Amazon’s Texas data center project, a massive off-grid power plant, is raising serious concerns about the environmental impact of artificial intelligence and the future of the US grid. While the plant’s size is significant, its reliance on simple-cycle turbines – turbines that mix fuel and air to create energy – is a major source of greenhouse gas emissions. This shift, driven by the increasing demand for data centers, poses a substantial threat to the environment and could ultimately destabilize the US energy system.
Simple-cycle turbines, commonly used at traditional power plants, are significantly less efficient than their combined-cycle counterparts. This means they release fewer energy-equivalent emissions per unit of power generated. According to the US Environmental Protection Agency, even if Amazon’s facility emits half of what’s permitted annually, it would still contribute significantly more greenhouse gas pollution than 78 average-sized natural gas plants. The Texas plant’s permitted emissions of up to 33 million tons of greenhouse gases a year are a stark reminder of this issue. The company’s email to Wired Science indicates that Amazon is still committed to achieving net-zero emissions by 2040, despite these concerns.
This trend is accelerating as data centers increasingly require flexible, quick-to-install power. At least three power plants in Ohio by Williams, for Meta, are utilizing simple-cycle units, with at least three power plants being developed in Ohio by oil and gas company Williams for Meta, with 39 simple-cycle turbines, and a similar number of turbines at Crusoe’s Stargate campus. The permits for Google’s Goodnight data center at another Texas site being developed by Crusoe describe using 20 simple-cycle turbines, and both xAI’s Colossus 1 and Colossus 2 data centers in Memphis are powered by dozens of simple-cycle turbines. All of these facilities are permitted to release millions of tons of greenhouse gases.
Industrial Info Resources’ senior vice president, Britt Burt, notes that simple-cycle turbines are ‘not the most efficient way to use a gas turbine.’ He suggests that the energy needs of data centers, particularly those training AI models, are disproportionately large, requiring rapid startup and shutdown cycles. This rapid switching can strain turbines, and the extreme stress can cause them to break. Stephen Lynch, a professor of mechanical engineering at Penn State, echoes this concern, stating that ‘if you have a highly fluctuating load, it’s better to have 30 things that can individually turn on and off than to have one giant thing that doesn’t like to be turned on and off.’
Furthermore, the federal government and several states are attempting to encourage data center power providers to move away from these inefficient turbines. The US Environmental Protection Agency has issued guidelines to promote energy diversification, including the possibility of integrating data centers with the grid. Williams has indicated they might consider the Ohio plants for interconnection in the future.
The potential for a stranded asset scenario looms, if the AI bubble bursts or tech companies’ energy needs decline. However, the massive, inefficient gas plants could continue to operate for decades, threatening to exacerbate the climate crisis. The combination of simple-cycle turbines, the increasing demand for data centers, and the geopolitical implications of AI’s energy demands creates a significant risk – a parallel fossil fuel grid that generates a substantial amount of greenhouse gas emissions, potentially outweighing any benefits of this new data center infrastructure.
Watch Related Video
Source: Wired Science
