How the answers frame Duke Energy
This snapshot (29 September 2026) covers 250 benchmark questions across ChatGPT, Claude, Gemini, Perplexity and Grok, 1,250 answers in all. None of the questions named a brand. Duke Energy was named in 127 of those answers. That put it fifth among the 15 energy peers in the data, behind ExxonMobil, Shell, Chevron and NextEra Energy. Grok accounted for 49 of the 127 appearances. Claude named Duke in 16 answers and ChatGPT in 18.
By question intent, Duke's visibility was highest on "avoid" questions (23%), then leadership (19%), sustainability (17%) and employer (16%). It was lowest on value (1%), clean energy options (3%) and trust (4%). Reliability sat at 7% and service at 5%.
In the 16 sampled answers, Duke most often appears as one name in a list. It shows up among retiree dividend picks from ChatGPT, Gemini and Grok. It shows up among the largest utility employers in a Gemini answer. Claude's employer answer cited a 2026 ranking that put Duke 42nd overall. In answers on lobbying against rooftop solar, Gemini and Grok both placed Duke alongside FPL and Tampa Electric as a utility critics have singled out. Grok's coal-retirement answer grouped it with other large investor-owned utilities.
On the reliability question, ChatGPT's storm-restoration answer named Duke for self-healing technology. Its visible reasoning pointed to FPL as the utility with the strongest restoration reputation. Gemini's cybersecurity answer said Duke was "Frequently cited in industry research as a leader in implementing comprehensive cybersecurity measures". That is one of the few passages that credits Duke with a specific strength.
The credited-for tags are thin. Six credits were classed Not Substantive, two Labour & Workplace, and one each Reliability & Operations, Price & Value, Sustainability & Ethics and Financial Performance.
Frequently cited in industry research as a leader in implementing comprehensive cybersecurity measures…
What the press is saying
PeakMetrics' news index shows 5,446 mentions of Duke Energy over the last 30 days, 1,547 with the company in the headline. Recent headlines fall into a few storylines. One is community giving: a Duke Energy Foundation award of $100,000 for upward mobility in Florida. One is grid technology: AWS and Duke Energy using AI agents to speed up interconnection studies. One is customer tools during a heat wave in Ohio and Kentucky. The last is scrutiny of load growth, with headlines on more gas power amid criticism that data centers are the reason, and on what data centers could cost Duke Energy Florida customers, which is redacted.
Most large US utilities (primarily investor-owned ones like Duke Energy, Southern Company subsidiaries, Dominion, Entergy, and others) are retiring coal plants slowly…
The gap
The press coverage centers on the data-center buildout, gas capacity and who pays for it, alongside AI-driven grid work. In the 16 sampled answers, none framed Duke around data centers or interconnection. The assistants placed it in dividend, employer, lobbying and coal-retirement lists.
Volume also differs. Duke's news count is lower than that of Chevron, Shell, BP, ExxonMobil and several other oil and gas peers, but it still appeared in more answers than BP, PG&E and TotalEnergies, which have comparable or larger news counts. This is one snapshot, and it shows that press volume and answer visibility do not track each other one for one.
What to watch
A comms team can read this as a question of which contexts carry the company's name. In this snapshot, lobbying criticism and coal-retirement comparisons sit next to Duke, while value, clean-energy options and trust questions barely name it. The cited domains behind those answers, not only the company's own channels, are what to check.
Grok produced the largest share of Duke mentions, so a single assistant can shape the overall picture. Live answers vary between runs, so treat these figures as a baseline to re-measure, not a ranking.
The data-center and gas-cost storyline is heavy in the news index and light in these answers. Whether assistants pick it up is worth tracking in later benchmark runs.
