India wants to play in the AI top tier, and is making it clear that its strategy isn’t just about “having models,” but about building infrastructure, connectivity, and industrial capacity at a national scale. That idea has been hovering over the India AI Impact Summit 2026 in New Delhi, where the government has emphasized a recurring message: if AI is going to transform the economy, it cannot become a new factor of inequality.
Within this context, Google has taken the opportunity to reinforce its narrative of “useful and accessible AI” with a package of initiatives focused on infrastructure and the public sector. In a post signed by James Manyika, the company recalls its recent announcement of a $15 billion investment to establish foundational AI infrastructure in India and also presents America-India Connect, an initiative to open new strategic fiber optic routes that increase reach, reliability, and resilience of digital connectivity between the United States, India, and various points in the Southern Hemisphere.
This move is significant: in 2026, talking about AI at a national scale means talking about computing capacity and networks. And India is trying to solve two bottlenecks simultaneously: data centers (energy, land, permits, supply chains) and connectivity (routes, redundancy, and latency).
Infrastructure, State, and Talent: Google’s “package”
Beyond cables and connectivity, Google focuses on the public sector and the “human factor.” Among the announcements is a $30 million program via Google.org to promote AI innovation projects in governments, another fund of $30 million for science, and agreements aimed at strengthening institutional capabilities (including training and deployments on public platforms).
The subtext is clear: mass adoption isn’t just about technology; it’s about processes, governance, and skills, especially in administrations managing critical services and large volumes of data.
NVIDIA and the muscle of inference era: 20,000 Blackwell Ultra for Yotta
If Google emphasizes connectivity and ecosystem capabilities, NVIDIA is pushing the message of computing as an immediate leverage. Parallel to the summit, Reuters reports that Yotta Data Services will invest over $2 billion to create an AI computing hub in India with NVIDIA’s latest chips and plans to deploy more than 20,000 Blackwell Ultra chips by August.
The most revealing detail is the distribution: according to Reuters, NVIDIA itself will use half of these chips over four years for its DGX AI Cloud service, used by major Indian IT firms such as TCS and Infosys.
In other words: India isn’t just buying hardware; it’s entering a phase where part of the computing is organized around AI clouds and dedicated clusters, which is crucial at this stage, where inference and agents are driving demand for capacity.
Reuters adds another important point: India, still behind the US and China in AI technology development, is positioning itself as an investment destination for data centers, a stance also fueled by global reconfiguration of supply chains due to export controls and geopolitical tensions.
Macron also looks to India: alliances and AI geopolitics
AI is no longer just a technological race but a race of blocks. In Europe, there’s an increasing focus on sovereignty and strategic autonomy, and France is trying to play a visible role. In this line, European media have reflected Emmanuel Macron’s interest in approaching India’s AI ambitions and building cooperation bridges.
For India, these kinds of international support fit a carefully crafted narrative: the country aims to be a software factory, a data center hub, and a mass market all at once; a triangle that, if well executed, attracts investment, talent, and providers.
What do all these implications mean (beyond headlines)
- “AI for All” demands real infrastructure: Without electricity, space, networks, and supply chains, announcements remain demos. Fiber projects and GPU hubs may be less glamorous, but they are essential.
- The battle shifts towards inference: Training models matters, but business (and impact) are moving towards deploying AI in production, at scale, with controlled costs.
- Sovereignty is more than data: It also includes computing capacity, operational control, local talent, and public-private collaboration frameworks that prevent complete dependency.
- Risk of concentration: If access to computing is limited to a few, a country can achieve scale… but may lose diversity and innovation at the edges. The challenge will be balancing “gigantism” with a competitive ecosystem.
Frequently Asked Questions
What is America-India Connect and why does it matter for AI?
It’s a Google initiative to promote new strategic fiber optic routes between the US, India, and other locations in the Southern Hemisphere, enhancing resilience and reach of connectivity—critical for cloud services and AI workloads.
Why is it significant that Yotta deploys 20,000 Blackwell Ultra chips?
Because a deployment of that scale targets inference and large-scale computing capacity, useful for AI in production, training, and “AI cloud” platforms within the country.
What does “AI for All” mean practically?
Access: reliable connectivity, public services capable of adopting AI with guarantees, workforce and business training, and affordable computing resources.
What role do international alliances (Google, NVIDIA, France) play in India’s strategy?
They provide investment, technology, and geopolitical positioning, but also require India to clearly define its sovereignty conditions, capacity transfer, and local ecosystem development.
Image via Twitter