The rush to advance AI has brought with it a proliferation of data centers, and in turn, a significant increase in energy consumption. Tech companies have until now been seen as climate leaders, making market-leading commitments to reach net zero emissions by 2030. However, their ambitions of rapid AI development are putting these commitments in jeopardy.  

Open MIC and our partners are working to develop finance-focused strategies to improve transparency around the environmental costs of data center infrastructure and the risks these pose to investors. 

Since Big Tech’s embrace of Large Language Models (LLMs) and the applications based on them, carbon emissions across the sector have risen dramatically. Google’s emissions in 2023 were 13 percent higher than the previous year. From 2021 to 2023, Meta’s emissions jumped 66 percent. Microsoft plans to spend $50 billion to expand its data centers in the coming year, which they admit in their 2024 sustainability report is a significant source of emissions. 

This trade-off may not even be necessary. The ascendancy of DeepSeek, a Chinese artificial intelligence app with functionality similar to ChatGPT, has stunned investors in its ability to create impressive AI models while relying on fewer and less-advanced resources. This calls into question whether massive investments in data centers and related power infrastructure are even essential for the advancement of AI. 

Yet, as companies like Alphabet, Amazon, Meta, and Microsoft pour billions of dollars into new data centers, their ambitious goals of using 100 percent renewable energy seem less and less obtainable. Moreover, it is likely that their AI-related emissions are far greater than currently disclosed, as these tech giants do not publicly account for the electricity consumed by the companies they have investments and partnerships with, like OpenAI and Anthropic. 

Big Tech companies have also relied on renewable energy certificates (RECs) to meet their climate targets. RECs allow companies to claim their energy consumption is more environmentally friendly while still relying on fossil fuels. This significantly distorts their true carbon footprint. Both Amazon and Meta have been called out for misleading claims about their energy usage. (To its credit, Google has phased out the use of RECs in favor of matching carbon-free energy on an hourly and location-specific basis. Microsoft has made a similar commitment that it aims to fulfill by 2030.)

Increasing energy demand has left companies seeking other, potentially more harmful energy sources. Nuclear power projects, such as the previously-shuttered Three Mile Island plant, are becoming attractive prospects.

The energy demands of Big Tech’s data center expansion will attract regulatory scrutiny, potential carbon taxes, litigation and reputational risks, and may ultimately have a weak return on investment. Open MIC believes investors have a right to know exactly how much Big Tech’s emissions have increased as a result of their AI ambitions and what steps these companies are taking to meet their stated climate goals.