India AI push at Davos 2026: Infrastructure, sovereignty and real-world use cases take center stage

India AI Takes Center Stage at Global Forums

India's artificial intelligence journey is entering a decisive phase, moving from theoretical potential to practical deployment. At the World Economic Forum in Davos, industry leaders and policymakers outlined a vision where India AI becomes synonymous with sovereign capability rather than dependency 1

1

. Maharashtra Chief Minister Devendra Fadnavis highlighted how AI is already transforming public administration, with the state embedding AI in governance and service delivery through initiatives like AgriStack, which digitizes land records, crop data, and farmer information 1

1

. The state plans to build a 200-acre innovation city to attract AI-led investments and talent, signaling that AI Infrastructure development has become a strategic priority.

IndiaAI Mission Reframes National AI Strategy

The IndiaAI Mission represents a fundamental shift in how India approaches AI sovereignty, with its ₹10,300 crore allocation focusing on shared national infrastructure rather than private competition 2

2

. By provisioning access to more than 38,000 GPUs at rates 60-80 percent cheaper than global cloud pricing, the government is compressing the cost of experimentation for Indian founders 2

2

.

This democratization of compute addresses a persistent bottleneck where innovation died not from weak ideas but from prohibitive learning costs. The mission also invests in Indian-context data through initiatives like AIKosh and BharatGen, addressing the linguistic and cultural blind spots in global AI models trained primarily on Western datasets 2

2

. So far, 12 Indian firms have been selected to build India-centric Large Language Models (LLMs) and voice AI systems 4

4

.

AI Ecosystem Faces Deployment and Value Creation Challenges

While AI adoption is accelerating, a significant gap persists between deployment and value creation. PwC India chairperson Sanjeev Krishan revealed that only 12% of CEOs report returns on both top and bottom lines from AI use, largely because organizations fail to view it as a tool to transform enterprise fundamentally 1

1

. Bajaj Finserv demonstrates what's possible when AI integrates into real workflows: the company creates thousands of marketing videos annually using AI, customized 300,000 individual ads in 15 days for a Diwali campaign, and processes 30,000-40,000 loans monthly end-to-end through an AI bot that negotiates in Hindi, English, and mixed languages out of 4.5 million total monthly loans 1

1

.

These use cases deliver a 30% improvement in productivity, illustrating how AI creates value when embedded in actual business processes.

Data Centers as National Infrastructure Demand Policy Intervention

India's datacenter capacity is expected to reach 9.2GW by 2030 from the current 960MW, with Gartner projecting datacenter spending to grow at 20.5% in 2026 4

4

. Industry leaders argue that data centers as national infrastructure should receive treatment equivalent to roads and power grids. Manoj Paul of Equinix expects faster fiber deployment, cleared regulations to improve network density, and access to reliable, affordable power for energy-intensive AI workloads, noting that AI server racks consume up to six times more energy than traditional servers 4

4

. Raju Vegesna of Sify Technologies emphasized that India has surplus power capacity and renewable energy growth, but needs policy clarity around data hosting, taxation, and whether India positions itself as a consumption market or AI services country 5

5

.

AI Industrialization Requires Strategic Funding and Skilling Programs

The transition from AI adoption to AI industrialization demands a shift from rapid-exit Venture Capital cycles to patient, institutional-grade capital from Sovereign Wealth Funds and Private Equity 3

3

. The ET AI Impact Forum in February will bring together semiconductor giants and infrastructure operators to accelerate commissioning of hyperscale data centers that serve SMEs as effectively as global conglomerates 3

3

.

On talent, Krishan stressed that India must urgently overhaul its higher education system to remain relevant in an AI-driven economy, while Chakri Gottemukkala of 09 Solutions pushed back against fears of job elimination, arguing that higher-level engineering talent will become increasingly important through re-skilling that creates new value rather than automating old tasks 1

1

5

5

.

Building Bharat-Specific AI Models Over Western Replication

Zerodha cofounder Nikhil Kamath cautioned against replicating Western AI models, urging India to focus on building applications above core models rather than competing where risk capital is abundant 1

1

. Indian users accounted for more than 177 million AI application downloads in 2024, making India the second-largest AI consumer market globally, yet consumer AI revenue remains under $12 million annually 2

2

. This adoption-value gap reveals that scale without embedded utility doesn't compound. Nearly 90% of Indian AI startups funded since 2020 have been application-led, attracting close to $1.8 billion in capital, while AI Infrastructure and foundation model development received a fraction of that investment 2

2

. India operates across 22 official languages and hundreds of dialects with deeply informal economic structures, making models trained in Bharat's complexity better positioned to generalize globally. Financial services startups are already using cash-flow intelligence, GST data, and platform income signals to underwrite thin-file borrowers, while healthcare companies like Qure.ai deploy AI-driven diagnostics at population scale 2

2

. Niti Aayog projects AI adoption to add $1.7 to $2 trillion to India's economy by 2035, with Accenture estimating generative AI alone could contribute an additional $675 billion by 2038 4

4

. With 14% of the world's AI talent, India should invest selectively in Digital Public Infrastructure while focusing on application layers where it can export scale globally 5

5

.