52 Business The EconomistAugust 29th 2020
“O
ne neverreally knows who one’s
enemy is.” The words of Jürgen Ha-
bermas, a noted Frankfurt School philoso-
pher, are a good point of departure for un-
derstanding Palantir Technologies. On
August 25th the controversial software
firm, named after a magical orb in J.R.R.
Tolkien’s “Lord of the Rings” that lets users
see and speak across space and time, filed
the paperwork to list on the New York Stock
Exchange. Its direct offer of existing shares
to public investors, without raising fresh
capital, could happen within a month.
The company sells programs that gather
disparate data and organise them into
something usable for decision-makers,
from soldiers in Afghanistan to executives
at energy firms. More than a technological
project, it is a philosophical, even political
one. In the early 2000s its co-founder and
boss, Alex Karp (who used to sit on the
board of The Economist’s parent company),
wrote a dissertation about aggression in
politics at Frankfurt’s Goethe University,
though not under Mr Habermas, as is often
claimed. And Palantir itself is a child of the
9/11 terrorist attacks of 2001, which Ameri-
ca’s sundry law-enforcement outfits failed
to avert because they did not share data. In
a preface to the prospectus, bleak by the up-
beat standards of the genre, Mr Karp writes
of government agencies that “faltered” and
crises that “expose the systemic weakness-
es of the institutions on which we depend”.
Fixing these shortcomings is the com-
pany’s raison d’être. It could prove lucrative.
But it invites attacks from rivals and critics.
Like most startups that plan to go public
these days, Palantir is haemorrhaging red
ink. The 17-year-old firm has yet to turn a
profit. Last year it lost $580m on revenue of
$742m. It spends more on sales and mar-
keting than on research and development
(see chart). But its venture-capital backers,
who have poured $3bn into the firm over
the years, most recently valuing it at per-
haps $26bn, can draw comfort from things
moving in the right direction. In the first
half of 2020 revenue rose by 49%, year on
year, while losses got smaller. Sales may ex-
ceed $1bn for the full year, thanks to the use
of Palantir’s products to analyse pandemic
data. It will vie with Snowflake, another
data business about to list, for the year’s
biggest software flotation.
Palantir’s longer-term prospects are
murkier. Successful corporate-software
firms develop programs and services thatcan be offered without much customisa-
tion to many clients. This is trickier in the
data business, where every company has a
unique digital footprint. When Palantir got
going, it was in effect a professional-ser-
vices firm, chiefly creating bespoke data-
analysis systems for the likes of the ciaand
the Department of Defence. In recent years
it has developed more generic products for
corporate clients. But its scepticism of
standardisation means it continues to de-
ploy plenty of engineers to tweak them.
This increases costs and is likely to limit
how big and profitable it can get, says Mark
Moerdler of Bernstein, a brokerage.
Palantir’s origins bring other chal-
lenges, too. Because it came of age before
the rise of computing clouds, its software
often still inhabits customers’ data centres,
making it less nimble than younger cloud-
based rivals like c3.ai and Databricks.
Working for the government, particularly
its spookier agencies, has also created a se-
cretive and proprietary culture that is not
an easy fit with the sort of partnerships that
other tech firms often successfully use to
expand their business. And it remains
heavily reliant on government contracts.
Between January and June 55% of revenue
came from official sources, up from 45% in
the same period last year. It has only 125 cli-
ents, with the biggest three (unnamed)
ones accounting for nearly a third of sales.
Closeness to the state also points to Pa-
lantir’s biggest risk: politics. From its
post-9/11 beginnings it has seen itself as an
instrument of national security. “If we are
going to ask someone to put themselves inharm’s way, we believe that we have a duty
to give them what they need to do their job,”
Mr Karp writes in his missive. One of his
co-founders is Peter Thiel, a famed venture
capitalist of strong libertarian bent with an
authoritarian streak—and an occasional
supporter of President Donald Trump.
This—combined with work for Immi-
gration and Customs Enforcement, a feder-
al agency despised by progressives for its
heavy-handed treatment of migrants, or
the Pentagon’s Project Maven, to analyse
drone footage—has made Palantir one of
the most hated firms in left-leaning Silicon
Valley. “I’ve had my favourite employees
yell at me,” said Mr Karp earlier this year,
from a barn in New Hampshire where he
was self-isolating even before the pandem-
ic. Some engineers have left. Others are de-
manding high salaries to remain; in the
first half of the year Palantir paid $182m in
stock-based compensation, 38% of rev-
enue. Though being in bed with America’s
law enforcers and spies won’t scare off oth-
er government customers, corporate cli-
ents may take fright, particularly abroad.
As the prospectus concedes, “Our reputa-
tion and business may be harmed by news
or social media coverage.”
Palantir, which has recently decamped
from Silicon Valley to Denver, is trying to
make a virtue of the culture clash. It paints
itself as a patriotic problem-solver, es-
chewing the techno-Utopian pretensions
of the West Coast’s engineering elite. They
may know more than most about software,
Mr Karp writes. “But they do not know
more about how society should be organ-
ised or what justice requires.” That, he im-
plies, is the role of elected governments;
the prospectus rules out dealing with Com-
munist China. An unusual sales pitch in
tech. But a plausible one. 7SAN FRANCISCO
A secretive software-maker says hello to the stockmarket—and waves
goodbye to Silicon ValleyPalantir TechnologiesA prickly patriot
Can it break the spell?
Palantirfinancials,$mSource: Companyreports3002001000-100-200-300-4002019 2020Q3Q2Q1 Q2Q1Q4Net lossOther
expensesSales &
marketingResearch &
developmentRevenuesData wizardry