2 days ago
How AI can be a solution — not a problem — in the fight against climate change
written by Zenin Osho
In Maharashtra's drought-prone Baramati district, sugarcane farmers have long faced a tough trade-off: Maximise yields or conserve water. Now, with the help of artificial intelligence (AI), they are managing to do both. Farmers are using AI-driven predictions to optimise irrigation schedules leading to a 30 per cent reduction in water use. Crucially, it has also cut electricity costs for farmers by around 25 per cent, since less water means less reliance on power-hungry pumps. It hints at a broader truth: AI, despite concerns over its energy use, can help drive real-world climate solutions by making industries leaner, cheaper, and greener.
Much of the anxiety around AI stems from its growing appetite for electricity. Training large models consumes roughly 10 times more energy than a traditional web search. Greenhouse gas emissions from big technology companies have risen by nearly a third in recent years. With vast new data centres being built, further increases seem inevitable.
Yet the alarmism is often misplaced. In absolute terms, AI remains a relatively modest consumer of energy. According to the IEA, data centres account for about 1.5 per cent of global electricity use today, and that figure could double by 2030. But most of it is driven by streaming, social media and e-commerce, not AI. Even if AI's share grows sharply, its potential to decarbonise some of the hardest-to-abate industries — while tackling both carbon and short-lived climate pollutants like methane — is becoming increasingly difficult to ignore.
Take methane, for instance. Although less notorious than carbon dioxide, methane is a far more potent, if shorter-lived, greenhouse gas. Tackling it quickly could offer major climate gains. AI-powered startups are already rising to the challenge. GHGSat, for example, uses satellites equipped with advanced spectrometers and machine learning to detect facility-level methane leaks invisible to conventional monitoring.
Livestock, particularly cattle, are another major methane source. Startups like Rumin8 and Symbrosia are developing AI-informed feed supplements that curb emissions from digestion. Meanwhile, DSM-Firmenich's Bovaer, now approved for use in over 55 countries, can slash methane emissions from cattle by more than 30 per cent. Agriculture offers further opportunity: Flooded paddy fields, which produce significant methane, could also benefit from AI. Just as AI tools are helping sugarcane farmers in Baramati optimise irrigation and cut water use, similar approaches could reduce flooding periods in rice cultivation — lowering methane emissions while conserving water.
AI's promise in modernising energy systems is only just beginning to be realised. Use cases in renewable energy integration remain early, but encouraging signs are emerging. In the United States, Alphabet's Tapestry project, combining AI and cloud technologies, is helping grid operators automate the sluggish approval process for clean energy projects — speeding the deployment of wind and solar power. Similar challenges, albeit on a larger scale, loom in India. Integrating intermittent renewables into ageing, stressed grids remains complex. Distribution companies (discoms), which are entities responsible for buying electricity from generating companies and distributing it to end-consumers across different areas, many of which are financially strained, face acute difficulties in adopting new technologies.
Yet, AI offers powerful tools. It can improve demand forecasting, optimise grid load balancing, predict faults before they cascade, and automate grid planning, significantly expediting renewable integration. Crucially, Indian startups such as Ambee, Atsuya, and Sustlabs are actively deploying AI and IoT for sustainable energy solutions. Given India's ambitious goal of adding 500GW of non-fossil capacity by 2030, these efficiencies are simply no longer optional. While widespread AI adoption among discoms may still seem distant, the potential gains — in reduced losses, enhanced reliability, and lower costs — make a compelling case for phased, strategic deployment, supported by policy reform and investment.
Batteries, too, are critical to this transition. The ability to store renewable energy when the sun does not shine or the wind does not blow remains a bottleneck. Quantum computing, closely linked to advances in AI, offers a tantalising possibility. By simulating new battery materials, such as lithium nickel oxide, at the atomic level, researchers hope to design cheaper, longer-lasting storage solutions, accelerating the shift to a cleaner grid. Lithium nickel oxide is a promising material that could enable batteries with higher-energy density and lower costs compared to conventional lithium-ion designs.
Teams at Sandia National Laboratories and Google Quantum AI are already using quantum simulations to accelerate battery research. They are also applying quantum techniques to improve modelling of fusion reactions, potentially unlocking a future of abundant and carbon-free energy.
Industrial sectors that have long resisted decarbonisation are also beginning to show signs of change. Cement manufacturing, responsible for around 8 per cent of global emissions, is deploying AI to optimise kiln operations, cutting fuel use and emissions. In shipping, AI-driven navigation systems analyse real-time data on weather patterns and ocean currents to chart more efficient routes, saving time, fuel, and carbon.
Startups are crucial in pushing these frontiers. Their agility and willingness to bet on unproven ideas give them an edge over slower-moving incumbents. Startups need deep ecosystem support, including patient capital, reliable infrastructure, expert mentorship, and clear regulatory pathways. Initiatives like Google's startup programs provide a template, offering access to advanced AI models, cloud computing resources, and tailored guidance to help founders navigate technological and policy hurdles. The government's role in strategic investment in R&D, targeted support for climate-focused startups, and regulatory frameworks that encourage innovation without creating unnecessary barriers are all essential.
Transparency on AI's environmental impact is critical. From 2026, the European Union will require companies to report AI-related energy consumption; other jurisdictions should adopt similar measures. Data centres must evolve as well, shifting workloads to match renewable generation, investing in battery storage, and aiming for 24/7 carbon-free operations. Big technology firms should leverage their considerable purchasing power to accelerate the build-out of clean energy infrastructure, rather than relying primarily on offsets.
Combating climate change demands we tackle both carbon and super-pollutants like methane. While concerns about AI's energy footprint are valid, its powerful potential for deep decarbonisation and systemic change is undeniable. If policymakers, investors, scientists, and entrepreneurs unite, AI can transform from a perceived climate problem into one of our most potent solutions, with startups already blazing the trail towards a new era of innovation that matches the urgency of the challenge ahead.
The writer is Director, India Program of the Institute for Governance & Sustainable Development (IGSD)
