By HERMAN MORRIS
In 2019, responding to employee pressure and a walkout of thousands of workers, Amazon declared that it would be joining the climate pledge to reach net zero carbon emissions by 2040. Microsoft would also join the climate pledge, with Apple, Oracle, and Meta making similar promises. As part of these commitments, tech companies resolved to source carbon-free energy for data centers, rely on zero or low-emission transportation, and move toward recyclable materials for their physical electronic products. Given these lofty ambitions, one would expect a trend line of emissions going down. Despite these goals, however, Big Tech companies are emitting more emissions than ever, with many of them increasing due to investments into AI data centers.
On a fundamental level, Big Tech’s climate goals have always been a smokescreen. Their profits are built on the backs of extracting, transporting, and refining raw materials into chips, computers, warehouses, and cables. The processes to create these end products are inherently emission creating. The entire business model of Big Tech companies is tied to selling more physical computer devices and connecting more people to the internet and their services every year. With pressure to keep profits up as tech becomes a load-bearing pillar to the entire U.S. economy, Big Tech was always bound to renege on any promise to complete a green transition. However, the advent of AI data centers has introduced a complete about-face to pursuing one of the biggest accelerations of emissions in the modern global economy.
A false start
Data centers are already one of the biggest drivers of greenhouse gas emissions for tech companies. Powering data centers globally contributed to the IEA-reported 330 megatons of CO2 recorded in 2020, or roughly 1% of all greenhouse gas (GHG) emissions that year. Already, this measure is a lie that is inherent to net-zero emission targets. Net-zero emission targets (as opposed to “true” zero emissions) allow for an entity to still emit greenhouse gases, so long as they offset these emissions through the purchase of carbon offset credits or other certificates. The Guardian reports that through the use of renewable energy credits (RECs), tech companies are able to buy and use green energy at any other facility in the world, and then use that energy to mark down their data center emissions. This standard of reporting does not even meet the spirit of net-zero energy as there is no carbon capture or any other act of counteracting the total emissions from data centers. Through this creative accounting, data centers can under-report their emissions, which are estimated to be over seven times higher when RECs are removed.
A strain on the Earth and the working class
The development of AI data centers poses an even higher rate of emissions from tech companies. Training and operating AI models must be done with special hardware that most existing data centers are not equipped to provide. This has led to a huge build-out of new infrastructure from Big Tech companies. So far, $750 billion is estimated to have been spent on new data center construction.
In one case, Meta is proposing a data center the size of Manhattan as part of its push to create a new cluster of AI data centers. All these capacity additions have led Big Tech companies away from their climate goals, with Google, Amazon, and Microsoft now creating more emissions than they were in 2021, and Google increasing their emissions by 11 percent in 2024 alone.
While the cost of training the latest AI models is shrouded in mystery, the cost of operating them can be an order of magnitude more expensive than operating services such as Google. Goldman Sachs has estimated that a ChatGPT query requires nearly 10 times more electricity than a Google search. One analysis puts total potential AI-oriented hardware (based on annual GPUs sales for data center use) emissions output at 3.25 gigatons of CO2 a year, or 7% of all emissions produced globally in a year. The current strategy for developing AI tools is not only raising emissions but is creating a new class of software that will cost the planet more than previous web-based tools such as Google search or email.
The exorbitant energy demands of AI are causing tech companies to run up against the existing limits of global energy grids. Searching for a path out of this crisis, Amazon and Google have announced partnerships with energy companies to explore using nuclear energy for their AI data centers. Meanwhile, Meta has an open request for nuclear energy proposals to provide one to four gigawatts of electricity for its AI goals. Future projections place the average electricity bill increase due to AI for Americans to be 8% by 2030, with certain hotspots like northern Virginia to see around 25%.
There is also growing pressure on water use. In comparison to traditional data centers, the estimated total water footprint of Google hyperscale data centers that power Gmail and Google Drive was about 200 million gallons of water a year. At the low end, global AI data center water use is projected to be more than 1 trillion gallons of fresh water a year by 2027. One-fifth of those data centers will be in water-stressed regions.
Developing AI while the world burns
So how are tech leaders working through these contradictions of claiming to save the planet while pursuing more emissions than ever before? At an AI summit in Washington, D.C., former Google CEO Eric Schmidt was asked about balancing AI development with climate goals. His response was: “We’re not going to hit the climate goals anyway because we’re not organized to do it,” and that instead, more attention should be paid to accelerating AI development. Bill Gates made similar comments at a London event, claiming that any energy grid strain would be made up for by new efficiencies that AI unlocks.
Both comments reveal the actual motives behind AI investment for Big Tech companies. Right now, AI’s ability to positively impact climate change is highly theoretical, with no examples thus far being able to remove or capture emissions at the rate that its development releases. Climate change, on the other hand, is happening right now and destroying entire communities, with the UN forecasting a global temperature rise of 3.1C this century—a far cry from the 1.5C target. Even in a scenario where AI applications do unlock massive energy efficiencies that lower emissions, there is no reason to believe that this solves climate change. Without a fundamental change in how the global economy is structured, efficiencies discovered through new tools will only enable a further expansion of production and extraction from the earth. In truth, tech capitalists are throwing in the towel on the climate change debate and admitting they cannot solve it and would rather focus on their next scheme to keep profit margins up.
For workers’ control and nationalization of Big Tech
As of 2025, the richest seven tech companies in the U.S. account for over one-fourth of the S&P 500 in the U.S. By themselves, they have a market cap equivalent to the Japanese, Canadian, and UK economies combined. The wildly irresponsible investments in AI and its potential to deepen a growing climate crisis are a consequence of wealth belonging to a handful of bureaucrats who run the Big Tech firms. Their need to grow profits to stay in power naturally leads to the situation today in which they are actively making the world less habitable to try and find one more industry to break into.
On the other hand, working people will be forced to bear the environmental and economic burden of pursuing new AI developments. This, plus their collective experience in having to integrate AI tools into day-to-day work, makes them far better at rationally assessing and planning investment into AI.
To manage the climate crisis, it is necessary to take control of these firms away from the hands of the tech capitalists and put them into the hands of the working class. The Big Tech firms must be placed into public ownership, with workers in charge of determining how to structure and run production. Nationalization and workers’ control of the Big Tech firms can ensure actual debate and democratic votes on how much and where to invest the immense wealth that tech capital has produced, as opposed to leaving it up to an unaccountable bureaucracy.
What would this look like? First, the Big Tech companies and their immense wealth would be placed in public hands, with unions and shopfloor committees organizing production. Those unions and committees would have democratically elected leadership from the rank-and-file workers, running and planning production. Under this model, the debate of how much to invest into AI, where to try to apply it, and how to mitigate its climate impacts could be held and decided collectively by workers and not in a closed-door board meeting.
In the short term, this would create the conditions to argue for a dramatic decrease in AI investment regarding data center construction and operation. In the long term, this would allow for a more critical investigation into how to develop AI tools that are energy conscious and explore uses of AI that propose a benefit to humanity.
This structure has other benefits; it eliminates the irrational race to develop technology, as now there is no longer a profit motive to scramble and be the first to discover the next AI breakthrough. It also means being realistic about what is possible with the technology and no longer making sci-fi promises on AI’s potential to appease investors. Perhaps most importantly, it allows workers to determine how these tools will be used in the workplace. This ensures that new AI tools will help make workers’ lives easier, eliminate the less desirable parts of the job, and lower working hours, instead of using it to increase surveillance of workers and to try to impose an intensification of the workday.
Time is running out on meeting the climate goals put forward by the UN and the Paris Agreement. The experience of the past several decades has shown that capitalists are not up to the task of planning an economy that can maintain balance with the Earth. The example of Amazon workers walking out in 2019 shows that a mass impulse exists to protect the environment around us, even in the corporate workforce that is normally mollified through high wages and good benefits.
The clock is ticking for meeting the 1.5C limit that would prevent the worst of what climate change has in store, but there is still an opportunity to mitigate much of the catastrophic damage that climate change could cause. Workers can and must take control of the economy away from the capitalist class; to do anything less is to trust our future to a group that never really cared about the climate crisis in the first place.
