AI's Data Center Demand Shadows Climate Week as Founders Split

Climate Week drew its usual crowd of founders, investors and policymakers, but the loudest conversations in the halls were about data centers and artificial intelligence. TechCrunch reported that the same divide over AI's costs and benefits now splitting the wider U.S. has reached the climate tech community, turning what was once a fairly unified gathering into a debate about power, land and money.

The tension is easy to see in the numbers. Data center electricity use in the U.S. is climbing fast after roughly two decades of flat demand, driven by training and inference for large models. Utilities and grid operators have responded by delaying coal and gas retirements in some regions and by fast-tracking new gas plants, which sits badly with founders whose companies exist to cut emissions.

For some climate tech investors, though, the boom is a customer. Data center operators need firm, carbon-free power around the clock, and they are willing to sign long-term contracts that solar and wind alone struggled to attract. That has boosted interest in geothermal, advanced nuclear, long-duration storage and grid software, categories that spent years chasing smaller deals.

Others see a zero-sum fight. When hyperscalers outbid everyone for clean power and transmission capacity, smaller projects get squeezed, and the emissions math gets worse in the near term even if the long-term grid gets cleaner. Several founders at the event argued that AI companies should be paying directly for the grid upgrades their load requires, rather than passing the cost to ratepayers.

The split also runs through talent. AI labs pay salaries that climate startups cannot match, and a handful of engineers have left energy companies for model developers in the past year. Investors said that drain is real but limited so far, and some noted that AI tools are now speeding up materials discovery and grid forecasting in ways that help their portfolio companies.

What happens next depends on a few concrete decisions. Regulators in several states are weighing how to allocate the cost of new transmission and generation for large loads, and utilities are reworking interconnection queues that already stretch years. If data centers keep growing at the current pace, the climate tech community may have to settle whether AI is a customer, a competitor, or both.