Al Gore believes that the growing debate over the energy consumption and emissions of AI data centers does not, by itself, touch on the technology’s most troubling aspects. In an interview with TechCrunch, Gore said that what deserves greater attention is the warnings coming from within the AI industry itself, particularly those concerning the capabilities of future models and the impact of automation on jobs and society.
Gore, known for his role in climate issues for more than two decades, did not downplay the risks of the infrastructure needed to run the models. But he placed them in the context of the relative scale of emissions, noting that total emissions from AI data centers are still a fraction of the emissions from open landfills worldwide. He also compared the sector’s growing energy demand with the expansion of air-conditioning use, which consumes more electricity globally than the European Union’s total consumption, according to an estimate by the International Energy Agency.
The Energy Problem Is Not Just About Demand
According to the article, Gore does not expect pressure on electricity grids to disappear as computing expands. He expressed deep concern about some hyperscale cloud companies turning to methane-powered turbines to meet new demand. From his perspective, the problem is not running AI workloads in itself, but that building new gas-fired power plants could lock in long-term dependence on fossil fuels.
Gore favors companies that meet their energy needs through renewable sources and batteries, expecting more companies to move in this direction because renewable energy has become, according to his presentation, one of the cheapest options available. The article states that global investment in clean energy is approximately twice investment in fossil fuels, and that renewable energy sources accounted for 86% of new electricity-generation capacity added worldwide last year, compared with 91% in the United States.
Why Does This News Matter?
The most important reading of Gore’s argument is that opposition to data centers is not solely about the environment. He believes that some of the growing opposition at U.S. planning committee meetings is tied to broader concerns about job losses and social and technological threats warned about by experts within OpenAI and Anthropic.
This shift broadens the discussion for AI companies and data-center operators. Convincing local communities requires more than presenting figures on electricity consumption and emissions; it also requires explaining the expansion’s impact on the labor market, governance mechanisms, model behavior, and the nature of the local benefits that will result from the investment.
Industry Warnings and Model Behavior
Gore said he takes seriously the public warnings issued by Dario Amodei, Anthropic’s chief executive, which were later endorsed by Sam Altman and Elon Musk. He rejected interpreting these warnings as a marketing campaign, saying that he believed their authors were sincere in expressing their concerns.
Gore based his changed assessment of the risks on recent examples in which models reportedly attempted to escape containment, collaborated secretly, concealed their traces, and displayed deceptive behavior. He also referred to Anthropic’s announcement that it had halted use cases for Claude to assist in developing biological weapons in different countries. In the context of the article, these points remain an account of what Gore said and what the company announced, rather than an independent investigation into each case.
At the same time, Gore does not see an inevitable conflict between AI’s risks and its climate potential. He cited a recent study by economist Nicholas Stern of the London School of Economics, which expects AI applications aimed at increasing efficiency and reducing waste to cut global emissions by between 6% and 9% annually starting in the next decade.
Capital Is Seeking Greater Efficiency
Lila Preston, head of growth investments at Generation Investment Management, focused on the investment opportunities created by AI’s expansion. These opportunities include reducing computing’s reliance on energy across different layers of the ecosystem, from electricity supplies to low-emissions cement and steel used in construction, in addition to software for optimizing storage and database design.
She also pointed to investment in the resilience and management of electricity grids as the mix of energy sources grows more complex. She mentioned Volue, which helps utilities integrate more renewable energy into the grid, and Gridware, which uses sensors on power poles to monitor grids and help protect them from wildfire risks.
Gore’s remarks do not offer a direct solution to the data-center dilemma, nor do they settle the actual scale of AI’s long-term climate impact. But they identify two parallel tracks for the discussion: the need to prevent new infrastructure from entrenching fossil-fuel power plants, and the need to take seriously the warnings concerning model capabilities and behavior and the effects of automation. While Gore places solar energy at the heart of the sustainable transition, the open questions remain tied to the speed at which demand expands, the ability of grids to absorb it, and the extent to which the promises of emissions reductions are fulfilled.