

COP31 Panel Pushes Net-Impact Math to Settle AI Climate Debate
COP 31 Special Events • United NationsNovember 16, 2026
Locunity is an independent informational service and is not an official government page for this commission. All information is sourced from public footage and an analysis of official agendas with the aid of responsible, fact-checked AI. to report an error or omission.
A Nov. 16, 2026, COP31 side event set out to move the argument over artificial intelligence and climate from slogans to spreadsheets. Panelists backed the European Green Digital Coalition's net-impact methodology, which subtracts a technology's own emissions from the emissions it avoids. The proof points were a Japanese wheat pilot and Swedish apartment heating, along with a clear incentive: net-positive evidence can unlock green finance.
- Net-impact method credits AI projects with measured savings: NEC reports a net -3.0 to -4.6 tonnes CO2e per crop cycle.
- 74% of 140 analyzed national climate plans reference digital solutions. Only four address the digital sector's own footprint.
- Panel splits "solutionists" and "pollutionists" but agrees policy will decide AI's climate outcome.
- Equity gap acknowledged: the coalition says its social-impact framework is still in development.
Measuring AI's Net Climate Impact
Why it matters: Climate claims for AI are cheap. Verified net benefits are not. A shared accounting method could separate projects that cut emissions from those that only promise to, and it could steer financing toward the former.
The basics: The European Green Digital Coalition (EGDC) methodology nets the emissions a digital solution avoids against the emissions the solution creates. That includes embodied emissions, meaning those built into the hardware, and rebound effects. A rebound effect, sometimes called Jevons' paradox, occurs when efficiency gains drive more consumption.
Where things stand: Bob Bourgogne of the Carbon Trust walked through the method's core rules. It compares outcomes against scenarios, relies on primary data, requires comparability and sets transparency standards.
"It compares the carbon emissions that are saved by a solution against the emissions that are incurred by that solution," said Bourgogne.
Two case studies carried the argument:
- Japanese winter wheat: NEC Corporation's CropScope tool issued variable-rate fertilizer recommendations in a pilot. "Nitrogen usage was reduced by up to 22%," said Mari Ishikawa of NEC Corporation's Green Transformation Business Development Department. NEC put the net impact at -3.0 to -4.6 tonnes CO2e per crop cycle.
- Swedish housing: Bourgogne described AI heating control across 21,000 Swedish apartments that produced "389 tonnes of CO2 equivalent per year saved."
Luis Neves, CEO of GeSI, described a UN Solutions Hub and the Global Impact Accelerator as vehicles for scaling such projects. "We need to move to implementation," Neves said.
Decisions and stakeholder impacts: The panel took no formal action. The money logic was explicit, however. Moderator Ilias Yakovidis of the European Commission said green finance covers almost 50% of major bank commitments under the EU taxonomy, the bloc's classification system for sustainable investment.
"If you show that you're net positive with your IT, you can get green finance," said Yakovidis.
Technology vendors with documented savings stand to gain access to capital and, potentially, procurement preference. Developers without primary data or transparent footprints face a higher bar.
The other side: The method has a blind spot at the frontier of AI itself. "We don't understand the emissions associated with the training of AI models," said Bourgogne. He noted that training emissions are generally undisclosed. Net-impact claims for large AI systems therefore rest on incomplete inputs.
What's next: No dates or next steps were announced for the methodology's rollout.
Solutionists vs. Pollutionists: Policy Decides
Why it matters: Panelists said the AI-and-climate question has no fixed answer. The outcome depends on regulation, finance and who gets access.
Where things stand: A show-of-hands poll found both camps in the room: "solutionists" who see AI cutting emissions and "pollutionists" who see it adding to them.
Birgit Pickel, described by the moderator as General Director at the German Ministry for Sustainable Development and Climate, took the middle ground. "We know that AI is driving emissions at the moment," said Pickel. She added that AI offers large benefits for power grids and agriculture, and that policy will determine the net result. She also warned against deepening the digital divide between countries.
Pickel pointed to momentum at multilateral lenders. "Ajay Banga is a big fan of AI and digital solutions," she said. She noted the World Bank president wants AI built into funded projects.
Maria João Souza, Executive Director of Climate Change AI, described a "valley of death" where promising climate technologies stall between research and deployment. She said the gap is "where there needs to be more involvement from the private sector."
The other side: Neves of GeSI pushed back on skeptics. "There is going on a very negative conversation about AI, and that's the wrong approach," Neves said. He argued technology is needed for development.
Decisions and stakeholder impacts: Yakovidis noted that regulation lags technology. That gap leaves developing countries most exposed. They risk carrying AI's energy costs without the investment pipeline needed to capture its benefits. Industry groups want speed. Government and research voices want guardrails and deliberate financing.
ITU: Digital Climate Plans Ignore Their Own Footprint
Why it matters: Countries are writing digital tools into their climate commitments while largely overlooking the electricity those tools consume.
Where things stand: Tomas Lamanauskas, Deputy Secretary-General of the International Telecommunication Union (ITU), previewed findings from a joint ITU-UNFCCC stocktake. It covers 150 cases across 104 countries and reviews national climate plans, known as Nationally Determined Contributions (NDCs).
Of 140 NDCs analyzed with 2035 targets, "74% reference plans to deploy digital solutions," said Lamanauskas. Only four address greening the digital sector itself.
"Data centers already account for around 1.5% of global electricity consumption, with demand expected to double by 2030," he said.
Lamanauskas cited ITU standard TL 1480 for net greenhouse gas assessment and the Antalya AI Pledge, developed with the COP31 presidency. "The methodology exists. The technologies exist," he said. His point was that what's missing is their use in actual decisions.
What's next: Lamanauskas set three priorities:
- Measure impact from the outset.
- Build capacity in developing countries.
- Align financing with demonstrated benefits.
Audience Presses on Equity and Full Emissions Scope
Why it matters: The coalition's method measures carbon. It does not yet measure who benefits.
Where things stand: One attendee asked whether AI's climate role holds up if an AI economic bubble bursts. The same attendee asked whether emissions accounting runs cradle to grave, covering data centers, recyclability and Scope 3, meaning indirect emissions across a supply chain.
Yakovidis said users will keep using AI regardless of individual companies' fortunes. Bourgogne said all data should be incorporated, but data availability is the limiting factor.
Julie Collette of Plan International asked how assessments can account for social equity and the rights and participation of girls and young women. She also asked how Global South communities shape AI climate solutions designed in the Global North.
The other side: Yakovidis called social metrics the next frontier. "The social one is the hardest when it comes to numbers," he said. Neves cited solar-powered telecom base stations in Ethiopia as an example of inclusive deployment.
Decisions and stakeholder impacts: The coalition acknowledged its framework covers only the green axis. A social-impact framework is still being built. Until it exists, claims of a "triple win" for climate, economy and society remain unproven.
Minor Items
- Closing remarks: Nitin Arora, Head of the UNFCCC Global Innovation Hub, closed the session.
- Antalya AI Pledge: ITU flagged the pledge, developed with the COP31 presidency, as a governance anchor.
- ITU TL 1480: The ITU standard offers an existing tool for net greenhouse gas assessment of digital technologies.
Looking Ahead
No formal decisions were taken. Watch for release of the full ITU-UNFCCC stocktake report, progress on the coalition's social-impact framework, and whether lenders begin tying AI project financing to net-impact evidence.