India AI Impact Summit Insights
All sessions

Financing AI Futures: Digital Foundations for Asia-Pacific

Contents

Executive Summary

This panel discussion at the India AI Summit explores how Asia-Pacific countries can build sustainable AI infrastructure through public-private partnerships, digital public infrastructure (DPI), and coordinated regional approaches. The speakers present concrete examples from India, Kazakhstan, Tajikistan, and Uzbekistan, demonstrating that AI readiness requires foundational digital investments, institutional coordination, and realistic financing models rather than technology alone.

Key Takeaways

  1. DPI First, AI Second: Before deploying AI systems, build foundational digital infrastructure—identity, payments, data exchange—as the enabling layer. AI amplifies existing digital ecosystems; it cannot function without them.

  2. Regional Cooperation Over National Duplication: Developing nations should explore shared regional compute infrastructure, demand guarantees, and joint financing rather than building isolated national AI centers. This reduces per-country cost and increases negotiating power with chip suppliers.

  3. Institutional Design Matters More Than Algorithm Choice: Success depends on creating shared governance frameworks, clear data access rules, and cross-ministry coordination mechanisms—the organizational infrastructure matters more than the AI technology selection.

  4. Demand Guarantees Bridge the PPP Gap: Governments securing minimum capacity purchases at affordable rates allow private sector to invest confidently. This model has proven effective in healthcare and is directly applicable to compute infrastructure.

  5. Political Champion + Demonstrable Value = Adoption at Scale: Government-wide AI integration requires both top-level political commitment (making AI a KPI) and quick wins (showing real operational improvements) to overcome institutional resistance and black-box skepticism.

Key Topics Covered

  • Digital Public Infrastructure (DPI) as AI Foundation: Identity systems, payment systems, and data exchange platforms as prerequisites for AI deployment
  • Financing Models for AI Infrastructure: Public-private partnerships (PPPs), blended finance, demand aggregation, and long-term sovereign lending
  • Computing Capacity & GPU Democratization: Scaling compute access through regional cooperation and price discovery rather than monolithic government facilities
  • India's National AI Mission: Seven-pillar approach to building AI-ready infrastructure (compute, foundation models, datasets, talent, governance)
  • Government-Led AI Integration: Practical examples of embedding AI into public service delivery systems
  • Regional Cooperation & Central Asia: Cross-border partnerships to address landlocked countries' infrastructure constraints
  • Data Governance & Institutional Coordination: Creating shared data platforms and coordinating multiple ministries around AI deployment
  • Skills Development & Workforce Readiness: Training technical workforce in tier-2 and tier-3 cities for AI-ready jobs
  • Responsible AI Governance: Technology-enabled approaches to AI regulation without overburdening bureaucracy
  • From AI Pilots to Production Scale: Transitioning experimental projects to actual service delivery integration

Key Points & Insights

  1. DPI is the prerequisite for AI, not the alternative: Countries like India, Kazakhstan, and Uzbekistan demonstrate that robust identity systems, payment platforms, and data exchange infrastructure must exist before AI can operate effectively at scale. Without interconnected data flows, AI deployment remains fragmented and ineffective.

  2. India's GPU expansion model demonstrates demand-side alternatives to monolithic procurement: In less than two years, India scaled from 438 GPUs to 38,000 GPUs through public-private partnerships, demand aggregation, and price discovery—not centralized government procurement alone. This model is replicable for other countries with resource constraints.

  3. Institutional coordination, not just technology, is the hardest implementation challenge: Kazakhstan's success came from creating shared digital infrastructure that operates as national infrastructure, requiring clear governance agreements, data access rules, and responsibilities across ministries—not from AI tools themselves.

  4. Regional cooperation addresses the resource-constraint paradox for developing nations: Tajikistan and Uzbekistan illustrate how landlocked Central Asian countries can build compute capacity partnerships rather than duplicative national infrastructure. A UN resolution on Central Asian AI and infrastructure partnerships demonstrates the political will for this model.

  5. Demand guarantees unlock private sector investment in public infrastructure: Drawing from healthcare models, governments can assure minimum demand for compute capacity at nominal user charges, reducing investment risk for private companies while keeping services affordable for citizens.

  6. Foundation models must be contextualized for linguistic and cultural diversity: India's foundation model pillar explicitly addresses non-English language, social, and cultural contexts. Generic English-trained models underperform on local problems—requiring investment in indigenous models.

  7. High-level political commitment translates institutional resistance into KPI-driven adoption: Tajikistan's success in government AI integration came when the president made digitalization and AI top-level agenda items, converting ministry resistance into departmental performance targets.

  8. Data accessibility and governance frameworks are more important than specific tools or platforms: Kazakhstan emphasizes that the platform choice matters less than the shift from isolated digital services to a "connected operating environment" with clarity on data access, usage conditions, and accountability.

  9. Fiscal constraints require justifying AI as long-term productivity and economic investment, not technology hype: From Uzbekistan's Ministry of Finance perspective, AI infrastructure improvements—better tax compliance, fraud reduction, efficient expenditure—directly improve fiscal outcomes and competitiveness, justifying investment during tight budgets.

  10. Demonstrable operational value is the turning point for ministry buy-in: Kazakhstan's citizen service platform (egov.kz) conversion to AI-based request routing and processing convinced skeptical ministries through visible improvements—faster processing, reduced manual workload—rather than theoretical benefits.

Notable Quotes or Statements

"Computing power but also the algorithm is going to define the level of success." — Antonio (opening remarks, Asian Development Bank)

"Democratize both access to artificial intelligence as well as its benefits... every innovator in the country has an opportunity to create a competitive AI solution as well as every citizen to be able to benefit from that AI solution." — Kushal (India AI Mission)

"We went from 438 GPUs to about 38,000 GPUs... it's not 38,000 GPUs that government of India has procured all on its own money it's through public private partnership, demand aggregation, price discovery." — Kushal, on India's compute scaling strategy

"Becoming AI ready is really about long-term investment in the system and not only investment in the AI itself." — Adana (Kazakhstan)

"AI do not work effectively in a fragmented information... one of the most important shift for us was building structured data exchange." — Adana (Kazakhstan)

"When you deploy AI models of course you are going to lose it [decision-making control]... so you need to explain why [this is beneficial]." — Faruse John (Tajikistan)

"Investments in AI ready digital infrastructure must be viewed as long-term economic and productivity investments... these investments improve fiscal outcomes, better tax compliance, reduce fraud." — Aziz (Uzbekistan Ministry of Finance)

"It is not the specific platform and it's not the specific technology... it is the shift in how the system is organized." — Adana (Kazakhstan)

"The government guarantee was never called because demand was always more than the capacity but the safety net for the private sector to make the initial investment is important." — M Chri (ADB), on demand guarantees in healthcare PPPs

Speakers & Organizations Mentioned

Government & Institutional Representatives:

  • Antonio — Asian Development Bank (opening moderator, digital transformation strategy)
  • Kushal — India Mission / Government of India (AI strategy and national program implementation)
  • Carolyn Flory — Asian Development Bank, Digital Development Specialist (panel moderator)
  • M Chri (Mike) — Asian Development Bank, Director for Infrastructure Financing, Private Sector (South Central and West Asia)
  • Faruse John — Tajikistan, Deputy Chair, AI Council under Minister of Industry and New Technologies
  • Aziz John Akramov — Uzbekistan Ministry of Finance / Ministry of Economy
  • Adana — Kazakhstan (digital government and AI integration official)

Organizations & Initiatives:

  • Asian Development Bank (ADB) — multilateral development bank; strategic partner in supporting government AI decisions
  • India Mission (National Program on Artificial Intelligence) — 5-year, $1.2 billion national AI program
  • Digital India Corporation — government entity managing India's digital public infrastructure
  • Tajikistan AI Council — government coordination body
  • Kazakhstan's egov.kz — citizen service platform integrating AI
  • Saudi Data World Company — implementing 12-megawatt AI data center in Uzbekistan (completion 2026)
  • Global Tech Giants mentioned: Google, Microsoft, Amazon (partnerships with Uzbekistan data center)
  • UN Resolution on Central Asian Partnership — passed by all Central Asian countries supporting AI and infrastructure partnerships

Technical Concepts & Resources

India's National AI Mission — Seven Pillars:

  1. India Compute Capacity — Scaling GPU access from 438 to 38,000 GPUs through PPPs and demand aggregation
  2. Foundation Models — Indigenous, contextualized AI models for linguistic and cultural diversity (not English-only)
  3. AI Kosh ("Repository")** — Centralized access layer for datasets, models, development tools, and compute (similar to Hugging Face model)
  4. Application Development Initiative — Focus on "AI for good" use cases and end-user solutions
  5. Future Skills Pillar — Training technical workforce in tier-2/tier-3 cities; transforming industrial training institutions into AI/data labs
  6. Safe and Trusted AI — Technology-enabled governance tools and benchmarks to regulate AI without burdening bureaucracy
  7. Startup Financing — Patient capital for AI startups with limited immediate commercial viability

India's Digital Public Infrastructure (DPI) Scale:

  • UPI (Unified Payments Interface): 21.7 billion monthly transactions; ₹290 billion monthly transaction value
  • DigiLocker: 660 million users; 9.5 billion verified documents issued
  • Aadhaar (Digital Identity): 99.5% population coverage; direct benefit transfers generating $42 billion in savings

India Specific Use Case — Cybercrime Database Hackathon:

  • Created synthetic dummy version of sensitive National Cyber Crime Database
  • Opened hackathon to all innovators without compromising real data security
  • Shortlisted 20 innovators to on-premise secured environment
  • Ministry of Home Affairs now evaluating solutions for potential deployment

Kazakhstan's Digital Public Infrastructure:

  • Smart Bridge — National data exchange platform
  • Smart Data Commit — National data lake for government agency data sharing
  • e.gov.kz / eWitness — Citizen service platform processing large request volumes with AI-based classification and routing
  • Digital Government Infrastructure — Shared identity, integration layers, service platforms across institutions

Uzbekistan's Infrastructure Projects:

  • Data World Project — 12-megawatt AI-first data center in IT Park Tashkent ($150 million, completion 2026)
  • LED Gold Certification — International green data center standards compliance
  • Semiconductor & Electronics Development Strategy — Government-initiated sector development

Tajikistan's AI Applications:

  • Credit scoring in finance sector
  • License issuance for businesses (acceleration from committee-based to instant AI decision)
  • AI academy training (first cohort: 50 students sent to banks; evolved to startup incubation model)
  • Export of AI products to 25 countries

PPP & Financing Models Referenced:

  • Demand Guarantee Model — Government assures minimum service capacity purchase at nominal rates (proven in Indian healthcare dialysis scheme)
  • Blended Finance — Combination of public investment (sovereign + security-critical layers), PPPs, and development finance
  • Long-term Sovereign Lending — MDB support for long-duration, low-return infrastructure (unavailable in commercial markets)
  • Demand Aggregation — Regional cooperation to aggregate purchasing power and negotiate chip/infrastructure pricing

Policy/Governance Frameworks:

  • Technology-Enabled AI Governance — Tools and benchmarks that use technology itself to govern AI, reducing bureaucratic burden
  • Data Access Rules & Accountability Frameworks — Clarity on who accesses data, conditions of use, and responsibilities
  • KPI Integration — Making AI implementation a performance target for government departments

Note: This transcript appears to be automatically transcribed with some diction artifacts and repetitions. Where clarity was ambiguous, I have interpreted meaning from context but flagged uncertainty where appropriate.