In August, OpenAI quietly posted a job listing that said more about the next decade of enterprise risk management than most industry reports do. The role: a power-trading lead on its “Power & Land” team, tasked with turning the company’s “large, dynamic electricity and fuel exposures into practical hedging” strategies. The pay band runs from $181,000 to $285,000 plus equity. The job description reads less like a tech company posting and more like something you’d find at a utility or a commodity trading house — forwards, swaps, options, ISDAs, risk limits, internal controls.
That’s not a coincidence. It’s a direct consequence of what AI infrastructure has become: a business where electricity isn’t an overhead line item anymore, it’s a market position.
From utility bill to market exposure
OpenAI isn’t the first mover here, and that’s the real story. Meta filed with US regulators in September 2025 to start selling wholesale power, then formally entered power trading in November 2025 to support its AI buildout. Microsoft and Google face the same structural problem: campuses that need gigawatts of reliable, round-the-clock electricity in grids that were never designed for loads that large or that lumpy.
The scale makes the point. OpenAI is building a roughly $30 billion campus in Georgia and has reportedly discussed a 10-gigawatt site in Ohio backed by Nvidia. Energy pricing and regulatory friction already led the company to pause a UK Stargate project. More than 40% of US electricity still comes from burning gas, so a “power” trading desk at an AI company ends up needing gas-market expertise too. Meanwhile, grid operators are working on a regulatory “fast lane” for connecting very large loads — a process that will directly determine when and where these sites can even draw power.
Put plainly: when a single customer’s electricity demand can move a regional market, treating that demand as a fixed cost stops making sense. It becomes a risk to be measured, hedged, and governed — which is exactly the problem CTRM and ETRM systems were built to solve for commodity producers and utilities.
Borrowing a playbook that already exists
Nothing in the OpenAI job posting is conceptually new to anyone who has worked in energy trading. Quantifying exposure across markets and sites, evaluating fixed-price supply against forwards and options, and setting a framework for when and how much to hedge is the day-to-day work of an ETRM desk. What’s new is who’s doing it.
Hyperscalers are effectively being forced to stand up the same risk function an independent power producer or a refiner has run for decades, just under a different letterhead. That has a few practical implications worth watching:
- Demand for experienced power and gas traders is about to come from a new buyer category, which will tighten an already thin talent pool in commodity risk management.
- Vendors who built ETRM platforms for utilities and trading houses now have a logical adjacent market in hyperscale data-center operators, who need the same deal capture, position, and exposure reporting but may want it delivered differently — lighter, API-first, embedded in infrastructure planning rather than a standalone back office.
- Governance questions that energy trading desks settled years ago — limit-setting, segregation of duties, model validation for hedge effectiveness — are being reinvented from scratch inside companies whose core expertise is software, not commodities.
What this means beyond the AI giants
You don’t need a 10-gigawatt campus for this to matter. Any organization whose energy consumption has grown faster than its energy strategy — manufacturers adding electrified processes, logistics operators electrifying fleets, data-center operators at any scale — is edging toward the same fork OpenAI and Meta have already reached. The question isn’t whether to formalize energy risk management, it’s when the exposure gets large enough that informal management (a fixed-rate contract renewed on autopilot, a spreadsheet tracking usage) stops being adequate.
The signals to watch for inside your own organization look a lot like OpenAI’s job description, scaled down: electricity costs that swing enough quarter to quarter to affect margin forecasts, new sites or equipment that materially change your load profile, or exposure split across multiple grid regions with different price dynamics. Any one of those is a reason to ask whether your hedging approach — and the systems tracking it — were built for the exposure you have today, not the one you had three years ago.
The takeaway
The AI buildout is doing something unglamorous but significant to the energy trading world: it’s expanding the list of companies that need commodity risk management discipline well beyond the traditional producers, utilities, and trading houses. A job posting for a single power trader at one company isn’t the headline. The headline is that electricity has become a tradable, hedgeable, governance-worthy exposure for an entire new category of buyer — and the tools and talent built for that job are about to be in much higher demand.
