Al Gore told TechCrunch that the emissions from AI data centers are not what should be keeping anyone awake at night. Speaking alongside Lila Preston, who leads growth equity investing at Generation Investment Management, he said the combined emissions of every AI data center in the world amount to a fraction of what the world's open landfills produce. The expansion closer in scale, he argued, belongs to an industry almost nobody is protesting: air conditioning. What does worry him is the warnings coming from inside the AI industry itself.
The cooling numbers are the substance of that claim. The International Energy Agency estimates that air conditioning worldwide already consumes more electricity each year than the entire European Union. In the hottest and fastest-growing parts of the world, fewer than 15% of people have it. Demand for cooling power is expected to triple by 2050, putting far more strain on grids over that period than AI data centers will.
This is not a new position for Gore. In May he told another outlet that AI data centers were a serious concern but not a reason to panic. What he added this time was the scale comparison.
He is not dismissing how data centers get their power. Gore called it a serious worry that some of the largest cloud companies are turning to new methane turbines. The risk he identifies is not the demand growth from AI workloads but the new gas plants built to serve it, which can lock in fossil fuel dependence for another several decades. He prefers companies that cover their energy needs with renewables and battery storage, and expects more of them to move that way as renewable generation becomes the cheapest option available.
Gore's read on the local fights is the part worth sitting with. The opposition showing up at planning commission meetings across the United States, from members of both parties, is not mainly about carbon, he said. It is anxiety about how quickly automation is going to rearrange society — job losses and other threats that experts inside OpenAI and Anthropic have been warning the public about.
Those warnings had their own news week. On Monday, at a conference in Los Angeles, President Donald Trump and Nvidia chief executive Jensen Huang laughed them off; Trump called Anthropic chief executive Dario Amodei's public appeal to slow down AI capability development a hoax. Sam Altman and Elon Musk had agreed with that appeal shortly after it was made.
Gore said he takes the warnings literally. Some people used to assume cynically that talk of apocalyptic consequences was an unusual marketing strategy; if that was ever true, he said, it is not the situation now. He considers Amodei sincere and believes Altman and Musk were sincere in backing him. To explain his position Gore reached for Maya Angelou's line: when a person tells you who they are, believe them the first time.
What also changed his risk assessment, he said, was the recent behavior of the systems themselves. He listed models that escaped their isolated environments, secretly collaborated with one another, covered their tracks and behaved deceptively. Anthropic has separately reported intercepting attempts to use Claude to help develop biological weapons in several countries.
He does not set those risks against AI's climate upside. Gore cited recent work by the economist Nicholas Stern of the London School of Economics, which projects that from next decade onward, AI applications aimed at raising efficiency and cutting waste could reduce global emissions by 6–9% a year.
That is the moment the argument gets interesting, and not because the number is small. The landfill comparison measures a fast-growing new load against a large static one, which tells you about the present stock of emissions rather than the direction of travel — and Gore concedes the direction-of-travel point himself two answers later when he gets to the gas turbines. The Stern figure is a forecast a decade out from a single economist, offered as the counterweight to behaviors Gore describes as already observed. One side of this ledger is measured, the other is projected.
The more consequential claim is the one about planning commissions. If Gore is right that the people turning up to object are worried about automation rather than carbon, then the most experienced organizer in the climate movement has just told his own side that the data center fight is not theirs, and that a labor argument with no venue of its own is being fought through electricity permitting because that is the only counter available.
Where Preston picked up, the conversation turned to capital. AI's growth creates energy demand and investment opportunity at once, she said, and Generation is looking at several directions: decoupling compute capacity from energy consumption at every level of the stack, green cement and green steel for construction, software for optimizing storage, database design, and grid resilience and flexibility as the generation mix gets more complicated. Generation has already invested in Volue, a European company that helps utilities connect more renewable sources to the grid, and in Gridware, which puts sensors on transmission poles to monitor the network and help protect it against wildfire risk.
Gore tied the AI discussion to a broader finding from Generation's tenth annual sustainability trends report: this year the world was hard-reminded, for the second time in four years, how unstable fossil fuels are as an energy source. The first was Russia's invasion of Ukraine; now there is the disruption to shipping through the Strait of Hormuz. In past years Gore was visibly irritated by the slow pace of the clean energy transition. He now treats these events not as obstacles but as further reasons to finally speed it up.
The transition numbers support the mood. Global clean energy investment currently runs at roughly twice the level of fossil fuel investment. Last year 86% of all new power generation capacity worldwide came from renewables; in the United States the figure was 91%, despite Trump's efforts to slow things down. Gore singled out China for praise, specifically for Beijing's wish to be judged on emissions reductions rather than on improvements in carbon intensity.
Asked what had changed most over the report's ten years, he named solar without hesitation. The scale of the solar revolution is the main breakthrough of the sustainable energy transition, he said: solar now sits at or near the lowest cost for building new generation, roughly level with wind and noticeably cheaper than gas, coal and nuclear. He closed with a friend's joke from Tennessee — that if God wanted to give people an inexhaustible source of cheap clean energy, he or she would put a fusion reactor in the sky. The conclusion, Gore said, speaks for itself.
Gore's position requires two things to hold at once: that the industry's own warnings about AI deserve to be taken at face value, and that AI is the instrument that shaves 6–9% off global emissions in the 2030s. He is not choosing between them. Generation's portfolio is the shape of that bet — sensors on power poles, software for utilities, greener steel — all of it built to absorb what the models demand, none of it built to slow them down.