In April 2025, a coalition of more than 100 universities, scientists, and AI experts known as Climate TRACE began publishing monthly global greenhouse gas emissions data with just a sixty-day lag, tracking pollution from nearly 745 million individual assets across every country on Earth. That same month, the International Energy Agency released its Energy and AI report, warning that data centres consumed roughly 415 terawatt-hours (TWh) of electricity in 2024, about 1.5 percent of all global electricity, projected to more than double by 2030. These two data points capture the central paradox of our era: AI is simultaneously one of the most powerful tools we have to protect the planet, and one of the fastest-growing sources of strain upon it.

For ESG practitioners, investors, and corporate leaders, this tension demands concrete decisions about which AI applications to adopt, how to account for their environmental costs, and where the technology can deliver genuine climate value rather than sophisticated greenwashing.

Seeing the Atmosphere Differently

Weather forecasting was once the exclusive domain of supercomputers grinding through physics equations for hours. That paradigm is shifting. Google DeepMind’s GenCast, published in Nature in December 2024, demonstrated that an AI model could produce probabilistic weather forecasts that outperformed the European Centre for Medium-Range Weather Forecasts’ gold-standard ENS system on more than 97 percent of evaluated targets. Where traditional forecasts required hours on a supercomputer, GenCast generated a fifteen-day forecast in eight minutes on a single chip. Its successor, WeatherNext 2, announced in late 2025, pushed accuracy further still.

Open Climate Fix, working with Google DeepMind, has deployed AI forecasting models with the UK’s National Grid and a state grid operator in India, where the government has committed to 500 GW of non-fossil fuel capacity by 2030. Early results showed a 10 percent reduction in large forecasting errors and a 5 percent reduction in mean error, margins that at grid scale translate into millions of dollars in savings and meaningful emissions reductions.

Beyond weather, AI is reshaping how we track emissions. Climate TRACE now monitors greenhouse gas output across nearly 745 million assets worldwide, combining satellite imagery, machine learning, and ground-based validation to produce a near-real-time global emissions ledger. Their full-year 2025 data showed global emissions rose 0.50 percent to 60.63 billion tonnes CO₂e, yet power sector emissions declined for the full year, with China’s power sector posting its first year-on-year decrease since at least 2015. Before AI-enabled remote sensing, emissions inventories relied on self-reported national data that arrived years late.

Powering the Transition

The energy sector is where AI’s climate contribution may be most immediately tangible. According to BloombergNEF, global energy transition investment hit a record $2.3 trillion in 2025, up 8 percent from 2024, with electrified transport ($893 billion), renewable energy ($690 billion), and power grids ($483 billion) leading. Key highlights:

• Asia-Pacific accounted for 47 percent of the global total. China invested $800 billion; India climbed 15 percent to $68 billion.
• Climate-tech companies raised $77.3 billion in equity, up 53 percent year-on-year, led by multibillion-dollar deals from Asia.

“Despite policy and trade headwinds, the global energy transition is resilient and provides a number of opportunities for investors. Clean energy investment will continue to rise, especially as it relates to global data centre buildouts.”
- Albert Cheung, Deputy CEO, BloombergNEF 

AI is embedded throughout this surge. Google DeepMind’s machine learning has reduced data centre cooling energy by up to 40 percent. Applied to wind energy, its predictive algorithms have improved output predictability and increased energy value through smarter dispatch scheduling. In Australia, Tesla’s AI-managed Hornsdale Power Reserve has delivered over $150 million in consumer savings while providing grid stability.

Integrating intermittent renewables into ageing grid infrastructure remains the single biggest bottleneck in the energy transition. Google’s partnership with PJM Interconnection, the largest grid operator in North America serving 67 million people, aims to use AI to cut the interconnection approval process for new energy sources from years to months. For Southeast Asian markets like Vietnam and Indonesia, similar AI-driven grid planning tools could prove transformative.

Listening to the Forest

The World Wildlife Fund’s Forest Foresight uses AI and satellite imagery to detect early deforestation signs with approximately 80 percent accuracy. In the Amazon, Project Guacamaya, supported by Microsoft’s AI for Good Lab, deploys solar-powered microphones and bioacoustic analysis to monitor tropical forest soundscapes in real time. The Cornell Lab of Ornithology’s BirdNET identifies over 3,000 bird species from audio recordings. At sea, OceanMind uses AI to detect illegal fishing, a critical tool for Southeast Asia’s rich marine ecosystems. Investment in “naturetech” has grown at around 52 percent annually since 2018, with AI at the core. For financial institutions and investors, these tools offer the first scalable way to monitor biodiversity exposure across supply chains and asset portfolios, turning nature risk from an abstract concept into a measurable data point.

Spreadsheets and Systems

For corporate sustainability teams, AI is rapidly moving from aspiration to infrastructure. According to Deloitte’s 2025 C-suite Sustainability Report (surveying 2,100+ executives across 27 countries), 81 percent already use AI to advance sustainability goals. A Sphera survey found 73 percent of companies voluntarily disclose Scope 3 data, yet 45 percent have only limited assurance in that data. Platforms such as Persefoni, Watershed, and IBM’s Envizi use AI to automate emissions tracking across supply chains, mapping supplier-level emissions across hundreds of industries and geographies. Climate TRACE’s monthly data releases now feed directly into analytics engines that generate estimates in minutes rather than months.

The AI in ESG market, valued at $1.24 billion, is projected to reach $14.87 billion by 2034 (Market.us). With over 20 jurisdictions adopting ISSB-aligned disclosure frameworks, including Singapore, Hong Kong, Japan, and mainland China, AI-powered reporting is becoming essential for navigating overlapping requirements.

The Footprint of the Solution

None of this comes free. The IEA’s 2025 report made the scale of the challenge clear:

• Data centre electricity: 415 TWh in 2024, projected to reach 945 TWh by 2030, equivalent to Japan’s total demand. (IEA)
• The US and China account for roughly 70 percent of global data centre power. In the US, data centres will consume more electricity by 2030 than all energy-intensive manufacturing combined. (IEA)
• Data centre investment nearly doubled since 2022, reaching half a trillion dollars in 2024. (IEA)

“AI is a tool, potentially an incredibly powerful one, but it is up to us, our societies, governments and companies, how we use it.”
-Fatih Birol, Executive Director, IEA 

 

Water compounds the challenge. A Meta data centre in Georgia uses 500,000 gallons daily, ten percent of the county’s supply. An MSCI analysis of nearly 14,000 data centre assets found one in four may face increased water scarcity by 2050. A peer-reviewed study published in December 2025 estimated AI systems’ water footprint could reach between 312 and 765 billion litres in 2025 alone.

AI-enabled greenwashing is another growing risk: generative AI makes polished sustainability narratives trivially easy to produce, and the risk of “hallucinated” ESG data, where models generate plausible but fabricated statistics, is real enough that many vendors now anchor AI outputs in verifiable source documents. Equity matters too: Africa has less than one kWh of data centre electricity per capita versus over 540 kWh in the United States. If AI becomes the primary infrastructure for climate monitoring, those without access risk being excluded from the systems designed to protect the planet.

What Businesses Should Do Now

For ESG-conscious companies, particularly those operating across Asia-Pacific, the question is not whether AI will reshape sustainability practice, but how to engage with it responsibly.

1. Adopt AI for measurement before marketing -  Invest in AI-powered data collection and verification before using AI to generate sustainability reports. The 63 percent of companies already using or planning to use AI for ESG data (as reported by Veridion) are moving in the right direction, but adoption without governance creates liability.
2. Account for AI’s own footprint - Any credible ESG strategy incorporating AI must include the energy and water consumption of the AI systems themselves. Demand transparency from cloud and AI service providers about the environmental costs of compute resources, and factor those costs into Scope 3 calculations. Singapore’s approach, requiring industrial users to monitor and manage data centre water use, offers a regulatory model worth watching.
3. Watch the convergence of disclosure regimes - With ISSB-aligned standards spreading across Asia and the EU’s CSRD creating extraterritorial compliance obligations, companies need reporting systems that can map data across multiple frameworks simultaneously. AI tools that automate cross-framework alignment will shift from useful to essential within two to three years.
4. Invest in nature intelligence alongside carbon intelligence - Biodiversity and nature-related financial disclosures under the TNFD are following the trajectory climate disclosures charted a decade ago. Companies that build AI-powered nature monitoring capabilities now will be better positioned when mandatory nature reporting arrives, as it almost certainly will across major jurisdictions by the end of the decade.
5. Maintain human oversight - Every AI-generated emissions estimate, every satellite-derived deforestation alert requires human interpretation and institutional accountability. The technology can illuminate what is happening to the planet with unprecedented clarity. What we choose to do with that clarity remains, as it always has, a human responsibility.

AI can now map emissions, forests, oceans, and supply chains with unprecedented precision. But technology alone will not determine whether the planet stabilises or deteriorates. That decision still rests with the companies, investors, and policymakers interpreting the data.

 

Sources: IEA, BloombergNEF, Deloitte, Sphera, MSCI, Google DeepMind, Climate TRACE, WWF, Nature, Market.us. All statistics from 2024–2025 publications.