AI, innovation and the public good: A new policy playbook

Kilic, Bu cu
Wo king Pape
AI, inno a ion and he public good: A new policy playbook
CIGI Pape s, No. 318
P o ided in Coope a ion wi h:
Cen e o In e na ional Go e nance Inno a ion (CIGI), Wa e loo, On a io
Sugges ed Ci a ion: Kilic, Bu cu (2025) : AI, inno a ion and he public good: A new policy playbook,
CIGI Pape s, No. 318, Cen e o In e na ional Go e nance Inno a ion (CIGI), Wa e loo (On a io)
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/322416
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/4.0/
CIGI Pape s No. 318 — Ma ch 2025
AI, Inno a ion and he
Public Good: A New Policy
Playbook
Bu cu Kilic
CIGI Pape s No. 318 — Ma ch 2025
AI, Inno a ion and he
Public Good: A New Policy
Playbook
Bu cu Kilic
Abou CIGI
The Cen e o In e na ional Go e nance Inno a ion (CIGI) is an independen ,
non-pa isan hink ank whose pee - e iewed esea ch and us ed analysis
in luence policy make s o inno a e. Ou global ne wo k o mul idisciplina y
esea che s and s a egic pa ne ships p o ide policy solu ions o he digi al
e a wi h one goal: o imp o e people’s li es e e ywhe e. Headqua e ed
in Wa e loo, Canada, CIGI has ecei ed suppo om he Go e nmen o
Canada, he Go e nmen o On a io and ounde Jim Balsillie.
À p opos du CIGI
Le Cen e pou l’inno a ion dans la gou e nance in e na ionale (CIGI) es un
g oupe de é lexion indépendan e non pa isan don les eche ches é aluées
pa des pai s e les analyses iables inci en les décideu s à inno e . G âce
à son éseau mondial de che cheu s plu idisciplinai es e de pa ena ia s
s a égiques, le CIGI o e des solu ions poli iques adap ées à l’è e numé ique
dans le seul bu d’amélio e la ie des gens du monde en ie . Le CIGI, don le
siège se ou e à Wa e loo, au Canada, béné icie du sou ien du gou e nemen
du Canada, du gou e nemen de l’On a io e de son onda eu , Jim Balsillie.
Copy igh © 2025 by he Cen e o In e na ional Go e nance Inno a ion
The opinions exp essed in his publica ion a e hose o he au ho and do no
necessa ily e lec he iews o he Cen e o In e na ional Go e nance Inno a ion
o i s Boa d o Di ec o s.
Fo publica ions enqui ies, please con ac [email p o ec ed].
The ex o his wo k is licensed unde CC BY 4.0. To iew a copy o his licence,
isi h p://c ea i ecommons.o g/licenses/by/4.0/.
Fo euse o dis ibu ion, please include his copy igh no ice. This wo k may
con ain con en (including bu no limi ed o g aphics, cha s and pho og aphs)
used o ep oduced unde licence o wi h pe mission om hi d pa ies.
Pe mission o ep oduce his con en mus be ob ained om hi d pa ies di ec ly.
Cen e o In e na ional Go e nance Inno a ion and CIGI a e egis e ed
adema ks.
67 E b S ee Wes
Wa e loo, ON, Canada N2L 6C2
www.cigionline.o g
C edi s
Resea ch Di ec o , Digi al Economy S. Yash Kalash
Di ec o , P og am Managemen Dianna English
P og am Manage Jenny Thiel
Publica ions Edi o Susan Bubak
Publica ions Edi o Ch is ine Robe son
G aphic Designe Sami Chouhda y

Table o Con en s
i Abou he Au ho
i Ac onyms and Abb e ia ions
1 Execu i e Summa y
1 In oduc ion
2 Wha Exac ly Is Indus ial Policy?
3 Indus ial Policy: Pas , P esen and Fu u e
5 Building Compe i i eness in AI: Is The e a Pa h Fo wa d?
7 The Da a Cen e Dilemma: In as uc u e o Whom?
9 NIS
18 Conclusion: The Pa h Fo wa d
20 Wo ks Ci ed
i CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
Abou he Au ho
Bu cu Kilic is a CIGI senio ellow, and ech and
human igh s ellow a he Ca Cen e , Ha a d
Kennedy School. She has wo ked wi h a di e se
ange o o ganiza ions ac oss ci il socie y,
philan h opy and academia. He esea ch and
w i ings co e digi al igh s, in ellec ual p ope y
(IP), inno a ion and ade, and she has p o ided
echnical ad ice and assis ance in coun ies
in Asia, La in Ame ica, Eu ope and A ica.
As he o me head o policy o F on ie
Technology — a Minde oo Founda ion
ini ia i e — Bu cu guided he o ganiza ion’s
app oach o eme ging echnology, ad oca ing
o esponsible, equi able and jus solu ions.
Be o e joining Minde oo, she di ec ed he Digi al
Righ s P og am a Public Ci izen, a non-p o i
consume ad ocacy o ganiza ion in Washing on,
DC, and also led hei esea ch on access o
medicines. He in luence in ech policy, IP and
ade unde sco es he commi men o policy
en ep eneu ship and igh s-based ad ocacy. She
champions collabo a i e ci il socie y engagemen ,
policy en ep eneu ship and inno a i e policy
de elopmen on a global scale. In 2015, she was
ecognized as one o he 300 Women Leade s in
Global Heal h o he wo k on heal h and ade
policy. F om 2021 o 2022, she was a p ac i ione
ellow wi h he Digi al Ci il Socie y Lab a he
S an o d Cen e on Philan h opy and Ci il Socie y.
She comple ed he Ph.D. a Queen Ma y
Uni e si y o London and holds L.L.M. deg ees
in IP law om Queen Ma y Uni e si y o
London, and in o ma ion echnology law om
S ockholm Uni e si y. She ob ained he law
deg ee om Anka a Uni e si y, Tu key.
Ac onyms and
Abb e ia ions
AI a i icial in elligence
AMIs Ad anced Manu ac u ing
Ins i u es
AWS Amazon Web Se ices
G7 G oup o Se en
GDPR Gene al Da a
P o ec ion Regula ion
ICT in o ma ion and
communica ions echnology
IP in ellec ual p ope y
LLMs la ge language models
NAIRR Na ional A i icial
In elligence Resea ch Resou ce
NIS na ional inno a ion sys em
R&D esea ch and de elopmen
RIC Resea ch and Inno a ion Council
SMBs small and medium-sized
businesses
STEM science, echnology,
enginee ing and ma h
WTO Wo ld T ade O ganiza ion
1AI, Inno a ion and he Public Good: A New Policy Playbook
Execu i e Summa y
In la e Janua y 2025, a li le-known Chinese
s a -up, DeepSeek, made headlines wi h a
b eak h ough open-sou ce a i icial in elligence
(AI) model called R1. The model sen shockwa es
h ough he ech indus y and among Wall
S ee in es o s. R1 is epo edly ala mingly
good, pe o ming simila ly o OpenAI’s op- ie
models, bu i is also mo e cos -e ec i e and uns
on less-ad anced chips. DeepSeek changed he
con e sa ion abou AI and challenged he dominan
AI na a i e, which has long ocused on bigge
models, high-end chips, massi e in es men s
and expansi e da a cen es o p og ess. R1
demons a ed ha inno a ion and compe i ion
emain possible, e en on an une en playing ield.
This eye-opening momen coincides wi h he
e i al o indus ial policy as a s a egic ool o
go e nmen s aiming o build AI capaci y and
compe i i eness. Once dismissed unde neolibe al
economic amewo ks, indus ial policy is making
a s ong comeback wi h mo e go e nmen s
wo ldwide emb acing i o build digi al public
in as uc u e and os e local AI ecosys ems.
This pape examines how he na ional inno a ion
sys em (NIS) amewo k can guide AI indus ial
policy o os e inno a ion and educe eliance on
dominan ech companies. The concen a ion o
AI in as uc u e, compu e powe , aining da a
and cloud se ices in he hands o a ew dominan
ech companies has c ea ed a bo leneck o AI
inno a ion. I is di icul o small playe s o en e
he ma ke and compe e on ai e ms. Wi hou
a ge ed policy in e en ions, AI isks u he
consolida ing powe in a hand ul o ech companies.
Tha is why success ul AI policy mus go
beyond passi e adop ion and ins ead ocus on
unde s anding he local con ex and add essing
local needs. I should se clea p io i ies o enhance
domes ic inno a ion capabili ies o ensu e ha
AI de elopmen aligns wi h b oade economic
and socie al goals. S a egic in es men s in AI
esea ch and de elopmen (R&D) a e essen ial
o d i e independen echnological p og ess
and educe eliance on big ech in as uc u e.
Addi ionally, es uc u ing public ins i u ions
and adop ing a whole-o -go e nmen app oach
o AI go e nance can imp o e coo dina ion and
e ec i eness. In his con ex , aligning ade
policy wi h indus ial policy and compe i ion
is c i ical o os e ing a ai and dynamic AI
ecosys em ha suppo s local inno a ion and
ensu es long- e m echnological so e eign y.
The AI ace is no only abou echnological
b eak h oughs o bigge and as e models. The e
will be no single model o coun y domina ing
he u u e o AI. Ins ead, he u u e o AI will
be mul i-model and mul ina ional. Wi h bold
and s a egic policy making, go e nmen s can
shape AI’s ajec o y and ensu e ha inno a ion
se es no only a ew ech companies and
na ions bu also global socie y as a whole.
In oduc ion
No long ago, men ioning “indus ial policy” in
Wes e n capi als such as Washing on, DC, B ussels
o London was conside ed nea ly aboo. The
e m was so s igma ized ha i isked d awing
associa ions wi h Eu opean socialism o he
economic s a egies o de eloping na ions.
S a ing in he 1980s, he e m “indus ial policy”
ell ou o a ou unde he in luence o he
neolibe al economic o de , which suppo ed
ee-ma ke ideas. The e was a deep ideological
di ide: s a e in e en ion in ma ke s was
iewed skep ically while ee-ma ke p inciples
domina ed he poli ical and economic discou se.
The la e ’s in luence shaped global policy,
pushing o ma ke p i a iza ion, educed
go e nmen spending and ade libe aliza ion.
A shi began slowly a e he 2008 inancial c isis
as go e nmen s ied o add ess p essing issues
such as sus ainable job c ea ion, g een ansi ions
and supply chain esilience. “You ne e wan a
se ious c isis o go o was e,” decla ed an Obama
o icial in 2008, as he p esiden launched he
Ame ican Reco e y and Rein es men Ac , a
his o ic s imulus plan ha included unp eceden ed
in es men s in enewable ene gy (quo ed in Velu
2024). The e was also a g owing awa eness among
policy make s o China’s inc easing economic
powe and compe i i eness, aising conce ns
abou he impac s on he Uni ed S a es’ own
economy, jobs and wel a e (Mo ison 2019).
2CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
Wi h all hese ac o s in play, indus ial policy
e-eme ged as a opic in academic con e ences,
policy o ums, Wo ld Bank epo s, na ional
g ow h s a egies, iscal plans and e en elec ion
campaigns. I is now in he spo ligh , peaking as
a new buzzwo d in policy discussions (Si ipu apu
and Be man 2023; Ilyina, Paza basioglu and
Ru a 2024). Mo eo e , new empi ical s udies and
di e en expe iences wi h indus ial policies ha e
b ough esh insigh s o he deba e, o e ing a
mo e nuanced and con ex ual pe spec i e ha
add esses some o he in e p e a ional challenges.
This encou aged mo e p oduc i e discussions
among economis s, shi ing ocus om hea ed
disag eemen s o cons uc i e analysis and
unde s anding (Juhász, Lane and Rod ik 2023).
In sho , indus ial policy is back. Today, i is
inc easingly discussed in he con ex o AI.
AI and indus ial policy ha e become cen al
hemes, domina ing global discussions and
shaping policy agendas wo ldwide. E e y
coun y seems o be asking he same ques ions:
How can we build he in as uc u e needed o
become compe i i e in AI? How do we le e age
AI o d i e inno a ion and economic g ow h?
AI indus ial policies a e now widely discussed
in wo ld capi als and policy ci cles. Go e nmen s
wo ldwide a e eleasing s a egic plans o
pa icipa e in he global AI economy and le e age
i o economic g ow h and he ad ancemen o
socie y. Jus be o e lea ing o ice in Janua y 2025,
P esiden Joe Biden signed an execu i e o de o
ad ance US AI in as uc u e, ocusing on la ge-
scale da a cen es and clean ene gy acili ies (The
Whi e House 2025). The UK go e nmen ecen ly
in oduced he AI Oppo uni ies Ac ion Plan, a
mode n indus ial s a egy o AI (Depa men o
Science, Inno a ion & Technology 2025). Simila ly,
in Sep embe 2024, he Eu opean Union published a
long-awai ed indus ial s a egy aimed a enhancing
EU compe i i eness, pa icula ly in digi al
echnologies and AI (D aghi 2024). In Japan, he
Minis y o Economy, T ade and Indus y launched
he GENIAC p ojec ,1 aimed a le e aging gene a i e
AI o d i e economic g ow h and socie al bene i s.
Indus ial policy e o s a e no limi ed o
de eloped economies. Eme ging economies
such as B azil (Minis é io da Ciência, Tecnologia
1 See www.me i.go.jp/english/policy/mono_in o_se ice/geniac/
index.h ml.
e Ino ação 2024), India (Panday and Samdub
2024), Sou h A ica (Makumbi o a 2024), Tu key
(Minis y o Indus y and Technology 2021) and
many o he s — whe he la ge o small — a e
ac i ely explo ing pa hways o build AI capaci y
and pa icipa e in he global AI economy.
In his con ex , AI echnologies equi e a esh
look a he li e a u e and lessons om pas
indus ial policy expe iences, including he
di e si y o policies, s uc u al linkages and
go e nmen in e en ions. Wha is needed
is a mode n se o policy ins umen s ha
a e lexible, esponsi e and g ounded in
expe imen a ion, lea ning and con inuous
imp o emen . Gi en he apid pace o echnological
ad ancemen s, hey should be designed o
espond o he e ol ing na u e o echnology.
The g owing ocus on AI and indus ial policy
highligh s he need o well-c a ed policies ailo ed
o local con ex s, esou ces and ins i u ional
capaci ies. These policies mus ensu e ha
e o s a e sus ainable and aligned wi h b oade
economic and socie al goals. While he ambi ions
a e clea o build in as uc u e, os e inno a ion
and achie e compe i i eness, he e is no sil e
bulle . A ca e ully designed policy oad map,
howe e , can p o ide he di ec ion needed. This
pape seeks o p esen he big pic u e, o e ing
key insigh s and ac ionable ideas wi hou ge ing
in o he g anula de ails o each policy p oposal.
While each p oposal dese es a de ailed epo
o i s own, he pu pose he e is o p o ide a
s a ing poin ha inspi es hough and ac ion.
Wha Exac ly Is Indus ial
Policy?
The e is a lo o discussion abou indus ial policy
hese days. Policy make s wo ldwide emb ace
he concep , while academics, esea che s and
hink anks p oduce a icles and epo s, o e
ecommenda ions and hos con e ences (Millo
and Rawdanowicz 2024; P ojec Syndica e 2023).
Indus ial policy means di e en hings o
di e en people, and while he e m is widely
used, he e is no single way o desc ibing i .
9AI, Inno a ion and he Public Good: A New Policy Playbook
cen es — hey mus be cons uc ed wi h speci ic
uses o con ibu e o na ional inno a ion s a egies.
When go e nmen s in es in da a cen es, hey
should conside in oducing condi ionali ies,
s anda ds and gua d ails, including measu es
mi iga ing en i onmen al ha ms and s imula ing
local indus ial capabili ies, such as de eloping
he wo k o ce o enhancing skills.
In some cases, hos ing a da a cen e is iewed as
he only indus ial policy s a egy o coun ies,
pa icula ly smalle ones ha lack he scale o he
da a and compu e esou ces equi ed o de elop
la ge AI models. This iew is sho sigh ed. As
no ed ea lie , he AI landscape is highly une en
and domina ed by a ew companies. Howe e ,
his does no mean ha smalle coun ies mus
ely solely on hose companies o build hei AI
in as uc u e dependen on hem. Whe he i is
a la ge o small economy, he e a e pa hways o
es ablish some deg ee o digi al so e eign y and
de elop na ional AI models, e en on a modes
scale. The cu en ixa ion on la ge models leads
back o big ech, ein o cing he exis ing powe
imbalances. Wi h he igh policies and a clea
ision o building s a egic independence,
coun ies can b eak his icious cycle and os e
sus ainable and sel - elian AI ecosys ems.
NIS
The e i al o indus ial policy has inspi ed
ex ensi e policy discussions among economis s,
digi al policy specialis s and public policy expe s
on how much coun ies can use “old bu new”
indus ial policies o build in as uc u e, capaci y
and compe i i eness in AI. Each indus ial policy
discussion, whe he ocused on manu ac u ing,
g een ansi ion o AI, comes wi h i s own
unique se o s akeholde s, in e nal ensions,
powe dynamics and unde lying echnological
condi ions (Es e ez 2023), much like he AI
landscape i sel . Ea lie sec ions add essed some
o hose issues, including he “old bu new”
concep o indus ial policy, powe dynamics a
play and unde lying echnological condi ions o
AI. This sec ion shi s he ocus o he ques ion
o “how” and ou lines key policy p oposals.
In his con ex , ecen yea s ha e seen a g owing
unde s anding o indus ial policy and an expanding
li e a u e ha p o ides igo ous e idence on how
i wo ks a he han deba ing whe he i wo ks.
Mode n indus ial policy is inhe en ly complex
and o en consis s o many dis inc ou wa d-
o ien ed policy le e s. Inno a ion policy, a ma u e
ield ha o e laps wi h indus ial policy (Juhász,
Lane and Rod ik 2023), o e s aluable amewo ks
and lessons o de eloping be e insigh s in o
wha wo ks and unde wha condi ions.
A sys em o inno a ion ypically ou lines he
key economic, social, poli ical, o ganiza ional
and ins i u ional ac o s ha impac i s
de elopmen and dissemina ion. Collec i ely,
hese elemen s shape how inno a ions a e
gene a ed and emb aced wi hin an economy.
Beng -Åke Lund all in oduced he e m “inno a ion
sys em” in 1985 o desc ibe he in e ac ion be ween
i ms and ins i u ions in ol ed in knowledge
p oduc ion. He emphasized ha inno a ion eme ges
no om isola ed ac o s bu om he ela ionships
be ween a ious o ganiza ions, including basic
esea ch ins i u ions, applied esea ch cen es,
uni e si ies (as knowledge p oduce s) and indus ies
(as knowledge use s). This pe spec i e unde sco ed
he impo ance o he in e play be ween supply
and demand in d i ing inno a ion (Kilic 2014).
Mode n inno a ion heo y builds on his
concep by ocusing on he in e ac ions be ween
di e en ac o s and ins i u ions in ol ed in he
inno a ion p ocess. Ins ead o ea ing hese
ields sepa a ely, i combines hem in o a single
policy amewo k. This app oach also ocuses
on he b oade cul u al landscape o ins i u ions
engaged in scien i ic esea ch, knowledge
dissemina ion, employee educa ion and echnology
de elopmen . This includes policy amewo ks
( egula ions, laws, s anda ds) and go e nmen
in es men s in in as uc u e (Ca acos as 2007).
A coun y’s inno a ion success la gely depends on
an e ec i e NIS, which includes a ious sec o s
ha ex end hei in luence h ough esea ch,
en ep eneu ial ac i i ies o policy making. A c ucial
aspec o his sys em is he s ong connec ion
be ween lea ning and inno a ion, shaped by
ins i u ional amewo ks, es ablished p ac ices,
s anda ds and ules ha guide in e ac ions
among hese g oups (Kuhlmann 2003).

10 CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
I is widely ecognized ha no wo inno a ion
sys ems a e iden ical, jus as no wo socie ies
a e he same (Edquis 2005, 182). The p ocess
o building in as uc u e and dis ibu ing
echnological inno a ion a ies om coun y
o coun y, as does he ole o policy in
suppo ing i . This is whe e he NIS amewo k
becomes ele an : he o e all design o na ional
inno a ion policy should add ess echnological
inno a ion comp ehensi ely, conside ing each
coun y’s unique needs and capabili ies.
The essence o he NIS lies in i s dynamic na u e,
which is shaped by di e en ins i u ional ac o s,
such as laws, social and cul u al no ms, ou ines
and habi s, which guide he in e ac ions among
inno a ion ac o s (Nelson and Rosenbe g 1993).
Mode n inno a ion policies combine science,
echnology and indus ial policy and a e d i en
by i e essen ial p ocesses: knowledge, supply
o skills, demand o inno a ion, inancing o
inno a ion and shaping o ins i u ions ( o
example, laws and egula ions). Ra he han
unc ioning as subs i u es, hese elemen s wo k
oge he o c ea e and sus ain echnological
p og ess (Fage be g and Hu schen ei e 2020).
The NIS app oach le e ages exis ing knowledge
o build on skills, echnology and connec ions
while ailo ing s a egies o a coun y’s unique
needs and ci cums ances. I is g ounded in
ealis ic, achie able goals a he han policies ha
look good on pape . A obus NIS in AI depends
on how e ec i ely coun ies can build a s ong
AI echnology ecosys em, co e ing e e y hing
om basic esea ch o end-use applica ions.
I is no an easy ask o build suppo o an
inno a ion sys em. The e is no doub ha he e
would be esis ance om es ablished economic
in e es s, neolibe al schola s and o he s skep ical
o go e nmen -led ini ia i es. They may ques ion
he need o bene i s o indus ial policy on g ounds
such as a lack o in as uc u e, a small economy
o a lack o skills. The c i ical i s s ep is achie ing
consensus among policy make s and s akeholde s
on he impo ance o his mission and he necessi y
o echnological independence. Equally impo an
is unde s anding and e ec i ely engaging wi h he
exis ing s uc u es and ins i u ions cen al o local
inno a ion. De eloping a whole-o -go e nmen
app oach o AI wi hin he NIS amewo k is c ucial
o c a ing an indus ial policy ha is bo h o wa d-
looking and wel a e-o ien ed. This app oach can
suppo he domes ic indus y while add essing
social and en i onmen al challenges speci ic o
each coun y, ensu ing ha inno a ion aligns wi h
b oade na ional p io i ies and socie al needs.
To in eg a e AI wi hin a b oade NIS amewo k,
se e al complemen a y policy measu es can
be adop ed, including bu no limi ed o:
→ unde s anding local con ex and add essing local
needs;
→ se ing p io i ies o enhance domes ic
inno a ion capabili ies;
→ in es ing in AI R&D;
→ o ganiza ional inno a ion and es uc u ing o
public ins i u ions; and
→ shaping o policies (aligning ade policy wi h
indus ial policy and compe i ion).
The ollowing policy measu es d aw on pas
success s o ies and insigh s om cu en esea ch.
They a e in en ionally b oad and adap able,
encou aging inno a i e hinking ha educes
eliance on big ech. These ecommenda ions
can be ailo ed o align wi h each coun y’s
unique condi ions, p io i ies and goals.
They se e as a call o ac ion, u ging policy
make s o e hink con en ional inno a ion
s a egies, explo e new pa adigms and de elop
amewo ks ha empowe local ecosys ems.
Unde s anding Local Con ex
and Add essing Local Needs
Unde s anding he local con ex and add essing
speci ic needs is c ucial when designing an
inno a ion s a egy. Un o una ely, many
policies de eloped, ecommended and
p omo ed globally o en lack an unde s anding
o local needs and condi ions.
Many ech policies, o ins ance, a e c a ed in
Eu ope and hen adop ed in o he coun ies
unde a one-size- i s-all model. The Gene al
Da a P o ec ion Regula ion (GDPR) exempli ies
his end. As a pionee ing p i acy egula ion,
he GDPR quickly became a global empla e.
Howe e , many coun ies eplica ed i in hei
legisla ion wi hou ailo ing i o hei speci ic
con ex s. This led o limi ed en o cemen as many
coun ies lacked he ins i u ions, legal ools and
s akeholde s needed o e ec i e en o cemen .
11AI, Inno a ion and he Public Good: A New Policy Playbook
The much-celeb a ed “B ussels e ec ” has, in
many cases, u ned ou o be hype, unde mining
he signi icance o he GDPR as a g oundb eaking
amewo k o p i acy igh s and p o ec ions.
While GDPR-s yle laws a e widesp ead, he e is
no/limi ed en o cemen agains big ech companies.
Da a ee- low p o isions in ade ag eemen s and
p ac ices o en shield hese companies om local
p i acy egula ions, lea ing domes ic companies
o bea he compliance bu den. This c ea es an
une en playing ield, s i ling local inno a ion
while enabling big ech o domina e and shape
he digi al ecosys em in hese coun ies.
The same end is eme ging wi h AI- ela ed
egula ions and s a egies. As mo e coun ies
pu sue AI egula ion o de elop na ional AI
amewo ks, hey isk adop ing s a egies o
legisla ion ha may no ully align wi h hei
AI ecosys ems and u u e p ospec s and may
c ea e an une en ield o local inno a o s.
Indus ial policy is desc ibed as a sea ch p ocess
ha equi es embeddedness — a close collabo a ion
be ween go e nmen and local indus ies wi hou
allowing es ed in e es s o domina e o exe
undue in luence. This p ocess is illed wi h
unce ain ies, especially when pu suing highly
ambi ious goals, as go e nmen s o en lack de ailed
knowledge o he indus y and capabili ies and
he echniques a ailable o sol e hem. To add ess
his, i is ecommended ha go e nmen s and
businesses engage in meaning ul dialogue o ga he
in o ma ion, assess capabili ies and c ea e syne gies
wi h o he policies (Aiginge and Rod ik 2020).
Each coun y has di e en capabili ies and
in as uc u e ega ding AI. Policies should
be based on hese eali ies while add essing
he speci ic challenges ha a coun y
aces. Indus ial policy seeks o p o ide
ins i u ional solu ions o hose challenges.
Ra he han simply adop ing AI echnologies
o digi alizing public se ices elying hea ily
on he in as uc u e o dominan companies,
coun ies should conside de eloping policy
solu ions ha educe hei nea - o al dependence
on big ech in as uc u es, R&D esou ces, skills
and labou , and alue chains unde pinning all
digi al expe iences, whe he AI o b oade digi al
se ices. By doing so, hey can wo k owa d a mo e
independen and esilien digi al ecosys em.
I is essen ial o conside each coun y’s unique
con ex ual and ins i u ional ac o s. E en i wo
coun ies impo he same echnologies, hey
a e unlikely o achie e simila p og ess due o
di e ences in local ins i u ions, capaci y, skills and
social s uc u es. The NIS in ol es no only acqui ing
echnology bu also o ganizing, coo dina ing
and managing ela ed ac i i ies wi hin a ailo ed
ins i u ional amewo k. A coun y can build i s
social capaci y by wo king wi hin i s legal, economic
and scien i ic ins i u ions, adap ing hese s uc u es
o mee i s needs and eali ies (Odagi i e al. 2010).
While indus ial policy will na u ally di e ac oss
coun ies a di e en s ages o de elopmen ,
he e a e oppo uni ies o mu ual lea ning
and sha ed insigh s. In he con ex o AI, his
in ol es unde s anding he AI ecosys em,
add essing echnological dependencies, and
se ing p io i ies and goals ha align wi h local
eali ies o minimize and ul ima ely educe
deepening eliance on big ech companies.
Se ing P io i ies o Enhance
Domes ic Inno a ion Capabili ies
Indus ial policy is a lexible bu complex ool.
The goals o indus ial policy di e widely and
equi e di e en s a egies. I he objec i e is
inno a ion, hen R&D incen i es a e essen ial.
I he goal is building an AI-skilled wo k o ce,
s a egies o skill building a e needed — ocused
on se ices, STEM (science, echnology, enginee ing
and ma h) educa ion, linking uni e si y esea ch
o indus y, and suppo o SMBs. I i is abou
building public in as uc u e o AI, he ocus
shi s o add essing in as uc u al dependencies,
educing ma ke concen a ion and p omo ing
public good. The e is no magic o mula; each
p io i y demands a unique app oach.
C ea ing u u e jobs is o en highligh ed as a
c i ical goal o indus ial policies o AI. Fo
example, in he Uni ed S a es, e o s ocus on
wo k o ce de elopmen o p epa e bo h cu en and
u u e wo ke s o AI adop ion ac oss all sec o s,
wi h a s ong emphasis on upskilling in li e acy,
nume acy, p oblem-sol ing and p omo ing li elong
ca ee s in AI.4
Eu opean policy make s also p io i ize AI expe ise
o add ess skills and labou sho ages. Thei
4 SeeNa ionalA i icialIn elligenceAd iso yCommi ee(2023).
12 CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
plans include building in as uc u e, os e ing
public-p i a e pa ne ships, s eng hening
academia-indus y collabo a ion and enhancing
STEM educa ion ( ocusing on inc easing emale
pa icipa ion in ech ields) (Pal 2024).
Con empo a y indus ial policies o en p io i ize
AI skill de elopmen and STEM educa ion o
build an AI wo k o ce o he u u e. While he
c ea ion o an AI wo k o ce is an impo an goal
o indus ial policy, i mus be aligned wi h
local needs and p io i ies a he han na owly
ocusing on high- ech jobs a he expense o
o he essen ial occupa ions. Coun ies s ill
need doc o s, eache s, nu ses, plumbe s, uck
d i e s and ca pen e s, which suppo he
o e all unc ioning o he na ional economy.
Ano he c i ical poin o conside is ha
misaligned ma ke incen i es wi h socie al
objec i es o en dis o he inno a ion p ocess.
Fo ins ance, he impe ec ions o he labou
ma ke can c ea e a di e gence be ween
he social cos o labou and ma ke wages,
which may skew echnological de elopmen
owa d au oma ion a he han wo ke -
complemen a y echnologies (Acemoglu 2023).
To ensu e sus ainable g ow h, he AI labou
ma ke and skills de elopmen should be
in eg a ed in o he NIS amewo k o de elop
domes ic inno a ion capabili ies. A nuanced
app oach is necessa y o balance ma ke dynamics
wi h s a egic public in e en ions ha guide
inno a ion owa d b oade socie al bene i s. This
app oach will help build on exis ing s eng hs
while add essing gaps in he labou ma ke . Doing
so can c ea e a balanced ecosys em o skills and
indus ies, ensu ing bo h economic g ow h and
echnological esilience o e he long e m.
In his con ex , uni e si ies play a c ucial ole,
and building linkages be ween local indus y and
uni e si ies becomes equally impo an . Mode n
inno a ion heo y emphasizes he in e ac i e na u e
o he inno a ion p ocess, wi h uni e si ies playing
a key ole as collabo a o s. Knowledge-based
inno a ion sys ems inc easingly adop he iple-
helix model, which cap u es mul iple ecip ocal
ela ionships be ween uni e si ies, indus y and
go e nmen a a ious le els. Silicon Valley is
one o he mos ci ed examples o he iple-helix
model o inno a ion, wi h each helix building
upon and ein o cing he o he . S an o d Uni e si y
played a c ucial ole in he egion’s de elopmen ,
suppo ed by go e nmen ini ia i es ha enabled
Silicon Valley o become a global inno a ion hub.
This syne gy a ac ed and ci cula ed alen and
echnology in e na ionally, making i a leade
in inno a ion (E zkowi z and Zhou 2017).
Coo dina ion ailu es be ween uni e si y
esea ch and indus y can dis o he di ec ion o
inno a ion and hinde he e ec i e ansla ion
o esea ch in o echnological inno a ions.
B inging inno a ion om he esea ch lab o
he ma ke o en in ol es collabo a ion among
di e se s akeholde s, including esea che s,
companies, uni e si ies, go e nmen agencies,
angel in es o s, en u e capi alis s and o he i ms
along he supply chain. Howe e , companies o en
s uggle o ind he igh pa ne s— whe he a
uni e si y, a esea ch ins i u e o an in es o —
wi h he necessa y expe ise, esou ces and
us wo hiness. These ne wo k ailu es p e en
p omising inno a ions om eaching he
ma ke , s alling p og ess and limi ing inno a ion
oppo uni ies in he coun y (Acemoglu 2023).
The US go e nmen ’s s a egy o in es ing in
esea ch cen es ha b ing oge he publicly and
p i a ely unded echnologis s is widely ega ded as
a bes p ac ice in inno a ion policy. A good example
o his app oach is he Ad anced Manu ac u ing
Ins i u es (AMIs), a di ec ini ia i e o os e
collabo a i e inno a ion and p oduc ion. Building
on a adi ion o s a e-sponso ed collabo a i e
inno a ion and p oduc ion da ing back o he
1970s, he AMIs comp ise 45 ins i u es na ionwide.
They a e designed o add ess ne wo k ailu es,
connec ing a ne wo k o businesses, uni e si ies
and labo a o ies. Each ins i u e specializes in
speci ic echnology and se es as a hub o a local
clus e o companies and expe ise. The ins i u es
p o ide a pla o m o go e nmen leade ship and
in e en ion. They play a c i ical ole in b inging
pa ne s oge he , coo dina ing inno a ion e o s,
ce i ying compe ence and us wo hiness, and
mi iga ing conce ns abou in ellec ual p ope y (IP)
he . The AMIs ha e achie ed signi ican success
in enhancing ne wo k connec i i y, lowe ing
he cos s o sea ching o he igh pa ne s,
add essing collec i e ac ion p oblems, de eloping
echnology oad maps and in es ing in wo k o ce
de elopmen . AMIs con ibu ed o he de elopmen
o esilien , inno a i e and collabo a i e indus ial
ecosys ems (Block, Kelle and Negoi a 2020).
The Uni ed S a es is no he only example o
a coun y wi h go e nmen -d i en success ul
13AI, Inno a ion and he Public Good: A New Policy Playbook
inno a ion p og ams ha os e coope a i e
ela ionships by b inging oge he uni e si ies,
inno a o s, companies, in es o s and policy
make s. Success ul inno a ion policies ha
e ec i ely add ess ne wo k ailu es ha e been
implemen ed in coun ies such as Denma k,
Finland, I eland, Is ael and Taiwan (ibid.). These
p og ams acili a e sha ed esou ces, knowledge
exchange and collabo a i e p oblem-sol ing.
They also help educe ansac ion cos s and
minimize he isk o echnological s agna ion,
s eng hening domes ic local capabili ies.
The p oduc i e inno a ion policies adi ionally
applied o manu ac u ing can also be adop ed
o AI. They p o ide a amewo k o os e ing
echnological ad ancemen , enhancing domes ic
capabili ies, imp o ing wo k o ce skills, c ea ing
good jobs and p omo ing collabo a ion among
key s akeholde s. In he con ex o AI, such
p og ams can be designed o b ing oge he local
esea che s, uni e si ies, echnologis s, companies
and in es o s o build an equi able in as uc u e
ha p o ides access o compu e powe and
clean da a se s. This amewo k would p omo e a
collabo a i e model, empowe ing ci il socie y, local
communi ies, esea che s and local inno a o s
o pa icipa e in designing and de eloping AI
sys ems. In he long un, his would educe
eliance on big ech companies o in as uc u e,
public se ices and echnological needs.
A s a egic policy app oach should p io i ize
educa ional and ins i u ional capaci y building
o empowe local companies and s a -ups o
scale esponsibly. Go e nmen demand can be
a powe ul d i e o local inno a ion, whe he
by p ocu ing new AI sys ems o in es ing in
in as uc u e. This equi es designing go e nmen
p ocu emen p ocesses o suppo local inno a ion
whe e easible and p io i izing domes ic playe s
while aking in o accoun ade commi men s
unde a ious ag eemen s. Such p ocu emen
policies can c ea e oppo uni ies o le el he
une en playing ield and p omo e a mo e
equi able and dynamic inno a ion ecosys em.
In es ing in AI R&D
Economis s ha e long ecognized ha ma ke
o ces alone may no dis ibu e su icien
esou ces o esea ch and inno a ion,
jus i ying go e nmen suppo o inno a ion
h ough in es men in esea ch in as uc u e
o R&D ax c edi s (Acemoglu 2023).
In he ea ly wen ie h cen u y, he Uni ed S a es
and Ge many ad anced apidly in science-based
indus ies, la gely because hei uni e si y
sys ems we e highly esponsi e o he demands
o eme ging echnologies. Simila ly, Japanese
uni e si ies played a c ucial ole in Japan’s ea ly
indus ializa ion by helping local indus ies
upg ade hei echnological capaci ies. This
uni e si y-indus y collabo a ion ga e Japanese
indus ies a signi ican compe i i e edge, se ing
he s age o long- e m success (Kilic 2014).
Basic academic esea ch lays he g oundwo k o
inno a ion. His o ically, he e has been a signi ican
di ide be ween academia and indus y ega ding
basic esea ch. La ge co po a ions adi ionally
p io i ized sho - e m deli e ables and pa en s
o e undamen al esea ch o academic publishing.
Howe e , ech companies ha e inc easingly
shi ed his dynamic by hea ily in es ing in basic
esea ch and academic publishing, b eaking om
he indus y pa e n o elying on uni e si ies
o basic esea ch (Ahmed and Wahed 2021).
This shi highligh s a b oade issue: i is no
only ma ke s, in as uc u e and compu e powe
ha a e domina ed by ech companies, bu
esea ch i sel is also inc easingly monopolized
by big ech, consolida ing con ol o e
inno a ion and knowledge p oduc ion.
The linea inno a ion model, s uc u ed a ound
h ee s ages — basic esea ch, applied R&D
and di usion — ope a es on he p inciple ha
“science in en s, indus y adap s and socie y
con o ms.” Uni e si ies a e conside ed cen al
o his model, jus i ying he use o public unds
o basic esea ch and R&D ac i i ies. While he
model has been ins umen al in unding basic
esea ch, i aced c i icism o o e simpli ying
he inno a ion p ocess and limi ing uni e si ies
o basic esea ch. I ailed o accoun o he
sys ema ic and in e ac i e na u e o inno a ion,
igno ing he dynamic ela ionships be ween
a ious ac o s in he ecosys em (de Oli ei a 2014).
In he con ex o AI echnologies, basic esea ch
is essen ial, bu i canno be he only ocus o
R&D e o s. Wi h ech companies inc easingly
domina ing he basic esea ch space, public R&D
ini ia i es should go beyond basic esea ch o
suppo applied esea ch, os e collabo a ion wi h
he local indus y, and add ess b oade socie al
and economic dimensions o AI de elopmen .
14 CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
The public has a s ong in e es in ensu ing
AI models a e us wo hy and suppo ed.
The e is a clea dis inc ion be ween p i a e and
public in e es s in AI esea ch. Big ech esea ch
inc easingly ocuses on ad ancing on ie
LLMs, while c i ical a eas such as obus ness,
in e p e abili y, ai ness and secu i y ecei e
a less a en ion. To add ess hese gaps, public
unding should p io i ize a eas ha align wi h
b oade socie al needs, including in e p e abili y,
de ensi e cybe secu i y, benchma king and
e alua ions, and p i acy-p ese ing machine
lea ning. These domains a e essen ial o
ensu ing AI sys ems a e eliable, equi able and
secu e in hei applica ions (Wa ney 2023).
AI esea ch demands cos ly compu a ional
esou ces, including specialized ha dwa e
designed o mee he immense demands o
la ge machine-lea ning models. This c ea es a
compu e di ide, whe e ech companies (wi h
hei ex ensi e in as uc u e and esou ces)
and eli e uni e si ies (o en backed by simila
esou ces o pa ne ships wi h ech i ms) hold
signi ican ad an ages o e non-eli e uni e si ies,
which lack bo h (Ahmed and Wahed 2021).
Tech companies and, o a ce ain ex en , eli e
uni e si ies bene i om hei abili y o ec ui
op alen and access he la ge, high-quali y
da a se s essen ial o aining ad anced AI
models. Wi hou such esou ces, non-eli e
uni e si ies s uggle o con ibu e o mode n AI
esea ch. This esou ce imbalance h ea ens o
unde mine he long- e m esea ch and aining
unc ions adi ionally pe o med by uni e si ies,
hobbling hei abili y o sus ain inno a ion and
educa e he nex gene a ion o AI alen .
I has been sugges ed ha go e nmen s in es
in a “na ional esea ch cloud” o add ess he
compu ing di ide be ween ech companies and
uni e si ies.5 I is c ucial ha such in as uc u e
emains independen o ech companies and
p omo es public-in e es esea ch ee om
co po a e in luence. I is also essen ial o ensu e ha
his in es men does no inad e en ly become a
esea ch subsidy o ech companies (Ing am 2021).
5 S an o d Ins i u e o Human-Cen e ed AI o P esiden Donald T ump and
Membe s o Cong ess, n.d., h ps://hai.s an o d.edu/
na ional- esea ch-cloud-join -le e .
The Uni ed Kingdom, o ins ance, has announced
plans o es ablish a “Na ional Da a Lib a y,” a
comp ehensi e da a se o go e nmen -con olled
public eco ds. The ini ia i e aims o a ac
leading AI companies o collabo a e wi h he
Uni ed Kingdom. I o e s access o in-dep h,
eal-wo ld da a o d i e he de elopmen o
homeg own AI models. Howe e , beyond he
p i acy and secu i y conce ns al eady aised
(Milmo and S acey 2025), c i ical ques ions emain
abou who will be he p ima y bene icia y o his
public da a. While he Uni ed Kingdom appea s
de e mined o es ablish a homeg own so e eign
AI ecosys em, he success o his ini ia i e will
depend on he adop ion and execu ion o e ec i e
complemen a y policies. Wi h he igh s a egies
in place, he Uni ed Kingdom could become an
AI success s o y discussed in he yea s o come.
Fos e ing Pa icipa o y and
Coo dina ed Go e nance
His o ically, indus ial policies ha e been
designed op-down, a ge ing p e-selec ed sec o s
and elying on a s anda d lis o subsidies and
incen i es. This model was p e alen in coun ies
such as Japan, Sou h Ko ea, Taiwan and some
Eu opean coun ies (Aiginge and Rod ik 2020).
Bo h Japan’s and Sou h Ko ea’s success s o ies we e
buil on s ong, o malized, go e nmen -d i en
policy planning and s ong poli ical leade ship. In
bo h cases, he go e nmen s eo ganized hemsel es
unde hei NIS goals, ac i ely engaged wi h key
s akeholde s and ecalib a ed policies o ma ch hei
espec i e s ages o de elopmen . While he e we e
di e ences in app oaches and speci ic policies, bo h
e o s we e highly s uc u ed and well o ganized.
Ne e heless, mode n indus ial policies ha e
shi ed away om op-down app oaches. They
a e designed o os e sus ained collabo a ion
be ween he public and p i a e sec o s o
ad ance p oduc i i y and achie e social goals. The
design o NIS equi es no only well-de eloped
analy ical capabili ies in policy making bu also
e ec i e coo dina ion among a ious ac o s. Such
coo dina ion can help policy make s connec
inno a ion policy wi h b oade s a egic goals ( o
example, add essing socie al challenges such as
clima e change, social inequali y o echnological
esilience). A sha ed ision o mission can help
achie e he necessa y coo dina ion in inno a ion
policy (Fage be g and Hu schen ei e 2020).
Consequen ly, a success ul NIS s a egy equi es

15AI, Inno a ion and he Public Good: A New Policy Playbook
he public sec o o a icula e a clea ision and
pu pose and p io i ize c ea ing collabo a i e
and pa icipa o y ins i u ional s uc u es. These
s uc u es a e subjec o con inuous moni o ing and
e ision based on ou comes, ensu ing ha policies
emain esponsi e o changing ci cums ances
and e ec i ely se e hei in ended pu poses.
A good example o his coo dina ed app oach is
Finland, an indus ial la ecome ha achie ed
signi ican p og ess in he wen ie h cen u y,
pa icula ly in he in o ma ion and communica ions
echnology (ICT) sec o . The Resea ch and
Inno a ion Council (RIC), chai ed by he p ime
minis e , b ough oge he public and p i a e
inno a ion ac o s. The council played a c ucial
ole in e alua ing policy e ec i eness and shaping
inno a ion policy in he coun y. Finland aced
signi ican challenges wi h i s ICT indus y in he
ea ly 2000s, pa icula ly wi h he decline o i s
na ional champion, Nokia. The RIC became less
p ominen han i once was. Howe e , Finland’s
ecen shi owa d mo e c oss-sec o al and
ans o ma i e R&D and inno a ion p og ams
ein o ces he impo ance o high-le el policy
coo dina ion. These p og ams equi e adap a ions
in he ins i u ional and egula o y amewo k o
succeed (Fage be g and Hu schen ei e 2020).
Fo mode n indus ial policies o succeed, hey
mus be delibe a ely sus ainable, public-o ien ed
and led by local inno a ion. These policies should
be coo dina ed as pa o a holis ic package and
implemen ed in coope a ion wi h go e nmen
agencies and local indus ies. Chinese NIS is
ano he powe ul example o how s a egic
go e nmen in e en ion can d i e apid
economic di e si ica ion and s uc u al change.
China has in es ed billions in sec o s such as
elecommunica ions, in o ma ion echnology, ca
manu ac u ing and s eel. By s a egically di ec ing
esou ces and os e ing coo dina ion be ween
go e nmen , indus y and esea ch ins i u ions,
China’s NIS suppo ed i s echnological ca ch-
up and buil global compe i i eness, especially
in AI (Lund all and Rikap 2022). While many
coun ies ha e ied o copy aspec s o China’s
s a egy, hey o en all sho due o insu icien
capi al, lack o ins i u ional connec ions,
weak poli ical will o misaligned p io i ies
( o example, ocusing on o e ly ambi ious
goals such as de eloping la ge AI models).
Simila o Finland, Sweden and he Ne he lands
ha e ins i u ionalized high-le el policy coo dina ion,
wi h he p ime minis e aking he lead o se
he di ec ion and acili a e policy coo dina ion.
This kind o high-le el coo dina ion ex ending
beyond he emi o inno a ion agencies has
led o he eme gence o “whole-o -go e nmen ”
app oaches. These amewo ks emphasize he
impo ance o policy coo dina ion ac oss a ious
policy a eas and o ganiza ional bounda ies o
sec o minis ies. Such coo dina ed e o s a e
especially ele an o con empo a y inno a ion
policies, which span mul iple policy domains
and in ol e di e se s akeholde s. In ac , as
inno a ion policies e ol e o add ess complex
challenges and g ow mo e ambi ious, e ec i e
coo dina ion — ex ending beyond con en ional
s akeholde s o include academia, esea che s
and ci il socie y — becomes e en mo e c i ical
(Fage be g and Hu schen ei e 2020).
Maximizing AI’s economic po en ial elies
hea ily on access o compu e, da a, ene gy and
human capi al. Howe e , wi h an unp eceden ed
concen a ion o capi al, da a, alen and esou ces
in he hands o a ew big ech companies, he
na a i e su ounding AI and indus ial policy has
become hea ily dependen on hese companies.
Many policy p oposals all sho o o e ing an
al e na i e ision because doing so equi es
no only e ec i e policies bu also bold and
inno a i e hinking ha challenges long-held
economic heo ies and assump ions, along wi h
subs an ial inancial in es men and poli ical will.
Go e nmen s should no gi e up in he ace o
challenges and unce ain y. Ins ead, hey should
emb ace hese complexi ies as oppo uni ies o
shape policies ha ebuild he AI ecosys em om
he g ound up. This calls o bold and decisi e
poli ical leade ship, a “whole-o -go e nmen ”
app oach and pa icipa o y policy making, whe e
ci il socie y, local communi ies, wo ke s and
esea che s can help design AI policies. S a ing
small, ocusing on smalle AI models can p o ide
a mo e p ac ical and achie able ounda ion o AI
de elopmen . Rejec ing he blind eplica ion o he
US ech model cha ac e ized by da a ex ac ion,
commodi ica ion and ma ke concen a ion
c ea es a space o esponsible, equi able and
democ a ic inno a ion ha p io i izes p oduc i i y
and social goals. Complemen a y policies, such
as IP e o m, compe i ion policy, algo i hmic
anspa ency and accoun abili y, a e c i ical o
shaping indus y ou comes and os e ing a obus
inno a ion ecosys em. Ul ima ely, i all comes
16 CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
down o balancing he need o os e inno a ion
wi h se ing he public in e es . Poo ly c a ed
policies can ha e signi ican and a - eaching
consequences o public in e es and digi al
public in as uc u e. They isk concen a ing
powe wi hin exis ing dominan playe s, limi ing
oppo uni ies o local inno a o s, and u he
deepening he knowledge and echnology gap
be ween big ech companies and small local playe s.
Fo mula ing and implemen ing e ec i e inno a ion
policies equi es a b oad unde s anding o local
con ex s. The e is no uni e sal, one-size- i s-
all s a egy ha e e y coun y can adop . The
applica ion o NIS di e s ac oss coun ies, bu he
key elemen s emain consis en . O ganiza ional
inno a ion and a “whole-o -go e nmen ”
app oach can be ans o ma i e o he public
sec o , no necessa ily h ough digi aliza ion
alone bu h ough go e nance and egula ion o
inno a ion. I equi es e hinking o ganiza ional
s uc u es, imp o ing p ocesses and de eloping
policies based on new na a i es ailo ed o each
coun y’s unique challenges and oppo uni ies.
Shaping o Policies: Aligning
T ade Policy wi h Indus ial
Policy and Compe i ion
Ano he c i ical componen o NIS is i s ins i u ions,
which include policies, ules and egula ions
designed o suppo he coun y’s inno a ion
goals and p io i ies. Mode n indus ial policies
employ a dynamic mix o policies and egula ions.
Ensu ing policy alignmen is c ucial o success.
In ecen yea s, he “whole-o -go e nmen ”
app oach has become inc easingly impo an
in add essing complex policy challenges, such
as inno a ion, compe i ion and de elopmen .
Applying his app oach o AI inno a ion could
p o e ins umen al in add essing some o he
p essing challenges go e nmen s ace.
One o hese challenges is he ma ke concen a ion
in AI and he g owing in luence o AI companies
in shaping echnological, poli ical and policy
landscapes ac oss bo de s. Fo big ech, inno a ion
o en se es as a ool o maximizing p o i s o
ex ending in luence. Wi h only a ew companies
con olling essen ial digi al in as uc u e, da a
lows and compu ing powe , meaning ul p og ess
equi es add essing his concen a ion o powe .
The success o indus ial policies depends hea ily
on hei abili y o build a digi al ecosys em ha
is independen o US o Chinese ech gian s. This
is why i is c ucial o ensu e AI emains open and
compe i i e. Coun ies can le e age hei exis ing
an i us powe s o in es iga e and p ohibi
un ai and an i-compe i i e p ac ices such as
sel -p e e encing, ying, exploi ing cus ome s
and es ic ing access o key inpu s. Exis ing
laws al eady co e many o hese monopolis ic
beha iou s, and c acking down on hem would
p o ec he abili y o new en an s o challenge
ech gian s, d i ing inno a ion and expanding
choices o businesses and consume s.
S uc u al in e en ions ( o example, blocking
an i-compe i i e me ge s o imposing binding
condi ions such as di es men o asse s) a e mo e
e ec i e han beha iou al emedies, which ocus
on egula ing a company’s beha iou a he han
changing i s s uc u e. They a e o en complex
o moni o , easy o bypass and quickly become
ou da ed. Compe i ion au ho i ies wo ldwide,
including hose in A ica, Asia, Aus alia, Canada
and La in Ame ica, ha e he powe o block me ge s
o impose condi ions o p e en dominan ech
companies om solidi ying hei con ol o e
AI ma ke s h ough me ge s and acquisi ions
o small playe s ( on Thun and Hanley 2024).
Compe i ion policy should no unc ion in isola ion
bu a he align wi h a coun y’s indus ial and
ade policies, ensu ing i is no ea ed as an
a e hough . E ec i e compe i ion policy equi es
s iking a balance be ween sho - and long-
e m p io i ies, p ice e ec s e sus in es men
incen i es and consume in e es s e sus local
indus ies. Ye con en ional compe i ion policy
o en ails o ully add ess hese adeo s, as i
emains g ounded in a neolibe al amewo k ha
p io i izes p oduc i e e iciency and elies on
ou da ed ools o maximize he “na ow concep ”
o consume wel a e, o en p io i izing consume
in e es s o e b oade na ional economic goals. I
p esumes ha compe i ion na u ally deli e s low
p ices, inno a ion and op imal le els o compe i i e
in es men , wi h limi ed sc u iny o whe he
his holds ac oss sec o s, ypes o in es men
o inno a ion landscapes (Ca a a 2024).
In eg a ing compe i ion policy in o na ional
inno a ion s a egies can es ablish new pa ame e s
beyond he ou da ed consume wel a e s anda d. A
compe i ion policy ha p ima ily a ge s consume
wel a e may con lic wi h he indus ial policy ha
os e s p oduc i e and dynamic indus ies (Aiginge
and Rod ik 2020). Aligning compe i ion policy
17AI, Inno a ion and he Public Good: A New Policy Playbook
wi h a coun y’s inno a ion agenda allows i o
e ol e in o a ool ha os e s bo h inno a ion and
economic de elopmen , making i a mo e in eg al
pa o na ional economic policy. When ca e ully
designed, indus ial and compe i ion policies
can wo k oge he o os e inno a ion, ma ke
ai ness and sus ainable de elopmen . Wi hou
such e o s, coun ies isk emaining pa icipan s
a he han c ea o s in he digi al economy.
In addi ion o compe i ion policy, ade policy
plays a c i ical ole in shaping economic s a egies,
pa icula ly in indus ial de elopmen and
echnological p og ess. His o ically, ade policy
has been a key componen o indus ial policies
h oughou he wen ie h cen u y. Howe e , in
he con ex o AI indus ial policy discussions,
i s signi icance has been la gely o e looked. This
can be a ibu ed o se e al ac o s, including
he complexi y o ade policy, he lack o dep h
in discussions on digi al echnologies, and he
endency o sideline ade issues because hey a e
seen as oo echnical o no popula . T ade policy
and i s connec ion o indus ial policy ecei ed
mo e a en ion a e he elease o he D aghi (2024)
epo on he u u e o Eu opean compe i i eness.
The D aghi epo ou lines ambi ious measu es
o add ess Eu ope’s economic challenges,
including a la ge-scale p oac i e indus ial
policy, an inno a ion-d i en compe i ion
policy and he s a egic use o s a e aid. I also
emphasizes aligning compe i ion and ade
policy wi h he Eu opean indus ial s a egy
h ough ca e ul, case-by-case analysis a he
han adop ing b oad, gene ic posi ions (ibid.).
While he epo o e s aluable insigh s, a solid
c i ique o he EU egula o y amewo k and bold
policy p oposals, i alls sho ega ding ade policy.
Al hough i acknowledges he impo ance o ade
policy — a posi i e s ep — i emains oo ed in an
ou da ed neolibe al pe spec i e on global ade.
Fo ins ance, i ecommends main aining low ade
ba ie s o digi al goods, se ices and in as uc u e
wi h he Uni ed S a es o ensu e access o he la es
AI models and p ocesso s. While such low ba ie s
may a ou big ech companies — especially
hose wa y o Eu opean egula ions on p i acy,
wo ke s’ igh s, compe i ion and democ acy, o en
dismissed as non- a i ba ie s — i is unclea
how his app oach aligns wi h Eu ope’s indus ial
s a egy o helps he Eu opean Union educe i s
eliance on big ech’s dominance in he AI ma ke .
In ac , indus ial policy and he neolibe al ade
agenda a e undamen ally incompa ible. T ade
policy has adi ionally been designed o es ic
he e y ools on which indus ial policy depends.
Measu es o en dismissed as “ ade ba ie s” by
companies a e, in ac , essen ial o indus ial
policy (Kilic 2024). The neolibe al global ade
sys em ope a es wi h a winne - akes-all mindse ,
limi ing he policy space coun ies need o de elop
indus ial s a egies ha p o ec wo ke s, ci izens,
he en i onmen and democ acy (F ase 2017).
Since he 1980s, he neolibe al consensus has no
only ma ginalized indus ial policy bu also shaped
global ade policy. T ade ag eemen s disman led
ba ie s o ade and inancial lows, educed
egula ion and minimized go e nmen in ol emen
in he economy. Ins i u ions such as he WTO and
ade policy make s ejec ed indus ial policy,
iewing ma ke shaping o na ional in e es s and
alues as incompa ible wi h hei amewo k.
Mil on F iedman, o en ega ded as he mos
p ominen a he o neolibe alism, amed he
dynamic clea ly: he ma ke mus domina e and
democ a ic ins i u ions mus ecede. Ins ead o
democ acy egula ing he ma ke , he ma ke was
asked wi h egula ing democ acy. Poli icians and
policy make s in e nalized his mindse , lea ning
o ope a e wi hin a sys em whe e he absence
o ules became he only ule (Zubo 2022). Fo
decades, his co po a e-cen ic app oach has
esul ed in a ickle-down mindse shaping ade
policy making, u ning i in o a ool o libe alize and
de egula e ma ke s o bene i big playe s and hei
in e es s. Policies we e c a ed om a consume
wel a e pe spec i e, emphasizing he u ili a ian
bene i s o ma ke libe aliza ion, and we e execu ed
in ways ha a ou ed la ge co po a ions.
S uc u al adjus men loans, o eign aid and
WTO egula ions equi e coun ies o educe
and — in some ci cums ances — elimina e many
o he e y indus ial policy ins umen s ha
ha e p o en e ec i e (Schwa zenbe g 2024).
As indus ial policy makes i s comeback, long-
held assump ions abou ade policy also
need e isi ing and e hinking. E ec i e ade
policy oday mus add ess he eali ies o he
compe i i e landscape, including he disad an ages
wo ke s ace due o digi al echnologies, ma ke
concen a ion and un ai compe i ion. I
equi es policy space and lexibili y o suppo
he digi al ans o ma ion o economies.
18 CIGI Pape s No. 318 — Ma ch 2025 • Bu cu Kilic
Unde he Biden adminis a ion, he Uni ed S a es
has shi ed om neolibe alism owa d a mo e
in eg a ed ade and compe i ion policy app oach,
ocusing on imp o ing esilience and cu bing
excess ma ke concen a ion. The adminis a ion
had adop ed a “whole-o -go e nmen ” s a egy o
ensu e ha ade and compe i ion policies we e
p oac i ely aligned and complemen ed US indus ial
policy, especially in he con ex o he digi al
economy (The Whi e House 2021). As an i us policy
mo ed away om solely ocusing on consume
wel a e o b oade conside a ions o keeping
ma ke s open, ee and compe i i e, enhancing
ma ke access and ecognizing wo ke s’ igh s, US
ade policy had simila ly shi ed i s ocus om he
in e es s o US co po a ions and consume wel a e
o wo ke s, a me s, small business owne s and
communi ies, guided by an explici an i us agenda.
In his con ex , he US shi in digi al ade policy
was no able. In Oc obe 2023, he Uni ed S a es
T ade Rep esen a i e e ised i s app oach o se e al
key digi al ade p oposals, g an ing policy make s
g ea e lexibili y and policy space o cu b ech
powe and in oduce indus ial policy measu es.
These p oposals add essed da a ee lows, se e
loca ion equi emen s, and ade sec e p o ec ions
o sou ce code and algo i hms, which all ha e
signi ican socie al and economic implica ions.
Ra he han solely p omo ing big ech’s inno a ion
and compe i i eness, his e ised app oach
challenged he unchecked powe o ech
companies. I emphasized accoun abili y and
esponsibili y in he digi al economy while gi ing
SMBs a ai chance o compe e. This app oach
complemen ed, no o e ode, indus ial policies.
Conside ing he need o coun ies o e ain policy
space o add ess hei digi al de elopmen and AI
in as uc u e needs, he US shi in digi al ade
should encou age mo e coun ies o e isi hei
ade policies and s ep back om he neolibe al
su eillance-capi alis digi al ade amewo k
embedded in he WTO alks and bila e al ade
ag eemen s. T ade ules should align wi h — no
cons ain — compe i ion policies and suppo
measu es ha p omo e economic and digi al
esilience. They should no domina e o dic a e
wha coun ies can o canno do. The cu en
global digi al ade amewo k ails o suppo AI
indus ial policy; ins ead, i ein o ces s uc u al
dependencies and inc eases eliance on big ech
companies. Shaping ade ules o align wi h
indus ial policy objec i es is essen ial o c ea ing
a ai e and mo e inclusi e digi al economy.
Conclusion: The Pa h
Fo wa d
Indus ial policy is simila o o he policy domains,
such as educa ion o heal h ca e, whe e he e a e
s ong jus i ica ions o go e nmen in e en ion
bu no necessa ily de ini i e e idence o suppo
such measu es. In hese ields, esea ch and policy
deba es ypically ocus on unde s anding wha
wo ks, unde wha condi ions he go e nmen
should in e ene and how policy should be
implemen ed (Juhász, Lane and Rod ik 2023).
A simila app oach should apply o indus ial
policy. I is essen ial o look a he success
s o ies o iden i y he mos e ec i e s a egies
and e alua e di e en o ms o implemen a ion
o guide policy design and execu ion.
Since he elease o Cha GPT, he dominan
na a i e a ound AI has ocused on adop ion and
deploymen . This na a i e sugges s ha o achie e
compe i i eness in AI, coun ies should p io i ize
adop ing AI and digi alizing public se ices, heal h
ca e, educa ion and o he sec o s (Be glind, Fadia
and Ishe wood 2022; Delmolino 2024).6 Acco ding
o his iew, a lack o adop ion and deploymen
cons ains AI inno a ion, and inno a ion will
ollow once he e is a wide adop ion o AI.
Bu eal inno a ion does no wo k ha
way— especially ou side he big ech domain.
While a lack o b oad adop ion may hinde
“inno a ion” in a coun y, he mo e c i ical issue
is o en he lack o s a egic di ec ion. Ra he
han simply pou ing esou ces in o blind AI
adop ion and c ea ing AI capaci y dependen
on a ew companies, i is c ucial o hink
s a egically abou how cu en in as uc u e
and capabili ies can be le e aged o suppo
g ow h and he app op ia e policies o building
an independen , publicly d i en AI ecosys em.
A he co e o indus ial policy is he p inciple o
di ec ionali y. The way o wa d equi es a g ea e
6 See h ps://wwps.mic oso .com/blog/ai-public-sec o .