Via costar.com
The AI giant is building out a Wall Street-style energy trading desk to hedge billions of watts in power consumption across its expanding data center fleet.
OpenAI is looking for someone who can trade electricity the way its models trade tokens. The company has posted a job listing for a Power Trading Lead, a senior role responsible for managing hedging strategies across electricity and natural gas markets as OpenAI races to build out the data center infrastructure its AI ambitions demand.
The position sits within OpenAI’s Power & Land team and comes with a salary range of $181K to $285K plus equity.
What the role actually involves #
The Power Trading Lead will own OpenAI’s hedging strategy for electricity and natural gas exposure. That means building out positions using forwards, swaps, options, and basis-risk mitigation tools, the same instruments that energy traders at utilities and hedge funds use daily.
Candidates need more than a decade of experience in power trading or energy markets. OpenAI wants someone who deeply understands US wholesale power markets, utility tariffs, and natural gas pricing dynamics.
The scale of OpenAI’s energy problem #
The company’s upcoming facilities are projected to require around 3,200 megawatts of power. For context, that’s roughly enough electricity to power a mid-sized American city.
OpenAI has already locked in part of that supply. In July 2026, the company signed a 25-year agreement with a subsidiary of Southern Company securing up to 1,000 MW of power. That’s a substantial commitment, but it still leaves more than two-thirds of OpenAI’s projected demand uncovered.
The gap between secured supply and total need creates significant exposure to energy price fluctuations. A 10% move in power costs at 3,200 MW scale translates into potentially hundreds of millions of dollars annually.
Why this matters beyond OpenAI #
The 25-year term on the Southern Company deal is itself remarkable. That’s the kind of contract length you see in utility-scale renewable energy projects or industrial power purchase agreements.
The role also hints at how OpenAI thinks about its cost structure going forward. Energy is increasingly one of the largest variable costs in running large language models. Training runs for frontier models consume enormous quantities of electricity, and inference at scale, serving hundreds of millions of users, compounds that demand continuously.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our