Companies that are creating technologies to reduce carbon dioxide emissions or scrub CO2 from the atmosphere could be some of the surprising winners of the international race to deploy artificial intelligence.
A Morgan Stanley analysis provided to clients Monday found that the global building boom of data centers — needed to meet the demand for generative AI — will likely emit the equivalent of 2.5 billion metric tons of carbon dioxide between now and the end of the decade. That total is comparable to 40 percent of annual U.S. emissions and will increase pressure on Silicon Valley to ramp up support for carbon-cutting technologies, according to the investment bank.
Most of that expected spike in climate pollution would come from the immense amount of energy needed to power generative AI processes — which identify patterns in vast data sets of information to create new texts, images and videos. Carbon is also released when constructing vast data centers from cement, steel and other industrial products.
“We see it as unlikely that governments will look to limit [data center] development given the numerous benefits associated with the [pursuit] of GenAI,” the analysts wrote, noting its potential to increase the efficiency of complex systems such as transportation networks. “This underscores the importance of [data center] decarbonization solutions.”
