and oth-
er ride-sharing companies also usually
charge less than taxis to attract riders.
Each additional driver adds value be-
cause transportation options for riders
increase. Uber also owns no cars, so its
capital costs are minimal. Nevertheless,
Uber lost $4.5 billion in 2017 and $1.8
billion in 2018 (with income boosted
from selling some overseas operations).
For the first six months of 2019, Uber
reported sales of $6.2 billion and operat-
ing losses of $6.5 billion, including costs
related to the IPO.
12 Uber has been rais-
ing prices and cutting staff, yet operating
losses in the billions are expected to con-
3 Why does a company with such a
valuable service lose so much money?
First, Uber (like Lyft and WeWork)
is not really a digital business. Only the
transaction process is digital. Transport-
ing people or goods is a physical service
with potentially high costs. In this case,
Uber must pay a lot of money to find
drivers, and it keeps prices low to attract
riders. In addition, Uber’s economies of
scale and scope are primarily local since
each area needs its own drivers.
Second, a platform rather than a
traditional product or service compa-
ny should bring together two or more
market sides, rely on network effects to
grow, and then charge for a product or
transaction fee without the expense of
having the same number of employees
or large capital investments as a con-
ventional business. Uber’s platform
works best if the driver side is heavily
populated so that customers can always
get a ride quickly when they need one.
However, Uber drivers frequently quit
because of long hours and low compen-
sation, with no employee benefits since
they are independent contractors.
According to data released prior to its
IPO, Uber drivers were quitting at a rate
of 12.5% per month and the company
had to pay approximately $650 to hire
each new driver. With at least three mil-
Median values for Forbes Global 2000 industry control sample and platforms, 1995–2015.
Number of Firms
Sales (Million$) $4,845 $4,335
Operating Profit 12% 21%*
Market Value (Million$) $8,243 $21,726*
Mkt Value-Sales Multiple
1. 94 5. 35*
Sales Growth vs. Prior Year 9% 18%*
Source: Michael A. Cusumano, Annabelle Gawer, and David B. Yoffie,
The Business of Platforms (2019), p. 23.
* Differences significant at p < 0.001 for Industry Sample vs. Platforms using
two-sample Wilcoxon rank-sum (Mann-Whitney) test.
Mkt Value-Sales Multiple = ratio of market value compared to prior year sales.
Average of 13 years of data for 18 innovation platforms and five years for 25 transaction platforms.