Wildfire smoke creates orange skies over the Bay Area on Sept. 9, 2020. San Francisco is the capital of the AI boom, but the industry creates significant emissions that are making climate change and wildfires worse.
Jessica Christian/The Chronicle
Last week, a disorienting wintertime heat wave shattered March temperature records across the Bay Area. Thanks to human-caused climate change, every year we’re seeing many more high temperature records fall than record lows being set, and that discrepancy is growing as the world continues to warm.
Like many cities around the world, San Francisco has taken action to lower the greenhouse gas emissions that are causing climate change. City ordinance and its 2021 Climate Action Plan set a goal of net zero emissions by 2040. But buried in the plan is a quieter truth: Even if San Francisco implemented every strategy (including eliminating natural gas from every building, replacing all gas-powered cars with electric vehicles, and shifting most transit in the city away from driving altogether), it would still fall short.
That shortfall matters even more now because San Francisco has positioned itself as the global hub of artificial intelligence. The companies building these systems are in the city, and the political and economic decisions that enable their growth are being made there, too. But much of the energy required to power the AI boom, which is resulting in significant emissions that are making climate change worse, is being generated in data centers located in other places. Because of this, San Francisco’s Climate Action Plan no longer represents an honest accounting of the problem.
Article continues below this ad
The plan still focuses on what are known as Scope 1 and Scope 2 emissions generated within the city’s borders and from the energy consumed there. But the majority of the city’s climate impact sits outside that frame. Scope 3 emissions, as they’re called in the city’s climate literature, are made up of everything else, including the goods we buy and the services we use. In a wealthy city like San Francisco, those emissions are enormous. And they are growing quickly as AI becomes embedded across the economy — powering everything from media to customer service to logistics and transportation — because these systems rely on energy-intensive computing in data centers outside the city. The more our economy runs on AI, the less our climate accounting reflects the reality of our impact.
This presents a challenge, as AI has become one of San Francisco’s clearest economic bright spots. But the same industry helping fuel the city’s recovery is also helping drive one of the world’s fastest-growing sources of emissions. Yearly carbon dioxide emissions from data centers will reach 300 million metric tons by 2035, according to the International Energy Association — an amount equivalent to roughly 65 million more cars on the road for an entire year.
San Francisco Chronicle Logo
Make us a Preferred Source to get more of our news when you search.
Add Preferred Source
If San Francisco’s leaders in City Hall and in the tech sector are going to claim the economic benefits of the AI boom, they should also have to account for the climate impacts that result from it.
True, the city is facing a huge budget deficit, and we’re competing for jobs and investment. Keeping the tech sector in California, and San Francisco specifically, is seen by many as an economic imperative. But when the industry you’re busy promoting is also the source of a much bigger problem, the short-term economic benefits start to feel less essential.
Article continues below this ad
It is an implicit gamble that someone else — another state, the federal government, the tech companies themselves — will deal with the emissions being generated by AI. And everyone, including residents of San Francisco, will end up paying the price, whether through more dangerous storms that slammed Hawaii this week, the scorching heat waves gripping the Western U.S. or the enormous public cost of building seawalls to address sea level rise.
AI data centers located in other states can be powered by coal plants, for instance, the emissions from which will negatively impact the Earth’s atmosphere for hundreds of years. If San Francisco wants to claim the economic upside of AI growth, it cannot treat the climate impacts as someone else’s problem simply because the infrastructure powering the industry is located elsewhere.
Guest opinions in Open Forum and Insight are produced by writers with expertise, personal experience or original insights on a subject of interest to our readers. Their views do not necessarily reflect the opinion of The Chronicle editorial board, which is committed to providing a diversity of ideas to our readership.
Read more about our transparency and ethics policies
We can’t manage what we don’t measure. At the very least, San Francisco should require tech companies to report the emissions associated with their activities, including those that occur outside the city. Similarly, we should not overlook the growing use of AI by city employees. A single ChatGPT query uses about 0.3 watt-hours of electricity — trivial on its own, but multiplied across dozens of prompts a day, across thousands of employees, it adds up quickly. That doesn’t mean city workers shouldn’t use AI. But it does mean the city should track that usage, account for the associated emissions and make sure it aligns with our climate goals. Every local policy — including attracting new business — is climate policy, whether we acknowledge it or not.
San Francisco likes to see itself as a leader when it comes to innovation. Right now, thanks to the rise of AI, that title has become sorely outdated when it comes to fighting climate change. True leadership must be demonstrated, again and again, through choices that match the scale of the problem. If we are not willing to do that here, in one of the nation’s wealthiest cities, it’s hard to imagine where it will happen.
Article continues below this ad
Michael Redmond is a native San Franciscan. He holds a master of engineering degree in applied climate from the University of Michigan and a bachelor’s in environmental science from UC Santa Cruz.