Mumbai: Artificial intelligence (AI) is simultaneously creating a massive new climate challenge and offering some of the most powerful tools yet to manage it, noted Stephen Byrd, Global Head of Thematic and Sustainability Research at Morgan Stanley, speaking at the first Mumbai Climate Week Tuesday.
At the ‘AI and Climate Change’ session, on the opening day of the three-day event at Jio World Convention Centre that runs till 19 February, Byrd said the surge in AI-driven data centres could add 2-3 billion tonnes of carbon dioxide emissions globally before the end of the decade. But, he added, AI, if deployed correctly, could also deliver net climate benefits by optimising power systems and accelerating the shift to cleaner energy.
At present, India is estimated to have more than 270 data centres. Maharashtra leads with 89 centres followed by 34 in Chennai (Tamil Nadu) and 31 in Bengaluru (Karnataka). Mumbai alone hosts 44 data centres, the highest for any single city in the country. According to a report by Macquarie Equity Research published in October 2025, India has 1.4 gigawatts of operational data centre capacity, which is estimated to double by 2027.
Byrd framed India’s emerging advantage in the global data centre and power grid race squarely within the climate change debate, arguing that electricity availability is no longer just an infrastructure issue but a climate-defining one in the AI era. As AI-driven technology expands at an unprecedented pace, where and how that power is generated will increasingly determine whether the technology becomes a climate liability or a climate solution, he said.
Contrasting India with the US that has over 4,000 data centres and Europe that has around 3,500 data centres, he warned that advanced economies are already running into hard limits on grid capacity.
“To be very honest, in the United States, I don’t think we will have enough power access,” he said, adding that Morgan Stanley’s estimates show the US could be “10-20 percent short of the power needed”. Europe, he noted, faces similar challenges due to constrained grids and slower capacity additions, forcing difficult trade-offs between reliability, costs and emissions.
India, by comparison, was in a relatively stronger position, he said.
“A nation that does have enough power is an incredibly important national asset that not many nations have,” Byrd said, explicitly pointing to India’s power resources as a strategic advantage at a time when global demand for electricity from data centres is surging.
He suggested that access to reliable electricity could increasingly determine where the data centres, compute capacity and AI-led economic value are located.
Also Read: ‘India to attract investments worth $200 billion in data centres’ — Vaishnaw at AI Summit
Data centres & climate stakes
Byrd, who has spent over three decades in the energy and power sector, said discussions earlier in the day underscored four themes that resonated with him: the idea of the world as one family facing interconnected climate impact; the centrality of collaboration; the need to decouple economic growth from emissions; and the reminder that “climate is not a graph, climate is real people”.
Turning to AI, Byrd said Morgan Stanley’s analysis shows emissions linked to the construction and operation of data centres are set to rise sharply.
“Just to build and power data centres globally, we’re looking at over 2 billion tonnes of CO2 before the end of the decade, and possibly over 3 billion tonnes,” he said, noting “the scale is significant when compared with the United States’ annual emissions of just over 5 billion tonnes”.
Byrd described an unprecedented surge in power demand driven by AI. According to Morgan Stanley’s chip-level, “bottoms-up” modelling, the world will need to build about 130 gigawatts of data centre capacity between 2025 and 2028. To put that in perspective, Byrd said, “a large US city such as Philadelphia consumes around three gigawatts annually”.
He warned that this growth will strain power systems, particularly in the US and Europe.
“Our math suggests the United States would be 10-20 percent short of the power needed,” he said, adding that India’s relatively stronger power availability could become a strategic national asset in a world competing for electricity to fuel data centres.
Byrd also cautioned that India’s advantage comes with significant climate risks if not managed carefully. He repeatedly underlined the scale of the emissions challenge, stating: “Carbon emissions from data centres globally are likely to cross 2 billion tonnes of CO2 before the end of the decade and based on what I’ve seen, I think we’re likely to be above 3 billion tonnes.”
By this he implied that simply hosting more data centres could push emissions higher unless power systems are redesigned.
Drawing lessons from the West, Byrd warned of the dangers of poorly integrated data centre loads. Recounting an incident in the US triggered by a lightning strike, he said grid operators were shocked by how quickly data centre shutdowns amplified instability and nearly caused a regional blackout.
“We’re learning as we’re going, and that is not a great feeling,” he said, stressing that such volatility highlights the need for smarter grid management.
India, AI and climate action
Looking ahead, Byrd said AI capabilities are improving in a non-linear fashion, driving a “virtuous but concerning” cycle of rising compute and power demand. He pointed to data from OpenRouter showing that global weekly AI compute usage rose by over 2,000% in just 12 months, far outpacing growth in semiconductor supply.
He further noted that studies such as OpenAI’s GDPval suggest AI is rapidly approaching the ability to perform more than half of knowledge-economy tasks at human-expert levels.
For India, Byrd argued, the climate opportunity lies in avoiding those mistakes by deploying AI directly within the power system.
He said “the power grid is much like an incredibly complex biological system”, adding: “AI is a perfect solution for that complex problem because it can forecast demand and supply, manage variability from renewables and optimise the use of energy storage.”
If used well, this could mean “less fossil fuel generation and more effective utilisation of renewables”, Byrd asserted.
He also linked India’s clean energy push with AI-enabled flexibility on the demand side.
Referring to demand response, Byrd said flexible consumption is “incredibly helpful from an emissions point of view” because it allows grids to align usage with renewable generation.
“We’ve only uncovered about 10% of the benefit of demand response,” he said, calling it one of the most promising frontiers for cutting emissions while keeping power affordable.
Byrd positioned India at the intersection of two accelerating forces: climate change and artificial intelligence. With global AI power demand rising faster than expected, he noted that “every three months those numbers go up, not by a little bit but by a lot”.
According to him, India’s decisions on grid design, clean energy integration and AI deployment will shape not just its own emissions trajectory but also the global carbon footprint of AI.
For Byrd, one of the most promising frontiers is dynamic demand response, that is “using AI to flex consumer and industrial electricity demand to match renewable generation”. According to him, this could cut emissions, lower costs for consumers and unlock far greater value than has been realised so far.
He also highlighted AI’s already established potential to transform battery chemistry. “Cheaper, more efficient batteries could dramatically accelerate renewable energy adoption,” he said.
“The world is going to look very different next year, and it’s up to us to make sure AI delivers a net benefit for the climate,” Byrd said, while underlining that AI’s pace of development is likely to outstrip current forecasts, intensifying both emissions challenges and problem-solving capacity.
Byrd said that “if managed well, India could demonstrate how rapid digital and economic growth can be aligned with climate goals, rather than coming at their expense”.
(Edited by Nida Fatima Siddiqui)
Also Read: Inside Chennai’s data centres. Fortresses of the AI era