US
Saudi arabia
uae
South korea
japan
eu
China
Computing power
100
0
90
80
70
60
50
40
30
20
10
Capital power
Talent
Energy
IP
Data
Capital power
The US leads on AI venture capital investment between 2019 and 2024 ($303 billion)
US-based tech companies spend the most on R&D investments ($203 billion in 2024)
Data
IP
Energy
Talent
Capital power
Computing power
100
90
80
70
60
50
40
30
20
10
0
Relative performance across enablers, on a normalized scale of 0 to 100
Data
IP
Energy
Talent
Capital power
Computing power
Data
IP
Energy
Talent
Capital power
Computing power
Data
IP
Energy
Talent
Capital power
Computing power
Data
Talent
IP
Capital power
Computing power
Energy
Data
Talent
IP
Capital power
Computing power
Energy
Relative performance across enablers, on a normalized scale of 0 to 100
United States
Second-highest number of AI patent applications (18,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
US-based scholars and institutions have the second-largest number of citations in AI academic publications (~385,000) since 2019
Net exporter of energy; benefits from comparatively low electricity prices
Largest AI talent pool, with some ~500,000 AI specialists as of 2024
Largest share (~60%) of the world’s top AI researchers
Ranks third in mobile broadband subscriptions, with moderate contextualization of data (i.e., ease of integrating and using data from diverse types and sources)
Ranks in the top quintile of the E-Government Development Index, a measure of government digitization
Controls a key bottleneck on the chipmaking manufacturing value chain and has the most reliable access to cutting-edge chips for AI workloads
World’s largest data center, with ~45GW in data center capacity as of 2024
Superpowers and Middle Powers: The Emerging GenAI Landscape
Talent
United States
Intellectual property
United States
Data
United States
Energy
United States
Computing power
United States
Intellectual property
China
Largest share of AI patents since 2019 (60% of total worldwide)
Largest number of AI papers published (173,000) and third-largest number of citations in academic publications (313,000) since 2019
12% of notable machine-learning models have been developed by or in partnership with Chinese scholars and institutions
Capital power
China
Ranks second in government R&D spending ($50 billion) and first in sovereign/pension fund assets for AI funding
Government VC funds have invested ~$110 billion in AI since 2019
Chinese tech companies invest $63 billion annually in R&D (compared to $203 billion for US tech companies)
Talent
China
Largest production of STEM graduates
Second-largest share (18%) of the world’s top AI researchers
Ranks first in mobile broadband subscriptions (~1.5 billion), with a high degree of contextualization of data (i.e., ease of integrating and using data from diverse sources and of various types)
Ranks in the top quintile of the E-Government Development Index, a measure of government digitization
Data
China
Energy
China
While dependent on imports for its electricity production, China benefits from comparatively lower energy prices ($0.09/kwh)
Computing power
China
Able to compete through “good-enough chips” and has pledged ~$40 billion in investment in chipmaking capabilities
Second-largest data center capacity (~20GW), with significant unused capacity
100
90
80
70
60
50
40
30
20
10
0
Relative performance across enablers, on a normalized scale of 0 to 100
Computing power
European Union
Access to cutting-edge chips is supported through ASML, the sole manufacturer of the EUV lithography equipment required for state-of-the-art chipmaking
Third-largest data center ecosystem, at ~8GW in capacity
Energy
European Union
Average energy costs are high, at ~$0.27/kwh, but some member states have considerably lower costs (Sweden at $0.11/kwh, Finland at $0.11/kwh, and Spain at $0.14/kwh)
Intellectual property
European Union
EU-based scholars and institutions have the largest number of citations in AI academic publications since 2019 (528,000)
Fifth in AI patent applications since 2019 (4,000)
10% of notable machine-learning models have been developed by or in partnership with EU scholars and institutions
Talent
European Union
Second-largest global AI talent pool, with ~275,000 AI specialists as of 2024
Third-largest share (10%) of the world’s top AI researchers
Data
European Union
Despite being home to the fourth-largest number of mobile broadband subscriptions (~500 million), GDPR limits access and use
Ranks, as a union, in the top 15% in the E-Government Development Index, a measure of government digitization
Capital power
European Union
GenAI fundraising stands at only $3.5 billion as of June 2024 (compared to $65 billion in the US)
Top tech companies in the EU spent just $41 billion annually on R&D (compared to $203 billion in the US)
100
90
80
70
60
50
40
30
20
10
0
Relative performance across enablers, on a normalized scale of 0 to 100
Computing power
Japan
Key provider of equipment necessary for chipmaking and a leading exporter of arsenic
Third-largest data center ecosystem, with a total capacity of ~2GW in 2024
Energy
Japan
Relies heavily on imports due to scarce domestic resources, leading to comparatively higher electricity prices ($0.19/kwh)
Capital power
Japan
Third-largest government R&D spending ($13 billion per year)
Ranks fourth in R&D spending by domestic tech conglomerates ($26 billion per year)
Ranks sixth in mobile broadband subscriptions, with moderate contextualization of data (i.e., ease of integrating and using data from diverse types and sources)
Ranks in the top 15 of the E-Government Development Index, a measure of government digitization
Data
Japan
Intellectual property
Japan
Only ~1% of machine-learning models have been developed by or in partnership with Japanese scholars and institutions
Sixth-largest AI patent application producer since 2019 (1,600)
Small share of top AI researchers, with ~1% of the top 2,000
Talent
Japan
100
90
80
70
60
50
40
30
20
10
0
Relative performance across enablers, on a normalized scale of 0 to 100
Energy
South Korea
Relies heavily on energy imports, but average electricity prices remain low at $0.13/kwh
Computing power
South Korea
Second-highest exporter of indium
Holds a ~30% market share among manufacturers of sub-10 nm chips
Data center ecosystem is comparatively small, with capacity of ~1GW in 2024 (compared to the EU at ~8GW and the US at ~45GW)
Fourth-largest government R&D spending ($8 billion per year)
Third-largest amount of R&D spending ($28 billion per year) by domestic tech companies
Capital power
South Korea
Only ~2% of notable machine-learning models have been developed by or in partnership with South Korean scholars and institutions
Third-largest producer of patent applications since 2019, primarily driven by private institutions (13,000)
Intellectual property
South Korea
Talent
South Korea
Small share of top AI researchers, with less than 1% of the top 2,000
Data
South Korea
Ranks 23rd in mobile broadband subscriptions, with moderate contextualization of data (i.e., ease of integrating and using data from diverse types and sources)
Ranks fourth in the E-Government Development Index, a measure of government digitization
100
90
80
70
60
50
40
30
20
10
0
Relative performance across enablers, on a normalized scale of 0 to 100
Energy
United Arab Emirates
Comparably low electricity prices ($0.11/kwh); running and training GenAI models is much cheaper in the Gulf than in the US or EU
US recently eased restrictions imposed on chip exports to the UAE, but potential caps on exports of chips have been discussed
Small data center capacity (~0.4GW), but that is expected to increase significantly in the coming years
Computing power
United Arab Emirates
Capital power
United Arab Emirates
Strength lies in the centralization of capital in a few, government-controlled sovereign wealth funds (~$2 trillion in AUM) and strong support for AI investment
Intellectual property
United Arab Emirates
Increased share of AI publications by ~200% (2013–2021 vs. 2022–2023) but still trails other middle powers
Talent
United Arab Emirates
With 6,700 AI specialists, its talent pool remains comparatively small—but it has grown by 36% between 2022 and 2024
Small share of top AI researchers, with less than 1% of the top 2,000
Data
United Arab Emirates
Ranks 46th in number of mobile broadband subscriptions, with moderate contextualization of data (i.e., ease of integrating and using data from diverse types and sources)
Ranks in the top 15 of the E-Government Development Index, a measure of government digitization
100
90
80
70
60
50
40
30
20
10
0
Relative performance across enablers, on a normalized scale of 0 to 100
Capital power
Plans to invest $40 billion in AI development on top of its $100 billion investment in data center expansion and its $100 billion tech fund
Benefits from heavily centralized capital, notably in its large sovereign wealth fund, PIF (AUM of ~$900 billion)
Pool of talent remains comparatively small but has grown by 17% between 2022 and 2024, for a total of 4,500 AI specialists
Talent
Intellectual property
Increased its share of AI publications by ~160% (2013–2021 vs. 2022–2023) but still trails other middle powers
Data
Ranks 27th in number of mobile broadband subscriptions, with moderate contextualization of data (i.e., ease of integrating and using data from diverse types and sources)
Ranks in the top 10 of the E-Government Development Index, a measure of government digitization
Energy
Saudi Arabia
Electricity costs are among the lowest in the world ($0.07/kwh); running and training GenAI models is much cheaper in the Gulf than in the US or EU
Saudi Arabia
Saudi Arabia
Saudi Arabia
Saudi Arabia
Computing power
Saudi Arabia
Exports of US chips to Saudi Arabia are currently restricted; while discussions about easing the restrictions are ongoing, as of November 2024 the restrictions are still in place
Small data center capacity (~0.4GW), but that is expected to increase significantly in the coming years
United States
Intellectual property
Intellectual property
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
Computing power
Select a region:
Data
Saudi Arabia
Saudi Arabia
Saudi Arabia
Saudi Arabia
Saudi Arabia
Relative performance across enablers, on a normalized scale of 0 to 100
South korea
Japan
100
100
Capital power
Computing power
Energy
Data
IP
90
80
70
60
50
40
30
0
Intellectual property
United States
Second-highest number of AI patent applications (18,000) since 2019
US-based tech companies spend the most on R&D investments ($203 billion in 2024)
The US leads on AI venture capital investment between 2019 and 2024 ($303 billion)
US-based tech companies spend the most on R&D investments ($203 billion in 2024)
Capital power
United States
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Intellectual property
United States
Intellectual property
United States
Ranks in the top quintile of the E-Government Development Index, a measure of government digitization
Talent
United States
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
United States
Second-highest number of AI patent applications (18,000) since 2019
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
China
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
China
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
China
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
China
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
China
Computing power
Energy
Data
IP
Talent
Capital power
Computing power
Energy
Data
IP
Talent
Capital power
Computing power
Energy
Data
IP
Talent
Capital power
Computing power
Energy
Data
IP
Talent
Capital power
Computing power
Energy
Data
IP
Talent
Capital power
Computing power
Energy
Data
IP
Talent
Capital power
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
China
Second-highest number of AI patent applications (18,000) since 2019
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Talent
China
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Intellectual property
China
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Data
China
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Energy
China
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
European Union
Second-highest number of AI patent applications (18,000) since 2019
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Talent
European Union
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Intellectual property
European Union
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Data
European Union
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Data
European Union
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
Japan
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
Japan
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
Japan
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
Japan
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
Japan
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
South Korea
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
South Korea
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
South Korea
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
South Korea
Second-highest number of AI patent applications (18,000) since 2019
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
South Korea
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
United Arab Emirates
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
United Arab Emirates
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
United Arab Emirates
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
United Arab Emirates
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
Capital power
United Arab Emirates
Capital power
Saudi Arabia
Capital power
Saudi Arabia
Capital power
Saudi Arabia
Capital power
Saudi Arabia
Capital power
Saudi Arabia
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
US-based scholars and institutions have the second-largest number of citations in AI academic-publications (~385,000) since 2019
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions
70% of notable machine-learning models have been developed by or in partnership with US-based scholars and institutions