OpenAI builds an energy desk to power its AI empire

For most of the past two years, the story of AI infrastructure has been told in gigawatts and billions of dollars. OpenAI’s own hiring page is now telling a quieter but arguably more revealing version of that story. Over the past month, the company has posted job listings for a Power Trading Lead and a Clean Energy and New Technology Lead, both sitting inside its infrastructure organisation rather than bolted on as external consultants. Taken individually, either hire might read as a routine staffing update. Taken together, they mark OpenAI treating energy not as something it buys, but as something it actively manages.

A trading desk for a company that makes chatbots

The Power Trading Lead role, first reported by Bloomberg in early August, is the more striking of the two. Sitting within OpenAI’s Power & Land team, the position will run commodity hedging strategy across the company’s electricity and natural gas exposure, using the same instruments, forwards, swaps, options, that utilities and energy trading desks rely on daily.

It carries no direct reports initially and pays between $181,000 and $285,000 plus equity, and candidates are expected to bring at least a decade of power trading or utility strategy experience.

OpenAI’s own listing frames the stakes in unusually direct language for a job posting. The Power & Land team, it states, exists to help “secure reliable, scalable, and economically resilient power” for the company’s global data centre portfolio.

That is not the language of a business paying a monthly electricity bill. It is the language of a business that has decided its electricity exposure is large enough, and volatile enough, to need its own dedicated risk function, the same conclusion Meta reached when it stood up a power-trading operation in November 2025.

A second hire suggests this goes beyond financial hedging

Roughly two weeks after the trading role emerged, OpenAI posted for a Clean Energy and New Technology Lead inside its Industrial Compute organisation, the group responsible for site selection and buildout. The role will evaluate clean power, advanced storage, grid flexibility and low-carbon backup technologies, and identify which of them can move from pilot projects to repeatable deployment across OpenAI’s growing campus footprint.

Neil Osnato, founder of Persistence Analytics Group, whose reading of the interconnection queue crunch in Texas Capacity covered earlier this year, argues the significance of the role lies less in its title and more in where it sits.

“The key phrase in the posting is not ‘clean energy,’” Osnato said, quoted in trade coverage of the hire.

His point is that a role owning “emerging-energy strategy and execution” from inside the buildout organisation is a different animal to a sustainability team issuing recommendations from the sidelines. It suggests OpenAI wants the capability to judge, project by project, which power architecture is actually deployable, rather than treating energy as a procurement decision made late in a project’s timeline.

The clearest existing example of that logic in practice is Project Camellia, OpenAI’s 3.2 gigawatt campus planned for Effingham County, Georgia, where the company has already committed to provide Georgia Power with up to 1,000 megawatts of flexible demand response under a 25-year agreement.

Why the rest of the industry should read this as a signal

None of this is happening in isolation. Capacity has reported that OpenAI paused its Stargate UK project outright, citing the cost of energy and an unresolved regulatory environment, and that Microsoft subsequently stepped in to take over capacity at the Stargate Norway site after OpenAI failed to reach an agreement there.

Building internal capability to hedge power costs and evaluate flexible generation is, in part, a direct response to exactly the kind of exposure that stalled those two projects. A company committing hundreds of billions of dollars to data centre capacity cannot afford for electricity price volatility to be the reason a flagship site gets shelved.

There is also a harder financial context behind the timing. OpenAI’s own January pledge to “pay its own way on energy” only carries weight if the company has the in-house expertise to make good on it project by project, across wildly different grid conditions from rural Georgia to the industrial Midwest.

The trading desk and the clean energy role look like the operational machinery being built to support a public commitment that was, until now, largely aspirational, and one that sits alongside the grid capacity pressures Capacity has tracked across PJM’s territory, where large loads increasingly need their own power strategy to connect at all.

Energy has stopped being a line item that site selection teams negotiate once and forget. It is becoming a discipline that the largest AI companies are choosing to own directly, with dedicated trading, technology evaluation and community engagement functions sitting inside the infrastructure organisation itself. Meta got there first. OpenAI is now visibly following.

Whether Google, Microsoft and Amazon formalise the same internal structures, or continue leaning on external power purchase agreements and utility partnerships, will likely shape how confidently each of them can commit to the next wave of gigawatt-scale campuses, and how well they can defend those commitments when energy markets turn volatile.

VIRTUS Data Centres appoints Adam Eaton as new CEO to drive digital growth

What is behind-the-meter power and how is it reshaping the data centre industry?