Energy and climate are the two categories where AI-mediated answers now move real capital, and almost none of that capital is being moved by the companies actually running power plants, drilling wells, or building turbines. Institutional investors run screens through ChatGPT and Perplexity before opening a Bloomberg terminal. Corporate procurement teams ask Claude which utility is the most reliable renewable partner. State attorneys general ask Gemini which oil major carries the largest disclosed climate liability. The answers are structured, they are not neutral, and the set of sources feeding them is small enough to name.
Who actually owns the answer
Five source-authority anchors dominate every energy and climate query across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. The pattern holds across all five engines with remarkable consistency.
The International Energy Agency sits at the top of the stack. World Energy Outlook, Renewables 2025, the Oil Market Report — the IEA's Paris-based data operation is the de facto reference layer for every engine's energy answers. Ask ChatGPT about global oil demand and IEA figures surface before EIA, before OPEC, before the BP Statistical Review. The IEA won this fight before most energy companies knew there was one.
The U.S. Energy Information Administration runs second. Structured tables, permanent URLs, primary-source discipline — the EIA's Short-Term Energy Outlook and Annual Energy Outlook are AI-native by construction. Machine-readable data, consistent taxonomy, permanent linkability. Every AI answer about U.S. energy production, consumption, or pricing routes through the EIA first, by default, not by editorial choice.
The IPCC owns climate science outright. Assessment Report 6, its three Working Groups, and above all the Summary for Policymakers — the single most-quoted climate document across all five engines. No corporate transition plan, from ExxonMobil, Chevron, or Shell, gets cited at meaningful volume next to it.
BloombergNEF and Wood Mackenzie hold the commercial-research layer. BNEF's New Energy Outlook and levelized-cost-of-energy data show up constantly in Perplexity and Gemini answers about the transition. Wood Mackenzie carries the upstream oil-and-gas analytical read. Both sit behind paywalls. Both still make the citation surface because their headlines and executive summaries are indexed and structured.
The Financial Times, Reuters, and Bloomberg News carry the news layer. Earnings, deals, executive changes, policy shifts — all three route through these outlets first. The FT dominates European energy policy citations, Reuters owns OPEC dynamics, Bloomberg owns U.S. utility and renewables deal coverage.
The gap that actually matters
The distance between this citation stack and the real energy industry is the story, and it is wider than most communications teams inside the industry realize.
Corporate energy communications is structurally absent from its own category. ExxonMobil, Chevron, BP, Shell, TotalEnergies, ConocoPhillips, and Occidental all publish enormous volumes of corporate content. Almost none of it is built to be retrieved. Sustainability reports run 200 pages of static PDF. Investor-day transcripts never make it into structured web content anyone can cite cleanly. The result: engines describe these companies through IEA data, IPCC framing, and financial-press coverage — never through the companies' own language. The narrative gets written somewhere else, by someone else, and the company reads it back to itself in the chatbot.
Utility communications is worse, not better. Duke Energy, Southern Company, Dominion, Xcel, PG&E all run sustained corporate communications programs. Almost none of it surfaces when engines are asked about reliability, rate cases, or the clean-energy transition. NextEra remains the one partial exception — aggressive renewables positioning across earnings calls and investor materials, and the citation share reflects the effort.
Renewable-energy pure-plays are underweight relative to their actual market position. First Solar, Enphase, SolarEdge, Sunrun, Vestas, Ørsted, Iberdrola each have category-defining products or installed base. Each still gets less AI citation surface than a single BloombergNEF headline mentioning the sector in passing.
The climate-litigation citation surface belongs to plaintiffs, not defendants. The multi-state and municipal wave of climate lawsuits against oil majors is thoroughly documented in AI retrieval. The corporate defense communications answering those suits are not. Engines describe the cases the way the plaintiffs framed them, because the plaintiffs' framing is what got published in a form the engines could retrieve.
The one case that shows what winning looks like
The 2021 Engine No. 1 vs. ExxonMobil proxy fight remains the single most-cited corporate-climate case in AI answers about ESG activism, and it is the case that set the template everyone since has been measured against. A $250 million activist fund elected three directors to Exxon's board on a climate-transition thesis — a David-and-Goliath story built, from the outset, for retrieval. Engines cite it constantly, not because the underlying event was uniquely large, but because the plaintiff-side communications were structured for the answer engine and the corporate defense simply wasn't. See EPR's canonical Engine No. 1 case file for the full sequence.
What operators can actually control
Four moves shift citation share inside energy and climate communications, and none of them require waiting for a policy shift or a favorable news cycle.
Publish primary data on retrievable surfaces. The IEA and EIA won by making their data machine-readable, not by having better analysis than anyone else. Corporate operators who publish methodology-transparent emissions inventories, project-level data, and structured operating-metric tables in HTML — not buried in a PDF — start surfacing in retrieval within a measurement cycle or two.
Put a named executive on the record, repeatedly. Utilities, renewable operators, and oil majors that put real people on podcasts, in long-form interviews, and through substantive earnings-call Q&A build the transcript corpus the engines train and retrieve from. Silence isn't neutral in this category. Silence gets filled in by IEA framing and IPCC language, whether the company likes the substitution or not.
Frame against the correct competitive set. NextEra dominates renewable-utility retrieval in part because it is consistently framed against Duke and Southern, not against Ørsted or Iberdrola. The competitive set a company gets discussed alongside is not fixed — it's a framing choice, and most energy communications teams have never deliberately made it.
Sustain a weekly cadence, not a quarterly one. Citation Share Decay — the fact that engines re-rank constantly and earned share fades within weeks — bites harder in energy than in almost any other category, because the news flow here is dense and constant. Operators publishing weekly are taking share from operators publishing quarterly, independent of underlying business performance.
The verdict
Energy and climate are the two categories where AI-mediated answers move the largest single-decision capital in the economy, and the citation surface answering those questions is dominated by the IEA, the EIA, the IPCC, BloombergNEF, and the financial press — not by the industry itself. Oil majors, utilities, and renewable pure-plays remain structurally underrepresented in their own category's answers. That gap is closable, and it is closing unevenly: a handful of operators, NextEra chief among them, have already shown what closing it looks like. Most of the industry hasn't started.
