If your global AI strategy map treats national AI policy as primarily an OECD phenomenon (concentrated in the US, EU member states, UK, Canada, Japan, South Korea), the 2024-2025 adoption data shows the geographic centre of gravity is shifting. More than half of newly adopted national AI strategies in 2024 came from emerging economies. New frameworks have surfaced across sub-Saharan Africa (Ethiopia, Ghana, Nigeria), South and Central Asia (Sri Lanka, Nepal), and Latin America and the Caribbean (Costa Rica, Jamaica). Mexico and South Africa have strategies in active development. The OECD-centric framing of national AI policy is now outdated by the data.

Countries with a national AI strategy, 2019–25:

The trajectory data shows the geographic shift clearly. In 2019, the map of countries with national AI strategies was concentrated in OECD economies plus a handful of strategic outliers (China, UAE, Saudi Arabia, India). By 2025, the map has filled in across sub-Saharan Africa, Central Asia, Southeast Asia, and Latin America. The countries that were "no strategy" five years ago are now substantially the countries with "in development" or "adopted" frameworks.

The driver pattern: countries that historically played smaller roles in AI policymaking are now adopting formal national strategies as the perceived stakes of AI competitiveness have risen. Three structural drivers are visible.

The first driver: international cooperation and technical assistance are diffusing AI policy frameworks. Multilateral organisations (UN, OECD, regional bodies like the African Union) have provided template strategies, technical advisors, and convening forums that lower the cost of strategy adoption for countries without prior AI policy infrastructure. The UN's 2025 launch of the Independent International Scientific Panel on AI and Global Dialogue on AI Governance, announced at the UN General Assembly in August 2025, formalises the international coordination mechanism. The African Union's April 2025 declaration of AI as a strategic priority signals regional alignment.

The second driver: domestic recognition of AI as a state-capacity lever. Emerging-economy governments increasingly view AI as a tool for state capacity (digital government services, agriculture optimisation, healthcare delivery, education infrastructure) rather than primarily as a frontier-technology challenge. The policy framing in countries like Rwanda, Vietnam, and Indonesia treats AI as infrastructure for state development. This framing produces different policy content than the frontier-risk framing dominant in OECD economies, and the policy adoption follows a different urgency curve.

The third driver: peer-country adoption pressure. Once neighbouring or peer countries adopt national AI strategies, the relative-competitiveness pressure on non-adopting countries rises. A country without a national AI strategy in 2025 is now a visible outlier in many regional groupings. The pattern is similar to the 2010-2015 surge in national broadband strategies: once peer countries adopted, the cost of not adopting (perceived loss of regional positioning) rose.

The trajectory through 2026-2028 suggests continued expansion. Strategies in active development in Mexico, South Africa, and several other countries will be published in 2026-2027. The remaining "no strategy" countries, concentrated in parts of Africa and Central Asia, will be subject to the same adoption pressures over the next 24 months. By 2028, the map of countries with national AI strategies will likely cover the substantial majority of UN member states.

Three structural implications follow for organisations operating globally.

The first implication: the geographic AI policy landscape is broader than the planning view most organisations use. Strategic plans that engage with US, EU, UK, China, Japan, and a handful of other jurisdictions are working against a 2020-era map. The 2026 map includes meaningful AI policy activity in 100+ countries, with varying levels of operational maturity. Organisations operating in emerging markets need to engage with the national AI strategy of those markets as a planning variable.

The second implication: the content of emerging-economy AI strategies differs from OECD AI strategies in important ways. OECD strategies typically focus on frontier model regulation, safety, and competitiveness. Emerging-economy strategies typically focus on state capacity, sector applications (agriculture, healthcare, education), and digital infrastructure. The compliance and engagement requirements differ accordingly. Organisations operating in emerging markets need to understand the specific framework of each jurisdiction rather than apply OECD templates.

The third implication: the implementation gap is the operational risk for emerging-economy AI strategies. The Oxford Insights dataset captures published strategies, not implementation progress. Many emerging-economy strategies are published but not fully resourced or operationalised. The 2026-2028 trajectory will reveal which strategies translate into binding regulation, investment programmes, and procurement preferences, and which remain primarily aspirational. Organisations need to track implementation progress separately from policy adoption.

Looking ahead: the national AI strategy landscape will continue to expand geographically through 2028. The implementation maturity will vary widely across countries. Three rough tiers will emerge:

  • Tier 1: countries with binding regulation, funded implementation, and operational AI procurement/oversight (current examples: US, EU member states, UK, China, Japan, South Korea, Singapore, UAE, Canada, Australia)
  • Tier 2: countries with published strategies and partial implementation (current examples: India, Brazil, Saudi Arabia, Indonesia, several emerging economies with active programmes)
  • Tier 3: countries with strategies but limited implementation (current examples: many sub-Saharan African and Central Asian countries with recent adoptions)
For organisations operating across multiple jurisdictions, the strategic anchor needs to engage with the tier distinction rather than treat all countries with strategies as operationally equivalent. The compliance and engagement load varies materially by tier. The opportunity for early influence on policy formation is highest in Tier 3 countries; the compliance load is highest in Tier 1 countries. The middle tier is where most operational complexity sits.

By 2028, the national AI strategy map will be substantially complete geographically. The implementation gap will be the primary differentiator across countries. Organisations that build strategic engagement infrastructure that engages with the actual implementation reality (rather than the strategy document) will be better positioned than organisations that rely on strategy-document signals alone.

Sources

  • Primary: Stanford AI Index 2026, Chapter 8 (Policy and Governance) 8.2 — hai.stanford.edu/ai-index/2026
  • National AI strategy tracking: Oxford Insights, 2024 — country-level AI strategy adoption database
  • 2025 geographic updates: AI Index 2026 — newly adopted strategies (Ethiopia, Ghana, Nigeria, Sri Lanka, Nepal, Costa Rica, Jamaica)
  • International coordination: UN Independent International Scientific Panel on AI; UN Global Dialogue on AI Governance (Aug 2025); African Union AI declaration (April 2025)