The rise of AI is putting immense strain on energy systems worldwide. As India emerges as a key player in AI, it must find smart and sustainable ways to meet this growing demand. 

Ann Dunkin, former chief information officer at the US energy department, spoke with AIM on the sidelines of the Invest Karnataka 2025 summit, sharing interesting insights into how India can balance AI growth with sustainable energy solutions.

A Multi-Faceted Energy Strategy

India’s AI future largely depends on a robust and sustainable energy framework. Dunkin believes that the right approach is to leverage all available renewable resources. “Look at India’s assets and where the country can monetise these to get clean power at the lowest cost.”

She spoke about the “All-of-the-above energy policy” that revolves around wind, geothermal, solar, clean hydrogen, and nuclear power. 

Interestingly, the term gained prominence during the 2012 US presidential election, particularly with President Barack Obama’s administration using it to balance fossil fuel development with investments in renewable energy.

While the political shift in the US has lately impacted climate policies, Dunkin remains optimistic about clean energy’s momentum. “Even if our president chooses not to, the market will drive clean energy forward.”

She added that, with abundant sunlight and significant wind energy potential, India is well-positioned to tap into renewable resources to support the power-hungry AI infrastructure.

According to the Economic Survey 2024-2025, India’s data centre market is expected to grow from $4.5 billion in 2023 to $11.6 billion by 2032 at a compound annual growth rate of 10.98%.

With coal generating nearly 75% of India’s electricity, the data centre industry’s reliance on fossil fuels raises sustainability challenges. However, businesses are increasingly turning to renewable energy sources such as solar and wind power. 

Hiranandani Group’s data centre venture, Yotta, plans to secure more than 80% of its energy from renewable sources within the next three to five years. Likewise, Hyderabad-based CtrlS is targeting full reliance on renewable power by 2030.

Meanwhile,  Reliance Group, led by billionaire Mukesh Ambani, is purchasing AI GPUs from NVIDIA to build a new data centre in Jamnagar, Gujarat. It is touted as the world’s largest data centre. With a projected capacity of three gigawatts, the facility will dwarf existing data centres, which currently operate below one gigawatt.

Jamnagar, home to Reliance’s oil refining and petrochemical operations, is key to the company’s renewable energy ambitions. Nearby, a 5,000-acre green energy complex, including solar, wind, and hydrogen energy projects, is under development. 

The Data Centre Dilemma

Major tech companies like AWS and Microsoft are investing heavily in Indian data centres, raising questions about their long-term sustainability. 

Dunkin pointed out that the focus should not just be on expansion but also on optimising energy and water usage. “Investing in data centres is good, but looking at how we’re going to reduce the cost, energy consumption, and water consumption is equally important.”

She added that AI’s energy footprint can be reduced through innovations in hardware and software. She used DeepSeek and other models to illustrate how customers will need less compute to run them. She also noted that smaller language models, which can operate directly on devices, are set to become more common.

Safety First

Beyond energy, Dunkin said that there is a need for AI governance in India. She stressed the importance of data security, regulatory safeguards, and ensuring AI models are built on diverse and unbiased datasets.

“If your model is trained by a bunch of American white men, you’re going to get a model that’s biased towards American white men. We need training data from all backgrounds.”

At the Paris AI Action Summit, Prime Minister Narendra Modi voiced similar concerns about biases in LLMs.  

“We must build quality datasets free from biases, democratise technology and create people-centric applications. We need to address concerns related to cybersecurity, disinformation, and other threats, and must ensure that technology is rooted in many ecosystems for it to be effective and useful,” said Modi. 

It becomes crucial for India to build AI frameworks that reflect its diverse linguistic and cultural landscape.

Agentic Systems 

Looking ahead, Dunkin sees agentic AI as a key productivity driver. However, she also cautioned against data privacy risks, advocating for personal AI models that do not share user data externally. 

Surprisingly, Dunkin doesn’t use much of the AI apps like ChatGPT due to privacy and security concerns. “I don’t use it a lot, in part because I don’t have a private model that I have access to, and so I don’t want my data out there,” she said. 

“Your personal model should live on your device or in your personal cloud. It should fetch external data without sending your private data out.”

This thought aligns with India’s push for digital sovereignty and greater control over data privacy in AI applications.