56 Business TheEconomistJanuary22nd 2022
China’s fiveyearold aimaster plan set
out a number of goals. For example, by
2025 the country is to create an industry
with global revenues of 400bn yuan,
achieve “major breakthroughs” in technol
ogy and lead the world in some applica
tions. Five years later it is to dominate the
industry (by then worth $1trn in sales),
having written its ethical code and set its
technical standards, just as Europe and
America defined the contours of the Indus
trial Revolution.
Elements of the Communist Party’s ap
proach are characteristically prescriptive.
The Ministry of Science and Technology
has instructed China’s tech giants with ex
isting ventures in certain subdisciplines of
ai—Tencent in medical image recognition,
Baidu in autonomous driving—to double
down on these. That said, the plan is less
handson than some of the country’s other
development projects, observes Jay Huang
of Bernstein, an investment firm. In the
words of Huw Roberts of Oxford University
and five coauthors, the blueprint acts
chiefly as a “seal of approval” which “de
risks” assorted aiinitiatives championed
by centralgovernment entities, local au
thorities and the private sector.
In practice, the derisking involves dol
ing out lots of public money. Some of this
takes the form of tax breaks and subsidies,
as in the “little giants” programme to nur
ture 10,000 promising startups across va
rious sectors, including ai. Local govern
ments, even in poor rustbelt provinces
such as Liaoning in the far northeast, have
also dangled similar incentives in front of
aicurious companies.
Another type of support comes from
government procurement. Firms do not
disclose how much revenue they derive
from publicsector contracts. But the share
is likely to be significant. Central and local
authorities use SenseTime’s surveillance
technology. Megvii, which also specialises
in image recognition, has extensive deal
ings with stateowned enterprises.
The state is also investing in aicompa
nies directly. The central government runs
several techinvestment vehicles. Local
governments are increasingly creating
their own, often armed with billions of
dollars. Tianjin, a coastal metropolis, an
nounced a $16bn aifund in 2018.
Government capital is increasingly
helping plug a gap left by foreign investors
scared away by American sanctions
against some of China’s aidarlings, which
are seen as being too close to the Commu
nist Party. A fund run by the Cyberspace
Administration of China, a regulator, has
acquired an undisclosed stake in Sense
Time, which last month was hit by another
round of American sanctions over its al
leged involvement in government repres
sion of the Uyghur ethnic minority. (Sense
Time says that the sanctions are based on a
“misperception”ofitsbusiness.)Asepa
ratevehicle,theMixedOwnershipReform
Fund,accountedfor$200mofthe$765m
thatthefirmraisedinitsinitialpublicof
fering(ipo). Localgovernmentschippedin
another$220m.
Lostintranslation
Statedosh,combinedwithaccesstoplen
tifulpublicdata,hashelpedturnChinese
aifirmsintopowerhousesincertainnich
es.AccordingtoBain,a consultancy,bylast
JunetheclouddivisionofAlibaba,China’s
ecommercebehemoth,wasoffering 62 ai
enabledservices,fromvoicerecognitionto
videoanalytics,comparedwith 47 fromits
closest Western rival, Microsoft. Sense
TimeandMegviimassproducecomputer
visionsoftwareandhardwarethatcanbe
adaptedtoandinstalledinindividualfac
tories.Despitebeinglockedoutofmost
Westernmarkets bytheAmerican sanc
tions,SenseTimerakedin762myuanin
overseasrevenuesin2020,comparedwith
319myuantwoyearsearlier,mostlyfrom
SouthEastAsia.
Forallthesesuccesses,though,China’s
aiindustry trailstheWest inimportant
ways.DespiteleadingAmericaintheover
allnumberofairelatedpublications,Chi
naproducesfewerpeerreviewedpapers
thathaveacademicandcorporatecoau
thorsorarepresentedatconferences,both
ofwhicharetypicallyheldtoa higherstan
dard.ItranksbelowIndia,andwellbelow
America,inthenumberofskilledaicoders
relativetoitspopulation.Theseshortcom
ingsarelikelytopersist,forthreereasons.
First,capitalmaynotbebeingallocated
efficiently.Itisunclear,forexample,how
muchofTianjin’s$16bnkittyhasactually
been deployed. More damaging, Beijing
hascreateda systemforrewardinglocalof
ficialsthatfavoursdebtfuelledspending
andseldompunisheswastefulness.
Manystateaiinvestmentshavebeen
“recklessandredundant”,saysJeffreyDing
of Stanford University. Zeng Jinghan of
LancasterUniversityhasdocumentedthe
riseoffirmsthatfalselyclaimtobedevel
opingaiinordertosuckupsubsidies.One
analysisby Deloitte,aconsultancy,esti
matedthat99%ofselfstyledaistartupsin
2018 werefake.Suchboondogglesnotonly
burnthroughpubliccash,MrDingnotes,
but alsoconsume scarcehuman capital
thatcould more usefullyhave been de
ployedelsewhere.
China’ssecondproblemisitsinability
torecruittheworld’sbestaiminds,espe
cially those working on highlevel re
search.A studyin 2020 byMacroPolo,a
Chicagobased thinktank, showed that
morethanhalfoftoptierresearchersin
thefieldwereworkingoutsidetheirhome
countries.AmericaandEuropelookmore
appealing to such footloose brainboxes,
including many Chinese ones. Though
abouta thirdoftheworld’stopaitalentis
fromChina,onlyatenthactuallyworks
there.A shortageofnonChineseresearch
ersfurtherhandicapsChina’scapabilities,
notesMattSheehanoftheCarnegieEn
dowmentforInternationalPeace,a think
tankinWashington.
Evenmoreproblematicfortheparty,its
master plan ignored the cuttingedge
semiconductorsthatpower ai. Sinceits
publication Chinese companies have
found itevermoredifficultto gettheir
Intelligence gathering
Sources:RefinitivDatastream;PitchBook;StanfordUniversity *January19th †Into unlisted companies
Private investment† in artificial intelligence
22, $bn
Australia
Singapore
Japan
India
Canada
Israel
Britain
European Union
China
United States
2520151050
China, selected artificial-intelligence companies
Market capitalisation or valuation, $bn
TransInfoTech.
Bairong
SenseTime
Hikvision
Zhejiang Dahua Tech.
1
150
120
90
60
30
0
2018 212019 22*
Eye robot
China,computer-vision
softwaremarket,yuanbn
Source:SenseTime
100
80
60
40
20
0
252423222120192018
1 19 21 23
FORECAST
City
management
Enterprise
Consumer
Automotive
%of world total
2