The Best Superintelligence in Sovereign AI

There is currently no verified artificial superintelligence (ASI) that can be identified as the world's best sovereign AI system. No country has demonstrated a machine intelligence that meets an established definition of artificial superintelligence.

However, countries are competing to build something strategically important: sovereign AI capabilities that they can develop, operate, govern, and control themselves.

According to the Sovereign AI Power Index (SAPI), Quarter 2, 2026, South Korea ranks first in overall sovereign AI power, followed by the United Arab Emirates (UAE), Singapore, and the United States.

Source: The Sovereign AI Power Index, Quarter 2, 2026, published July 2, 2026. Scores are the Index's composite assessments, not percentages of AI intelligence or model performance.

The United States leads the index in compute capacity, scoring 91.0, but ranks fourth overall. South Korea leads because the index evaluates a broader combination of national capabilities, including capital, regulation, data sovereignty, and the ability to turn AI infrastructure into operational intelligence.

The central finding is that sovereign AI leadership is not determined by who owns the largest data center or develops the most powerful individual model. It depends on who controls and can sustain the entire intelligence ecosystem.

Explore the Sovereign AI Power Index

What is sovereign AI?#

Sovereign AI is the ability of a country or organization to independently develop, deploy, operate, and govern artificial intelligence using infrastructure, data, models, talent, and operating systems that it can control.

It is about more than where data is stored. A country may host its data within national borders while relying on foreign-owned models, external cloud infrastructure, proprietary software, and overseas technology providers.

That arrangement can provide data residency without providing full AI sovereignty.

True sovereign AI requires meaningful control across the AI lifecycle: from energy and computing infrastructure to data acquisition, model development, intellectual property, deployment, governance, and decision-making.

McKinsey describes sovereign AI through territorial, operational, technological, and legal dimensions. It also emphasizes that sovereignty exists on a spectrum rather than as a simple yes-or-no condition.

Source: McKinsey — What is sovereign AI?

The five dimensions of sovereign AI power#

The Sovereign AI Power Index evaluates national capability through five dimensions. Together, they assess whether a country can turn AI ambition into durable strategic capacity.

1. Compute capacity#

Compute capacity is the physical foundation of advanced AI.

It includes access to high-performance GPUs and accelerators, data centers, energy supply, grid capacity, and the infrastructure required to train and operate advanced models.

A country may develop excellent AI research but remain dependent on foreign infrastructure if it cannot reliably access the computing resources required to operate its systems.

Compute sovereignty means having reliable access to the infrastructure that makes independent AI development and deployment possible.

2. Capital formation#

Developing sovereign AI requires sustained investment.

Capital formation measures the ability to direct public and private financial resources toward AI research, infrastructure, companies, and strategic national priorities.

This includes sovereign wealth investment, venture capital, government funding, institutional investment, and the financial mechanisms needed to support long-term AI development.

The objective is not simply to spend more money. It is to convert investment into lasting national capabilities, technology ownership, and productive AI ecosystems.

3. Regulatory readiness#

AI systems operate within legal, regulatory, and institutional environments.

Regulatory readiness evaluates whether a country has the legal clarity, procurement frameworks, licensing pathways, and national AI policies needed to develop and deploy AI effectively.

A strong regulatory environment can encourage investment, support responsible innovation, establish security requirements, and enable government and enterprise adoption.

4. Data sovereignty#

Data is one of the fundamental resources behind artificial intelligence.

Data sovereignty concerns who controls strategic datasets, where they are stored and processed, which jurisdiction governs them, and whether sensitive information can be protected from unauthorized access.

For governments and organizations, this includes national datasets, public-sector information, industrial knowledge, healthcare records, financial information, research data, and proprietary business intelligence.

Without meaningful control over data, a country may remain dependent on external AI ecosystems even when it operates its own infrastructure.

5. Directed intelligence#

Directed Intelligence is the Index's proprietary dimension. It evaluates how effectively a country or organization converts AI capability into measurable operational outcomes.

The distinction matters: possessing computing infrastructure, capital, data, and AI models does not automatically produce strategic advantage.

AI must be integrated into institutions, industries, government operations, and national priorities. It must support real decisions, deliver measurable results, and become part of durable operating systems.

The Index describes five stages of institutional adoption:

• Experimental: AI activity is fragmented and largely opportunistic.
• Emergent: Pilot programs exist, but national coordination remains inconsistent.
• Structured: Institutions have established mandates, funding paths, and operating models.
• Directed: AI capabilities are aligned with national priorities and supported by durable governance.
• Intelligence Fabric: AI is embedded across state operations as a coordinated national system.

Source: The Sovereign AI Power Index methodology

The 10 leading countries in sovereign AI#

The following ranking reflects the published SAPI Quarter 2, 2026 composite scores. It is a ranking of national sovereign AI ecosystems, not a ranking of individual AI models or a measurement of artificial superintelligence.

1. South Korea — The leading sovereign AI ecosystem#

SAPI score: 59.2 | Rank: 1

South Korea ranks first in the Sovereign AI Power Index, with a composite score of 59.2.

Its position reflects a coordinated approach to domestic AI development, industrial capability, national investment, and institutional adoption.

South Korea is pursuing sovereign foundation models through a government-backed program involving domestic technology companies and research organizations. The participating ecosystem includes SK Telecom, LG AI Research, Upstage, and other national AI developers.

The country's approach is particularly important because it seeks to reduce dependence on foreign foundation models by developing models and capabilities within its own technology ecosystem.

The Ministry of Science and ICT has defined its sovereign foundation model ambition around original model architecture, domestic pretraining, data acquisition and processing, and independently developed learning methods. This is a more demanding definition than simply fine-tuning a foreign model.

In the 2026 second-stage evaluation of the national project, SK Telecom ranked first, followed by Upstage and LG AI Research. These results are separate from the SAPI national composite score, but demonstrate the depth of South Korea's domestic model-development competition.

Why South Korea leads:

• A coordinated national sovereign AI program.
• Domestic foundation-model development.
• Strong technology and semiconductor industries.
• Collaboration between government, research institutions, and private companies.
• A focus on reducing dependence on foreign AI technology.

Assessment: South Korea is the current overall leader in the SAPI ranking and one of the strongest examples of a country building a coordinated domestic AI ecosystem.

Sources: South Korea's Ministry of Science and ICT | Yonhap — Sovereign AI project evaluation

2. United Arab Emirates — A strategic sovereign AI powerhouse#

SAPI score: 58.9 | Rank: 2

The United Arab Emirates ranks second, just 0.3 points behind South Korea.

The UAE has positioned AI as a strategic national priority, combining investment, computing infrastructure, research, and government-led technology development.

One of its most visible AI initiatives is the Falcon family of models, developed by the Technology Innovation Institute (TII) in Abu Dhabi.

Falcon demonstrates the UAE's ambition to build AI technology locally rather than rely exclusively on models developed by foreign technology companies.

The Falcon ecosystem includes work on language models, Arabic-language capabilities, and efficient model architectures. In May 2025, TII announced Falcon Arabic and Falcon-H1, extending the family's focus on regional language capabilities and model efficiency.

The UAE's broader sovereign AI ecosystem also includes infrastructure providers, government initiatives, research institutions, and companies working on enterprise and public-sector AI.

Why the UAE ranks highly:

• Strategic national investment in AI.
• Government-supported AI research and development.
• Domestic AI model development through TII.
• Growing AI infrastructure and data-center capabilities.
• A strong focus on Arabic-language and regional AI applications.

Assessment: The UAE is one of the strongest examples of a country using national investment and institutional coordination to build sovereign AI capabilities.

Sources: Technology Innovation Institute | Falcon models

3. Singapore — Sovereign AI through institutional strength#

SAPI score: 58.2 | Rank: 3

Singapore ranks third, demonstrating that sovereign AI leadership does not depend exclusively on building the largest domestic foundation model.

Its strategy emphasizes national coordination, AI infrastructure, research, talent, governance, and practical adoption.

Singapore's National AI Strategy 2.0, launched in December 2023, established a national framework for developing and deploying AI. In May 2026, the government announced refreshed priorities and the establishment of a National AI Council to provide strategic direction.

Singapore's position illustrates the importance of institutional readiness: a country must be able to deploy AI throughout its economy, establish effective governance, and develop the talent and infrastructure required to sustain its capabilities.

Why Singapore ranks highly:

• Coordinated national AI strategy.
• Strong regulatory and institutional environment.
• Advanced digital infrastructure.
• Investment in AI talent and research.
• Focus on practical adoption across government and industry.

Assessment: Singapore is a leading example of institutional and governance-driven sovereign AI development.

Source: Singapore Ministry of Digital Development and Information — National AI Strategy update, May 2026

4. United States — The global compute leader#

SAPI score: 57.5 | Rank: 4

The United States has one of the world's most powerful AI ecosystems, with major frontier-model developers, advanced semiconductor technology, cloud infrastructure, research institutions, and private investment.

Its ecosystem includes companies such as OpenAI, Anthropic, Meta, Google, Microsoft, and leading semiconductor and infrastructure providers.

The United States receives the highest compute capacity score in the SAPI ranking: 91.0.

However, its overall position is fourth.

This difference is central to understanding sovereign AI. A country can dominate one dimension without achieving equally strong results across every dimension required for durable national capability.

The Index highlights a significant imbalance between US compute capacity and capital formation. Its published capital formation score is 35.7.

Why the United States remains a major AI power:

• Leading access to advanced computing infrastructure.
• Major frontier AI research and development.
• Globally influential AI companies.
• Strong semiconductor and cloud ecosystems.
• Significant private-sector innovation.

Assessment: The United States remains a global frontier AI powerhouse and the SAPI compute leader, but the Index's broader methodology places it fourth in overall sovereign AI capability.

5. Estonia — Digital government and institutional capability#

SAPI score: 56.6 | Rank: 5

Estonia ranks fifth, demonstrating the importance of digital government, institutional coordination, and national technology infrastructure.

Its established digital-government ecosystem provides a foundation for integrating AI into public services and national operations.

Estonia's position is a reminder that sovereign AI is not exclusively a contest between the largest economies. Smaller countries can build meaningful strategic capabilities through effective institutions, digital infrastructure, and focused national priorities.

Assessment: Estonia is an important example of how digital-government maturity can contribute to sovereign AI development.

6. France — Building European AI independence#

SAPI score: 55.6 | Rank: 6

France is an important player in Europe's effort to reduce dependence on external AI providers.

Its sovereign AI ambitions involve domestic research, European technology partnerships, national infrastructure, and public-sector adoption.

The country's broader strategic challenge is to translate its research and industrial strengths into a competitive AI ecosystem that can operate with greater technological independence.

Assessment: France represents an important European pathway toward strategic AI independence.

7. Saudi Arabia — Investment-led sovereign AI development#

SAPI score: 54.2 | Rank: 7

Saudi Arabia is investing in AI as part of its wider economic and technological transformation.

Its strategy places significant emphasis on infrastructure, capital, national development priorities, and the use of AI across government and industry.

Its position in the SAPI ranking highlights the importance of converting national investment into durable technological and institutional capability.

Assessment: Saudi Arabia is an emerging sovereign AI power with a significant investment-led development strategy.

8. Oman — An emerging sovereign AI ecosystem#

SAPI score: 52.7 | Rank: 8

Oman ranks eighth in the SAPI assessment.

Its position places it among the leading countries in the index's Developing tier, reflecting emerging national capability across the dimensions assessed.

The ranking should be interpreted as a measure of the Index's composite assessment, not as evidence that Oman has developed a frontier or superintelligent AI model.

Assessment: Oman is an emerging participant in the regional sovereign AI landscape.

9. Japan — Advanced technology and industrial foundations#

SAPI score: 50.6 | Rank: 9

Japan brings substantial strengths in industrial technology, robotics, advanced manufacturing, research, and computing.

These capabilities provide a foundation for sovereign AI development, particularly in industrial and specialized applications.

Its strategic opportunity is to combine its established technology base with domestic AI models, infrastructure, and broader deployment.

Assessment: Japan has significant industrial and research foundations that can support the development of sovereign AI capabilities.

10. Italy — Developing European sovereign AI capacity#

SAPI score: 48.8 | Rank: 10

Italy completes the top ten in the SAPI Quarter 2, 2026 ranking.

Its position reflects the country's developing role in the European AI ecosystem and the broader effort to strengthen domestic technological capabilities.

Assessment: Italy is part of Europe's growing sovereign AI landscape, with further opportunities to strengthen infrastructure, investment, and domestic AI development.

What is the difference between sovereign AI and artificial superintelligence?#

The terms are related to the future of artificial intelligence, but they describe different things.

Sovereign AI concerns control: who develops, owns, operates, governs, and can independently maintain AI systems.

Artificial general intelligence (AGI) refers to AI with broad capabilities across a wide range of cognitive tasks, rather than competence limited to a narrow application.

Artificial superintelligence (ASI) refers to a hypothetical form of artificial intelligence that substantially exceeds human cognitive capabilities across most or all relevant domains.

A sovereign AI system could use an advanced foundation model without being AGI or ASI. Likewise, a future superintelligent system could be developed by a private company or operate across international infrastructure without being sovereign to any one country.

The distinction is important because the SAPI ranking does not measure which country has developed the most intelligent model. It measures the national capabilities and institutional conditions that support independent AI development and deployment.

A country could become a sovereign AI leader without developing the world's most powerful frontier model. It could also develop an exceptionally capable model while remaining dependent on foreign computing infrastructure, software, or data.

The long-term strategic question is therefore not only who develops the most intelligent AI, but who retains meaningful control over the intelligence systems that shape its economy, institutions, and national security.

What does a country need to own to achieve true AI sovereignty?#

AI sovereignty extends beyond owning servers, purchasing GPUs, or hosting data within national borders.

It requires control over the knowledge, technology, intellectual property, and decision-making processes that create and operate intelligent systems.

The most important assets include:

Process knowledge#

The practical knowledge of how AI systems are designed, trained, evaluated, deployed, maintained, and improved.

Intellectual property#

Ownership or enforceable control over the inventions, architectures, algorithms, software, research, and technologies that create strategic AI advantages.

Data#

Control over the collection, ownership, processing, storage, access, and governance of strategic national and organizational datasets.

Core product development#

The ability to design, build, improve, and maintain AI products without depending entirely on external technology providers.

Discovery and proprietary AI#

The capacity to conduct independent research, discover new methods, develop proprietary models, and create original AI capabilities.

Private code base#

Control over the software, source code, infrastructure tooling, and systems that enable AI development and operation.

Decision logic#

Understanding and controlling how AI systems process information, make recommendations, execute instructions, and influence important decisions.

Engineering practices#

The technical expertise, development processes, testing systems, security practices, and operational knowledge required to maintain advanced AI systems.

Invention#

The ability to create original technologies, architectures, methods, and applications rather than merely adopting or adapting technologies developed elsewhere.

Quality standards#

The ability to define, measure, verify, and enforce the performance, safety, reliability, and security requirements of AI systems.

Design trade-offs#

The knowledge required to make independent decisions about performance, cost, latency, privacy, security, reliability, energy consumption, and model capability.

The underlying principle is straightforward: sovereignty means retaining the ability to create, understand, control, improve, and operate the intelligence systems on which an organization or nation depends.

A country that owns its data but cannot control its models, software, or infrastructure has only partial sovereignty.

A country that owns its infrastructure but depends on external technology for its core intelligence capabilities also has only partial sovereignty.

The strongest sovereign AI ecosystems retain control over the complete chain of technological and institutional value.

Can open-source AI models provide sovereign AI?#

Yes, but open-source or open-weight models do not automatically guarantee sovereignty.

Using an externally developed model can provide a country with access to advanced AI capabilities. It may also allow local organizations to fine-tune, customize, and deploy models on domestic infrastructure.

However, the original model architecture, training process, weights, licensing terms, data, and development decisions may remain under external control.

The level of sovereignty depends on what can actually be controlled.

A useful distinction is:

South Korea's government-backed foundation-model program illustrates a more demanding approach. Its stated definition of a sovereign foundation model emphasizes domestic model design and pretraining rather than simply fine-tuning a foreign model.

That does not mean every country must independently manufacture every semiconductor or develop every software component. Sovereignty is a spectrum, and strategic independence can be achieved through carefully managed partnerships, trusted suppliers, and diversified infrastructure.

The objective is to avoid critical dependencies that another organization or country can unilaterally control.

Why sovereign AI matters for the future of artificial superintelligence#

As AI systems become more capable, control over their infrastructure, knowledge, and operating environments becomes increasingly important.

The development of more advanced AI could affect national security, economic competitiveness, scientific discovery, public administration, and the distribution of technological power.

Countries that depend entirely on external AI providers may face strategic risks involving service availability, foreign legal jurisdiction, technology restrictions, data access, and the ability to adapt systems to national requirements.

Sovereign AI offers a way to reduce these dependencies and build domestic capabilities.

However, sovereignty does not automatically make an AI system safe, aligned, transparent, or beneficial.

A nationally controlled AI system can still produce unreliable outputs, introduce bias, expose sensitive information, or create security risks. Strong governance, independent evaluation, accountability, and safety research remain essential.

The future of advanced AI will therefore involve two connected but distinct challenges:

1. Intelligence capability: Developing increasingly capable AI systems, potentially progressing toward AGI and ASI.
2. Intelligence sovereignty: Retaining meaningful control over the infrastructure, data, models, intellectual property, and governance that support those systems.

The countries that combine both capabilities may have a substantial strategic advantage.

Final verdict: The future belongs to those who control intelligence#

The competition for sovereign AI is not simply a race to build the largest data centers, train the biggest models, or attract the most investment.

It is a competition to control the complete intelligence ecosystem: the infrastructure, data, models, research, intellectual property, software, engineering expertise, and institutions that make advanced AI possible.

South Korea currently leads the Sovereign AI Power Index, followed closely by the UAE and Singapore. The United States remains the strongest compute-capacity leader but ranks fourth in the overall assessment.

These results demonstrate that national AI power is multidimensional. Infrastructure alone is not enough. Capital alone is not enough. A powerful model alone is not enough.

The most strategically powerful sovereign AI ecosystem will be the one that can independently create, understand, secure, govern, and continuously improve its own intelligence capabilities.

There is no established sovereign artificial superintelligence champion today. But the countries building the foundations of independent intelligence are already competing to shape who controls the next era of AI.